Ouvr - Taxation-Wealth-And-Saving

1
cover next page >
title : Taxation, Wealth, and Saving

author : Bradford, David F.
publisher : MIT Press
isbn10 | asin : 0262024705
print isbn13 : 9780262024709
ebook isbn13 : 9780585240213
language : English
Income tax--United States, Spendings tax--United States,
subject Taxation of articles of consumption--United States, Saving
and investment--United States.
publication date : 2000
lcc : HJ4652.B668 2000eb
ddc : 336.2/00973
Income tax--United States, Spendings tax--United States,
subject : Taxation of articles of consumption--United States, Saving
and investment--United States.
cover next page >
< previous page page_iii next page >
Page iii
Taxation, Wealth, and Saving

David F. Bradford
1
2
Â
< previous page page_iii next page >
< previous page page_iv next page >
Â© 2000 Massachusetts Institute of Technology
All rights reserved No part of this book may be reproduced in any form by any electronic or mechanical means (including phot
or information storage and retrieval) without permission in writing from the publisher.
This book was set in Palatino by Asco Typesetters, Hong Kong.
Printed and bound in the United States of America.
LibraryÂ ofÂ CongressÂ Cataloging-in-PublicationÂ Data
Bradford,Â DavidÂ F.,Â 1939-

Taxation,Â wealth,Â andÂ savingÂ /Â DavidÂ F.Â Bradford.
p.Â cm.
IncludesÂ bibliographicalÂ referencesÂ andÂ index.
ISBNÂ 0-262-02470-5Â (alk.Â paper)
1.Â IncomeÂ taxUnitedÂ States.Â 2.Â SpendingsÂ taxUnitedÂ States.
3.Â TaxationÂ ofÂ articlesÂ ofÂ consumptionUnitedÂ StatesÂ 4.Â Saving
andÂ investmentUnitedÂ States.Â I.Â Title.
HJ4652.B668Â 1999
336.2'00973dc21Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â 99-31172
ÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂ
Â
< previous page page_iv next page >
< previous page page_v next page >
Page v
CONTENTS
Introduction vii
I 1
Income and Consumption Taxation: Basic Concepts and Broad Policy Issues
1 3
The Case for a Personal Consumption Tax
2
3
2 41
The Choice between Income and Consumption Taxes
3 65
On the Incidence of Consumption Taxes
4 85
Fundamental Issues in Consumption Taxation
II 121
Income and Consumption Taxation: Theoretical Tools and Issues
5 123
Tax Neutrality and the Investment Tax Credit
6 141
The Economics of Tax Policy toward Savings
7 205
Issues in the Design of Savings and Investment Incentives
8 243
The Incidence and Allocation Effects of a Tax on Corporate Distributions
9 267
A Problem of Financial Market Equilibrium When the Timing of Tax Payments Is
Indeterminate
Â
< previous page page_v next page >
< previous page page_vi next page >
Page vi
10 281
Pitfalls in the Construction and Use of Effective Tax Rates (with Don Fullerton)
III 309
Income and Consumption Taxation: Transition, Complexity, and Implementation
11 311
Transition to and Tax-Rate Flexibility in a Cash-Flow-Type Tax
12 333
What's in a Name? Income, Consumption, and the Sources of Tax Complexity
13 343
Treatment of Financial Services under Income and Consumption Taxes
3
4
14 371
Fixing Realization Accounting: Symmetry, Consistency, and Correctness in the
Taxation of Financial Instruments
IV 429
Economically
Meaningful Measures
of National Saving
15 431
Market Value versus Financial Accounting Measures of National Saving
16 467
What Is National Saving? Alternative Measures in Historical and International
Context
Sources 517
Index 519
Â
< previous page page_vi next page >
< previous page page_vii next page >
Page vii
INTRODUCTION
The papers collected in this volume include most of those I have written on the subject of taxation, wealth,
and saving since chance focussed my attention on the income tax in 1975. That was when I had the good
fortune to succeed George S. Tolley as Deputy Assistant Secretary of the United States Treasury for Tax
Policy and Director of the Treasury's Office of Tax Analysis. At the time I regarded myself as what we would
now call a middle brow theorist, interested in subjects like second best taxation and discounting in public
investment decisions. I exaggerate only a little when I say that "income" to me was something captured by
the symbol y, the sum of a wage rate times the quantity of labor and an interest rate times the quantity of
capital. If I had been exposed to the concept of "Haig-Simons" income in graduate school, I had managed to
forget it.
At the Treasury I found myself

immersed in such practical issues
as how best to integrate individual
and corporation income taxes,
whether to count air travel
provided by airline companies to
their employees in their taxable
income, whether to allow workers
whose employers did not have
tax-preferred pension plans to
4
5
establish their own, and on, and

on. Collaborating with an
excellent team of economists and
lawyers, I discovered that not
only were many such questions of
practical importance but they also
engaged my economist's
puzzle-solving skills. As a
consequence, taking the position
at the Treasury turned out to have
an enormous impact on my life as
an economist.
Adding to the day-to-day business of preparing memoranda and testimony and to the ongoing process of
developing the Treasury's capacity for tax analysis was a special project assigned to us by Treasury Secretary
William Simon. In a speech in December 1975 he
Â
< previous page page_vii next page >
< previous page page_viii next page >
Page viii
called for a fundamental reform of the income tax, with broadened base and lower rates. He turned to the
Office of Tax Policy to produce the specifics.
One might have thought that by culling the output of scholars such as Brookings Institution's Joseph Pechman
and of study groups such as Canada's Carter Commission we could have put together a plan in short order.
Much to an impatient Secretary Simon's frustration, however, collecting the data and developing the software
to support the Treasury's traditional analytical tables for such a complex reform and resolving debates about
the myriad policy decisions involved were time consuming. It took the better part of a year to produce the
Secretary's basic tax reform plan. By then, President Ford had lost his bid for a second term and what was to
have been a Treasury reform proposal became, instead, a Treasury report. Entitled Blueprints for Basic Tax
Reform (U.S. Treasury Department, 1977), the study appeared, as I recall, on the desks of Congressmen on
the day of President Carter's inauguration and it received the attention one might expect for the product of an
outgoing administration.
Blueprints did, however, have an impact on the community of tax specialists and profoundly affected my
subsequent professional career. Economic analysis of the income tax system and tax policy advocacy
absorbed a large fraction of my work in economics from that time on.
Blueprints described two models for income tax reform, one based on the Haig-Simons income concept that
dominated (and arguably still dominates) American income tax thinking, the other based on consumption,
building on an intellectual tradition dating back to John Stuart Mill. Before we began the Treasury basic tax
reform project I had not given much thought to the consumption alternative, joining most economists in the
view that it was an interesting but impractical concept. But as I dug into subject, I came to the contrary
conclusion that consumption was superior to income as a tax base on just about any ground one might care to
consider, save one: It was not politically acceptable. Thus I did not succeed in persuading Secretary Simon
that his reform proposal should be built on a consumption base. He was, however, willing to let the staff
continue to work on a consumption version. When the political factor ceased to be central, with the
concurrence of President Ford he authorized the inclusion of the consumption-based system, dubbed the Cash
Flow
Â
< previous page page_viii next page >
5
6
< previous page page_ix next page >
Page ix
Tax, together with the Haig-Simons system, called the Comprehensive Income Tax, in the official Treasury
report.
The papers in Part I of this volume concern the broad issues in the choice between income and consumption
as the basis for the principal system directed at distributive concerns.
The first paper, "The Case for a Personal Consumption Tax," was written at the invitation of Joseph
Pechman, who, not long after I left the Treasury, organized a conference at Brookings to consider the relative
merits of traditional income- and consumption-, or expenditure-, based reforms. Knowing my views, he asked
me to argue the pro-consumption position. In that paper, as well as in the next one reproduced, "The Choice
between Income and Consumption Taxes," I present the case in favor of consumption as a base in terms of
the traditional categories of equity, efficiency, and simplicity, having in mind, in particular, the Cash Flow
Tax. That plan falls in the category that came to be called ''consumed income taxes," to suggest that the
taxpayer's liability is based progressively on the amount of income currently consumed.
The accounting device by which a consumed income tax is implemented is like that by which a traditional
income tax is implemented. One starts with measures of receipts by the taxpayer. The tax base is then derived
by a process of subtracting items to be excluded from the base. An example would be deducting from a
carpenter's receipts amounts paid for the rental of power tools. To get to a consumed income tax the starting
point is the individual's cash receipts from all sources, including borrowing and sales of assets, from which
are deducted (in addition to items like the carpenter's rental of tools) all amounts used to acquire assets
(including items such as deposits to savings accounts).
In the course of putting together the Treasury's plan, we came to appreciate the practical advantages in many
circumstances of the economic equivalence between this cash-flow treatment of saving transactions with the
alternative of simply ignoring both the amount used to acquire assets and any subsequent return flow from
the assets. We called the latter the "tax prepayment" approach to implementing a consumption-type tax and
used this alternative in a number of applications. The expression captures the idea that, because no deduction
is allowed for the amount saved, the taxpayer is paying in
Â
< previous page page_ix next page >
< previous page page_x next page >
Page x
advance the discounted value of tax that would be due at the time of future dissaving under the consumed
income approach. The tax prepayment approach is now also known as the "yield exemption" approach.
As I have mentioned, the notion of a consumption-based tax was sufficiently charged politically to take it out
of serious consideration as an Administration proposal in the Simon Treasury. I thought the ordinary citizen
would not be much concerned about the economists' technical view, but rather think of both the Blueprints
plans as variants of what they already called an income tax. So I wanted to borrow from William Andrews
(1974) to call what finally emerged as the Cash Flow Tax the Cash-Flow Income Tax. The Treasury's staff of
professional economists felt, however, that this would be intellectually dishonest and I reluctantly accepted
their view.
When Robert Hall and Alvin Rabushka put forth their Flat Tax proposal in 1983 they were not so punctilious,
describing their ingenious system as a flat income tax even though it was a form of consumption tax under
the economists' technical definition. The Flat Tax achieves great simplicity by dispensing altogether with the
individual-level accounting for financial transactions (such as payment and receipt of interest) and using a tax
on business firms to account for all but the tax on individual wage and salary earnings. From an individual
6
7
point of view, it puts the taxation of saving entirely on a tax prepayment basis.
Not only did I think Hall and Rabushka's terminology was appropriate, I also became persuaded that the basic
structure of the Flat Tax offered significant advantages over the consumed income method, both in
implementation and in transition. In Untangling the Income Tax (1986), a book prepared for the Committee
for Economic Development, I described that structure as a "two-tiered consumption tax" (the two tiers being
the business and employee levels). At some point, I labeled my preferred version of this generic approach the
"X Tax," attempting thereby to duck the politically charged question of its identity as an income or a
consumption tax. My own files leave me in doubt about exactly when or where I first advanced a plan by this
name, but two versions are described in the third paper reproduced here, "On the Incidence of Consumption
Taxes.'' I like to use this paper to teach students how important it is to look beyond the surface at which the
layman understandably tends to stop, to dig out the underlying economic structure of tax and
Â
< previous page page_x next page >
< previous page page_xi next page >
Page xi
transfer systems. In this case, two apparently very different systems, one taxing wages and one subsidizing
wages, lead to identical results.
The next paper, "Fundamental Issues in Consumption Taxation," illustrates how it is not only the layman who
can fail to think things through. This is a revised version of a paper prepared at the behest of Michael Boskin
for a Hoover Institution Conference in May 1995. It is well known to people in the field that a consumption
tax has the effect of a zero tax on "income from capital." So, for example, in general equilibrium simulation
analysis a shift from an income to a consumption tax is modeled by cutting to zero the tax on the rate of
return to capital. In a typical simulation for the United States the assumed rate of return might be six percent
per year or higher. Understood as a rough description of reality, this rate is supposed to reflect the average
inflation-adjusted yield in financial markets or perhaps the yield on real investment in the nation's capital
stock. In most models, the simulated impact of the shift in tax systems is fairly sensitive to this expected yield
on capitalif it were zero, there would be no economic difference between the two regimes. (Given the power
of compounding, the difference between six percent and some much smaller figure also has a major effect on
the potential accumulation at the personal level that one might reasonably project from a shift in base.)
In the Boskin conference paper I emphasized that the difference between income and consumption taxes,
thought about at the business level, is entirely in the timing of the write-off of investment, which is expensed
under the consumption approach, and depreciated under the income approach. Risk is treated in an
economically equivalent manner in income and consumption taxes as these are generally implemented. So,
for example, the high payoff from a risky business investment that succeeds generates extra tax revenue and
the low payoff generates less revenue or a deductible loss under either approach; only the timing may differ.
To make a short statement of a more complex argument, if losses and gains are treated symmetrically, neither
approach imposes a burden on the risk taker. Consequently, it is just taxing the risk-free rate of return in an
economy that distinguishes the income and consumption approaches, a number far below six percent, and
maybe not greater than one percent. As a result, the pure forms of the two taxes cannot produce very different
consequences in any dimension, distributive or
Â
< previous page page_xi next page >
< previous page page_xii next page >
7
8
Page xii
allocative. To me this fact came as a surprise after almost twenty-five years of thinking rather actively about
the subject. (This surprise did not affect my policy preferences much, since I have tended to emphasize the
differences between practically attainable tax systems and have stressed that an income-based tax is
inevitably plagued with inconsistencies that can be averted in a consumption-based system.)
In writing "Fundamental Issues in Consumption Taxation" I came to understand that the concept of "income
from capital," which analysts take for granted, is not at all clear. The analytically correct proposition that a
consumption talc exempts the yield from capital, understood as the risk-free rate of return to waiting, does
not translate easily into conclusions about its effect on, say, a successful entrepreneur. In my view, erroneous
conclusions based on misunderstanding this point have seriously impeded sensible reform. "Fundamental
Issues in Consumption Taxation" also addresses errors in the analysis of transition to a consumption base that
were and remain common in the public debates on this subject. I have more to say on transition in one of the
papers in Part IV.
The papers in Part II illustrate some of the interesting analytical challenges I encountered in thinking about
income and consumption taxes. "Tax Neutrality and the Investment Credit" deals with an issue that came up
while I was at the Treasury, namely, the conceptually correct way of calibrating an investment credit for the
durability of the property. (Reflecting professional confusion on the question, under the tax law at the time
the investment credit was greater for long-lived than for short-lived capital but there was also an adjustment
to the depreciation basis for the amount of the credit.) At the moment, such details are not of great policy
concern, since the general investment credit was repealed in 1986. The same issue arises in other contexts,
however. Furthermore, as I argue in one of the papers, an investment credit is superior to devices such as
accelerated depreciation as a technique for offsetting the disincentive to investment inherent in an income tax.
Perhaps such a credit will reappear in the policy portfolio.
The second paper, "The Economics of Tax Policy toward Savings," attempts to pull together the basic theory
relating to an issue of central importance for tax policy. Conventional wisdom had it that
Â
< previous page page_xii next page >
< previous page page_xiii next page >
Page xiii
whether the reduced (to zero) rate of tax on capital implied by consumption- rather than income-based
taxation is desirable from an allocative point of view depends upon the responsiveness of savings to the rate
of interest. If the savings rate is not sensitive to the interest rate, there was thought to be little or no efficiency
advantage to a consumption tax. Consequently, a great deal of effort was devoted to estimating the elasticity
of saving with respect to the rate of return. In this paper I use a simple two-period model incorporating labor
supply to develop the optimal tax conclusion that whether one "should" tax savings (on efficiency grounds)
depends on the cross elasticity between the supply of labor and the interest rate, and not on the elasticity of
saving or even of consumption with respect to the interest rate. Although I am naturally partial to my own
way of making the point, it is one that was also being made by others (Mervyn King [1980], for example) at
about the same time. A differentiating feature of my paper is its treatment of the underappreciated way that
multiple tax rates (e.g., tax-exempt and taxable individuals) affect the analysis, and how they can even totally
reverse the predicted effect of investment incentives.
The third paper, "Issues in the Design of Savings and Investment Incentives," is also distinguished by its
attention to the way typical tax treatments of saving and investment may be expected to work out in a world
where tax rates differ across individuals or companies. Unlike the world of the theoretical income tax, in the
real world, business and financial income are not correctly measured under conditions of inflation. Even with
stable prices, actual depreciation allowances differ from those implied by economic theory. Furthermore, at
8
9
that time an investment credit was an element of the U.S. tax system even though not part of a theoretical
income tax. This paper shows the surprising ways these features may interact. Among other things, I spell out
the difference between an investment tax credit and accelerated depreciation (one leads to a narrowing of the
after-tax rate of return across the population of taxpayers, and the other simply to a change in the general
level of the after-tax rate of return). I explain why an investment tax credit does not (and accelerated
depreciation does) lead to sheltering (which is defined) if interest is fully deductible (and taxable). The paper
also goes into questions of tax design. For example, I explain why partial write-off is the way to go if
acceleration is desired (in order to get the correct
Â
< previous page page_xiii next page >
< previous page page_xiv next page >
Page xiv
discrimination by asset durability) and how interest deduction (or inclusion) should be related to the
individual's tax rate in order to prevent sheltering distortions.
Integration of corporation and individual income taxes was an issue under active consideration when I was at
the Treasury. The Administration had, in fact, submitted a specific plan to Congress. Most economists took
for granted that eliminating the double tax on corporate equity (once at the level of the company and a second
tax on dividends at the shareholder level) would generate efficiency gains. I was bothered, however, by a gap
between the view of the tax as an extra burden on capital used in the corporate sector and the actual rules by
which the corporate tax was implemented. Henry Tulkens had invited me to spend the semester after leaving
the Treasury as a guest at CORE in Louvain, where I had time to think the issue through. "The Incidence and
Allocation Effects of a Tax on Corporate Distributions" contains the results. This paper spells out a model of
what came to be called the "new view" of the corporation tax, in which the second tax, the dividend tax, has
no distorting effect. Of relevance for policy, in the simple model world some versions of corporate
integration would accomplish nothing but dissipation of revenue. Roger Gordon and I (1980) subsequently
sought to put the model to an empirical test. He concluded that our evidence ran against the new view,
whereas I saw support for it in the same results!
Encountering "The Incidence and Allocation Effects of a Tax on Corporate Distributions," Joseph Pechman
said I was the first person whose work done after leaving the Treasury was harder to read than that done
before. It is true that to address skepticism I had encountered in lecturing on early versions of the analysis I
was at pains to dot the rational-expectations i's and cross the overlapping-generations t's. So, although the
basic insight is simple, the paper may unfortunately not be accessible to some for whom I hope most of the
rest of this book is of use. This is also true of the next paper, "A Problem of Financial Market Equilibrium
When the Timing of Tax Payments Is Indeterminate," which is the most mathematical in this collection. It
uses the same formal structure to consider an expanded set of financial instruments available to the
government beyond simple bonds. I chose to include the paper here because the puzzle it raises, without
solving, seems to me relevant to the current con-
Â
< previous page page_xiv next page >
< previous page page_xv next page >
Page xv
sideration of government investment in the stock market (beyond the investment already implied by the tax
system).
In the last paper of Part II, "Pitfalls in the Construction and Use of Effective Tax Rates," Don Fullerton and I
parse a tool commonly deployed by analysts to capture the net effect of complex rules relating to the taxation
9
10
of capital. The idea of the effective tax rate used in this context is simple. If the going interest rate is 5
percent per year and the net effect of a set of rules is that the saver ends up getting 4 percent, it makes sense
to say the effective rate of tax on capital income is 20 percent. Our paper describes so many different ways
one might calculate effective tax rates that I have never since had much confidence in them. The message of
the paper, however, is not that the device is meaningless but that it is critical to be clear about the question to
which the effective rate in question is the answer. As so often in the field of taxation, the devil is in the
details.
In Part III I have assembled papers addressing a variety of issues relating to the potential deployment of the
consumption approach to taxation. Writing the first one on the list led me to a change in my preferred policy.
In a fairly typical economists' view, a consumption base is preferable to an income base but it is harder to
implement. Perhaps more important, it is hard to make a transition to it from an income base (or even from
our existing "income" base, which is far from the theoretical concept generally used to model this choice). I
had long been convinced the first proposition (harder to implement) is wrong but had accepted that the
transition is a serious challenge. Starting with Blueprints, I consequently devoted a good deal of attention to
the issue and had reached the conclusion that the transition was not a more serious problem than had been
faced in actual major changes in the law in the past (in 1981 and 1986, for example). In the course of writing
"Transition to and Tax Rate Flexibility in a Cash-Flow Type Tax," it became clear to me the transition
problem is in one sense more serious than I had thought, since it is an ongoing one. That is, in the two-tiered
consumption tax approach (i.e., the generic Flat Tax approach) that seems to me the most promising one to
take, every change in the business tax rate, and even the risk of change in that rate, generates qualitatively the
same incidence and incentive effects as does the introduction of the system in the first place. On the other
hand, I realized that there is a reasonable
Â
< previous page page_xv next page >
< previous page page_xvi next page >
Page xvi
approach to minimizing this problem, which also addresses the onetime switch from an income to a
consumption base. The answer is a change in the treatment of business investment, away from the cash-flow
accounting (expensing) that is generally understood as the hallmark of the approach, substituting depreciation
and related elements of business income accounting, combined with a deduction for the yield on capital tied
up in the business. It may sound complicated but it isn't particularly so.
Complexity is, however, a serious problem in the existing system. I have argued in many places that the
traditional income concept cannot be implemented in a simple way, while the consumption approach can be.
The word "can" is chosen advisedly, however. Unless rules are based on a clear conception of the object of
the game, consumption-based taxation could be made as arbitrarily complicated as any other regulatory
system. The paper, "What's in a Name? Income, Consumption, and the Source of Tax Complexity," points
out the traps for the unwary in this important aspect of policy. The context is a comment on a paper by
Deborah Paul but I think it is reasonably self-contained.
Two-tiered consumption-type (or income-type, for that matter) taxes are confined to "real" transactions, that
is, purchases and sales of goods and services. In some common transactions real and financial elements are
mingled, however. An example would be sale of an automobile on an installment basis, combining sale of the
good, a car, and issue of a loan. Such situations call for carefully designed rules. For example, a simple way
to deal with the installment sale is to require the seller to account for the aggregated transaction on a
cash-flow basis. The taxation of financial institutions raises such issues in a particularly pronounced way and
it is generally assumed that these institutions would require a special tax regime. In "Treatment of Financial
Services under Income and Consumption Taxes" I argue that the problem is equally prevalent, albeit
disguised, in income tax systems but that reasonable remedies are available.
10
11
Students of taxation are familiar with the fact that to do an income tax "right" requires an annual accounting
for wealth at market value. That is, a person who does nothing but hold an asset from the beginning until the
end of the year has income equal to the change in market value of the asset (up or down). Actual income tax
systems, however, use "realization accounting" for many assets and liabilities. Thus, the person in our
example would not have income for tax
Â
< previous page page_xvi next page >
< previous page page_xvii next page >
Page xvii
purposes until the asset was sold,

at which time a "realized" gain or
loss would be recorded. Other
assets and liabilities are
accounted for on an accrual basis.
For example, the issuer of a
zero-coupon bond obtains a series
of deductions for interest during
the life of the bond, and the
holder of the bond is obliged to
include the accruing interest in
his tax base, even though no cash
changes hands until the maturity
of the bond. The inconsistency
between these two types of
accounting can be exploited by
clever taxpayers to drain money
from the fisc. Once taxpayers
discover such opportunities, the
tax administration must counter
with special anti-abuse rules.
Tax scholars have generally sought to deal with this set of problems by requiring taxpayers to use market
values in figuring their income with respect to certain classes of assets and liabilities, "mark to market"
accounting in our jargon. There are clear practical difficulties with using this approach for all assets and
liabilities, however. It is, in particular, ill suited for dealing with complex financial instruments (other than
those traded on active exchanges) that are currently being developed at a dizzying pace. In the last paper in
Part III, "Fixing Realization Accounting: Symmetry, Consistency and Correctness in the Taxation of
Financial Instruments," I describe a class of methods to solve these problems using realization accounting. I
believe that the approach described in that paper could, in principle, be deployed to render realization
accounting workable more generally. In that sense, the paper contributes to saving the income tax option.
Most readers seem, however, to agree with me that the method is unintuitive and unlikely to win acceptance.
To that extent, the paper demonstrates why the process of one complex ad hoc rule following another will
inevitably continue, barring a fundamental restructuring of the tax system.
The two papers in Part IV are quite different from the rest. They are linked, however, to the beginning of this
story. Secretary Simon was an ardent promoter of capital formation, who perceived the rates of saving and
investment as too low for the good of the nation. Many others, before and since, have lamented the low rate
of saving in the United States, in comparison with other countries and with our own past. Without implying a
position on the merits of this concern or even on the accuracy of the broad observation, I had always been
struck by the remarkable disparity between the statistical basis for drawing conclusions about saving and the
economic
11
12
Â
< previous page page_xvii next page >
< previous page page_xviii next page >
Page xviii
theory regarding saving behavior. The latter presumes that individuals and households are concerned with
their wealth, for example in relation to planning for retirement. The wealth in the theory is net worth at
market value, marked to market, in other words. The economic definition of saving is the change of wealth
over time. The statistics on aggregate saving that are generally used, however, are all effectively on a book
value basis.
Living as we do in a period of stunning variation in the market value of assets and liabilities, and, in
particular, of dramatic differences between companies' accounting net worth and their market values, it
should go without saying that book values are an insufficient guide to judging economic performance. In
dealing with information about individual companies, economists are traditionally skeptical of book net
worth, stressing the need to use data on prospects for the future as captured in capital market valuation. In
"Market Value versus Financial Accounting Measures of National Saving" I take up conceptual and data
problems that confront the construction of a measure of national wealth and saving corresponding to the
concepts in economic theory. The figures presented in the paper show that the difference between standard
statistical measures and the capital market's valuation of wealth is substantial and variable. As its title
suggests, the second paper, "What Is National Saving?: Alternative Measures in Historical and International
Context," extends the analysis to put it in the longer historical and the international perspectives.
Critics of the market valuation of wealth argue that it is ephemeral, an objection that does not address the
presumption of economic theory that it is the market's valuation that is the subject of individuals' behavior
and concerns. More telling, I think, is the objection that wealth is a poor statistic of economic performance.
Thus, for example, wealth can increase either because of an increase in the expected future flow of goods and
services or because of a fall in the price of a given flow of future services (a decline in the interest rate).
Recent work by Michel Reiter (1996) greatly sharpens our understanding of the underlying index number
problem, but it is a concern that seems little shared in the profession.
In concluding this rather personal introductory overview, I would like to acknowledge three people who had
important influence on the work presented in this volume. The first is Kenneth J. Arrow,
Â
< previous page page_xviii next page >
< previous page page_xix next page >
Page xix
who was my doctoral supervisor at Stanford University. I have tried to emulate his intellectual standards in
all of my work. His example of rigorous analysis of socially important questions has been a lifelong
inspiration. The second is Martin Feldstein, who has had a profound influence on economic research,
including research on taxation, and who has stimulated me to write several of the papers included here. At the
conclusion of my service at the Treasury, he invited me to participate in the transformation he was just then
beginning at the National Bureau of Economic Research. The position of director of the Bureau's program of
research on taxation provided me with a connection to a national community of public finance scholars that
continues to be enormously rewarding. Long before that he was also responsible for introducing me to
William Andrews of Harvard Law School, the third person on my list. William Andrews's insights into the
nature of the income tax problem fundamentally changed my view of it and greatly enriched my encounter
with the complex social institution we call income tax law. I hope the many others to whom I owe debts of
gratitude will not mind my singling out these three friends for special acknowledgment.
12
13
References
Andrews, William D., "A Consumption-Type or Cash Flow Personal Income Tax," Harvard Law Review 87
(April 1974): 1113-1188.
Bradford, David F., Untangling the Income Tax, Cambridge, MA: Harvard University Press, April 1986.
Gordon, Roger H., and David F. Bradford, "Taxation and the Stock Market Valuation of Capital Gains and
Dividends: Theory and Empirical Results," Journal of Public Economics 14 (October 1980): 109-136.
Hall, Robert E., and Alvin Rabushka, Low Tax, Simple Tax, Flat Tax, New York: McGraw-Hill, 1983; 2d ed.
issued as, The Flat Tax, Stanford, Calif: Hoover Institution Press, 1995.
King, Mervyn A., "Savings and Taxation," in G. A. Hughes and G. M. Heal (eds.), Public Policy and the Tax
System, London: George Allen and Unwin, 1980, pp. 1-35.
Reiter, Michael, "National Wealth, Saving and Asset Prices," MS, University of Munich, February 1996.
U.S. Government, Department

of the Treasury, Blueprints for
Basic Tax Reform, Washington,
DC: U.S. Government Printing
Office, January 1977, reissued
with an introduction discussing
some of my afterthoughts as
David F. Bradford and the U.S.
Treasury Tax Policy Staff,
Blueprints for Basic Tax Reform,
2d ed., Washington, DC: Tax
Analysts, 1984.
Â
< previous page page_xix next page >
< previous page page_1 next page >
Page 1
I
INCOME AND CONSUMPTION TAXATION: BASIC CONCEPTS AND BROAD
POLICY ISSUES
Â
13
14
Page 3
1
The Case for a Personal Consumption Tax
The publication of the U.S. Treasury's Blueprints for Basic Tax Reform in 1977 and of the Meade committee
report in the United Kingdom in 1978 1 reflects the renewed interest in the idea of an expenditure or
consumption tax as a practical and desirable alternative to an income tax.2 These reports have much in
common. Both regard the remarkably varied treatment of different forms of saving, the unnecessary
complexity of tax rules, and the great sensitivity of relative tax burdens to the rate of inflation as defects of
the existing income tax systems of the two countries. Both reports conclude that no simple set of
principlesand certainly no consistent concept of incomegoverns present ''income" tax policy.
When one considers the political process by which tax law is made, this is not surprising. But having an
effective policy to deal with the hundreds of rule interpretations and proposed rule changes each year requires
a clear objective. The dominant approach to tax reform, in the United States at least, has been to aim for
comprehensive income taxation.3 But both the Treasury and Meade committee studies conclude that a
comprehensive consumption base has advantages over an income base in terms of both the equity of the ideal
form of the tax and the relative ease with which the ideal could be approximated by simple, practical rules.
Furthermore, both studies noted that since existing personal income taxes have many features that favor
savings, a move to a consistent consumption base might well represent a less radical change than a move to a
comprehensive income base.
In this chapter I argue the case for a consumption base. "Arguing a case" is something that an economist
should perhaps avoid. Tax policy must reflect values, and economics as science cannot resolve the ultimate
clashes of interest that are bound to occur. Nonetheless,
Â
Page 4
I believe a strong case can be built for a consumption tax on the basis of values widely shared by the
participants in the debate.
In the first section of the chapter I

define the central concepts of
income and consumption. In the
second section I examine the
relative difficulty of implementing
a comprehensive income tax and a
comprehensive consumption tax.
This issue is given such high
priority because it is still
apparently widely believed that a
consumption base is more difficult
to put into practice than an income
base. 4 Surprisingly, this is not so.
Indeed, many of the most
troublesome aspects of income
taxation are avoided when
14
15
consumption is the base. Whereas

the discussion in the second
section is in terms of hypothetical
tax systems, in the third section I
look at the existing U.S. income
tax and show that some of its
more notable defects are in the
areas where the theory would lead
one to expect them. Together,
these two sections make it clear
that, because a satisfactory income
base is so much more difficult to
implement than a satisfactory
consumption base, the former
should be chosen only if there is
some compelling reason to do so.
In the final sections I consider two
major criteria for judging tax
systems: the degree to which they
distort resource allocation, and the
fairness of the distribution of
burdens they produce. I argue that,
while firm conclusions are hard to
establish, both efficiency and
equity considerations probably
favor the consumption base.5
Background: The Concepts of Income and Consumption
As only two years have passed since the Brookings conference on comprehensive income taxation,6 it is not
necessary here to devote much attention to the concept of income in its various versions. However, it is
important to have firmly in mind the relation between the income concept usually accepted in U.S. tax policy
debate, called Haig-Simons income, and the concept of consumption, and to relate both to the accounting
systems necessary to translate them into administrable tax bases.7
Haig-Simons income is usually defined in terms of the uses to which the tax unit puts its resources during the
year, according to the familiar accounting identity
Â
Page 5
where Y, C, and ∆W stand for

income, consumption, and change
in net worth ("savings") over the
accounting period, respectively.
To employ income as a tax base
requires putting operational flesh
on the concepts of consumption
and savings. For example, it must
be determined whether outlays for
15
16
medical treatment constitute

consumption. Similarly it must be
decided what sort of wealth will
be included in net worth; for
instance, human capital is usually
not included. It should be
emphasized that the terms
"consumption" and "wealth" are
not operationally defined a priori;
they are defined in the process of
determining tax policy.
To calculate either a consumption or an income tax base, it is normally most convenient to work from the
taxpayer's receipts rather than from his outlays. This can be seen most simply by starting with the assumption
that the wealth entering the definition of the tax base is an asset like a savings account, with readily
identifiable current yield, r. Let W1 stand for the wealth at the beginning of period i, and E1 for the
"nonwealth receipts" (wages, transfers, and so forth), understood as occurring at the end of period i, at the
same time as returns on wealth, rWi, and consumption, C1. For these concepts to form a satisfactory
accounting system, it is necessary that
or
The left-hand side of 3 is Haig-Simons income; the right-hand side, the sum of nonwealth receipts and returns
on wealth accumulated up to date i, is the usual calculation base. The same approach is normally taken to
calculate a consumption tax base; savings are subtracted from the sum of nonwealth receipts and returns on
wealth:
But an important simplification may be effected for a consumption base calculation by substituting for it an
equivalent in present-value terms. (I shall return to this point later.)
For the accounting system to balance, a certain consistency among consumption, wealth, and nonwealth
receipts is required. If it were determined, for example, that outlays for medical care should not be counted as
consumption for tax purposes, accounting consistency
Â
Page 6
would require subtracting these expenses from nonwealth receipts in the period (that is, these expenses
should be regarded as negative nonwealth receipts). Consistency also tells us how to treat returns on wealth
where the asset is of a type different from the standard savings account. For instance, if the discounted
present value of an inheritance at the beginning of period i were to be counted as part of wealth at the
beginning of period i - 1, the inheritance should not be counted in period i as a nonwealth receipt. 8
The difference between an income base and a consumption base lies entirely in the treatment of savings. It is
sometimes erroneously suggested that gifts and bequests received would be treated differently in income and
consumption accounting. This is not so. Assuming their anticipated value is not counted in wealth, these
receipts are simply included in Ei, in the appropriate period. Referring to accounting relationships 3 and 4,
16
17
one sees that such an increment in nonwealth receipts leads to the same change in both Haig-Simons income
and consumption.
The treatment of gifts and bequests given is less clear, since it may be argued that amounts given away are
not "consumed." If this approach were taken, the corresponding amount would need to be deducted from
nonwealth receipts in the period in question, an adjustment much like the one to exclude medical expenses
from consumption. Whether or not one thus removes gifts and bequests given from consumption has nothing
to do with the choice between income and consumption bases, so that in what follows I assume either that
gifts and bequests given are not regarded as consumption or that gifts and bequests given are nil.9
One conceptual problem with Haig-Simons income should be noted: the appropriate treatment of changes in
wealth associated with changes in the discount rate. The problem arises when wealth for tax purposes
includes the present value of future receipts (unlike the savings account example), as with a perpetuity bond.
To make matters simple, consider a taxpayer who holds a perpetuity yielding $10 a year and who consumes
exactly $10 a year. If the interest rate drops from 10 percent to 5 percent, his wealth will instantly jump from
$100 to $200, generating Haig-Simons income, even though his potential rate of steady consumption does not
change at all.10 This possibility is perhaps not unimportant in a period, like the present, of highly variable
interest rates.
Â
Page 7
The Relative Difficulty of Implementing Income and Consumption Bases
When turning from definition to practical implementation, one might expect consumption to pose greater
problems than income (since one normally reckons consumption by subtracting savings from income). Yet I
shall argue here 11 that precisely the opposite is the case: a consumption tax can be readily constructed on the
basis of current-year cash transactions only, virtually all of them also used in an income tax, while dispensing
with the elements of income calculation that are based on transactions in the (sometimes distant) past and are
designed to approximate unobservable quantities. These account for many of the most irksome features of
income taxation in practice.12
In considering the problems of implementation one should keep in mind desirable characteristics of a tax
accounting system:13
The transactions used to build up the tax base should be objectively observable. Imputed transactions should
be avoided.
The period over which records of transactions need be kept should be short.
The rules for constructing the base from recorded transactions should be understandable to ordinary people.
Complex transformations of historical data, such as inflation adjustments to put dollar figures on a
comparable basis, should be avoided.
All these criteria could be reasonably satisfied by a Haig-Simons income base implemented along the lines
suggested by equation 3 if the world resembled the simple model of savings deposit wealth and cash
consumption outlays. In practice, unfortunately, instead of using actual annual transactions one must deal
with a complex array of imputations to arrive at an estimate of the left-hand side of 3. Such imputations may
be subdivided into in-kind or "direct" nonwealth receipts in the form of (a) consumption services (for
example, use of the company car) or (b) increments of wealth (for example, additions to the value of pension
rights other than those arising from implicitly accruing earnings on pension wealth), and direct returns from
wealth, again in the form of (c) consumption services (for example, use of an owner-occupied house) or (d)
asset value changes (for example, depreciation of a piece of equipment). The search for practical methods of
17
18
making those imputations simple enough to permit
Â
Page 8
self-assessment by taxpayers and effective enforcement by the revenue agency has long bedeviled the
formulators of income tax rules. Most of the complexity and much of the deviation of the actual personal
income tax base from the Haig-Simons ideal stem from the difficulty of measuring consumption services
directly rendered by assets and, more important, measuring accruing wealth changes not reflected in current
transactions.
A number of difficult imputations, such as the notorious three-martini lunch, fall into category a. Since all are
as hard to deal with under a consumption base as under an income base, I devote no further attention to them
here. Problems with other imputations will be evident in the discussion that follows. Imputations c and d, in
particular, are necessary to measure what is usually called "income from capital."
Accruing Wealth Changes
For those forms of wealth that do not yield any direct consumption service, it is clear that the savings account
model is the exception rather than the rule. Particularly under inflationary conditions the cash flow from an
investment is a good measure of its yield only by coincidence. But even with stable prices, cash
flowdividends, coupon payments, and so forthis best regarded as simply a change in form of a portion of the
portfolio, to cash, and not as an adequate measure of yield. To use the right-hand side of equation 3 to
measure Haig-Simons income, one must track the market value of the components of the balance sheet.
The rules of thumb by which this counsel of perfection is approximated havenot surprisingly in view of the
stakes involvedproved enormously troublesome. The most prominent issues are depreciation (the allowance
for the reduction in value of assets due to wear and tear, obsolescence, and the like), capital gains (actually
unrealized increases in value of all kinds, but usually those associated with earnings retained in corporations,
changed expectations about the future, or changed discount rates), and accruing values of claims to future
payments like pensions and life insurance.
Depreciation. Although attempts have been made to determine patterns of depreciation through the
econometric analysis of used asset markets and although the U.S. Treasury attempts to collect systematic
information about actual use patterns of assets under
Â
Page 9
the asset depreciation range (ADR) scheme of depreciation allowances, remarkably little is known about the
degree of approximation of such allowances to true depreciation ("economic depreciation"). 14
Capital gains. Another group of changes in asset values not reflected in current market transactions is called
"accruing capital gains." To the economist these value changes are no different from those labeled
"depreciation," but U.S. tax accounting makes an important distinction between the two.15 Though it might be
possible to measure these accruals currently where active markets exist, as in the case of publicly traded
common stock in corporations, such procedures have not interested practical men (in part because they do not
seem to consider genuine the wide swings in wealth that these valuations imply). For real estate one could
imagine a system of self-assessmentbacked up by some sort of implied willingness to sell at the assessed
18
19
valuebut such an approach has never been suggested for an income tax and has never drawn support in the
local property tax administration setting. The difficulty of obtaining annual valuations and the potential cash
flow problems for taxpayers with large accrued income but no cash income have generally led to the
acceptance (for example, in Blueprints and in the Meade committee report) of a realization basis for capital
gains accounting. Although schemes can be designed to approximate the effect of accrual taxation when the
tax is collected on realization (as is proposed in the Meade report),16 these involve considerable complexity.
Retirement rights and life insurance. When retirement rights are clearly vested in individuals, and particularly
when they are "funded," they are essentially shares in mutual funds. They present an additional problem for
accrual taxation: the need to take into account possible restrictions on access to the market value of the
claims, for example, prohibition of or penalty on withdrawal, and limitations on the use of these assets as
loan collateral. With such restrictions the appropriate valuation of the rights depends upon the intention of the
holder and is thus even more difficult to approximate for tax purposes than the accruing value of other
securities.
Much more difficult is the evaluation of pension right accruals under "defined benefit plans," whereby
employers agree to provide retirement payments based on a formula related to earnings. Accurately
measuring individual wealth in this form is totally out of the question, as is the valuation of accruing social
security retirement
Â
Page 10
rights (a significant part of wealth in the United States). Life insurance presents similar, if less extreme,
measurement problems.
Corporations. Pensions and life insurance schemes are but two examples of the problem of allocating
investment earnings accruing within institutions to the individuals whose wealth is represented. This same
problem arises for all corporate income. Ideally, individual income accounts should encompass the
performance of corporations in which the individual has an interest. That such an imputation of corporate
income to shareholders is not attempted is presumably the chief justification for the separate tax on corporate
income, despite its incompatibility with the principle that income should be taxed according to the
circumstances of the individual. The Brookings conference on integration of the individual and corporation
income taxes showed how complex it would be to impute corporate income to its ultimate beneficiaries in a
satisfactory way. 17 None of the proposals considered recently deals with the problem of imputing to
shareholders earnings retained in the corporation, and all accept the imperfections in income measurement
rules already noted. Blueprints does offer a system of full integration of corporate and shareholder accounts,
and the Carter Commission in Canada also proposed the allocation of retained earnings to shareholders, but
those plans have not drawn much support.18
Advantage of a consumption base. It is in dealing with problems of the sort described above that a
consumption base has its most obvious administrative advantage. Under a consumption tax accruing wealth is
wholly irrelevant. There is therefore no need to measure it, no need to estimate depreciation, accruing capital
gains, or accruing rights in pension systems or life insurance policies. There is no need to measure the effect
on shareholder wealth of retained earnings or of any other events at the corporate level. If no cash transaction
takes place, there is no need to be concerned about those forms of wealth in calculating the base of a
consumption tax. It is that simple.
Wealth Yielding Direct Consumption Services
Owner-occupied housing is the most obvious, and quantitatively the most important, asset that provides most
of its yield in the form of direct consumption services. For income tax purposes the owner-occupier should be
regarded as being in the business of renting himself housing; then the income from this asset would be
19
20
measured
Â
Page 11
in the usual way. Otherwise some other method of imputing the return must be adopted if a portion of the
income is not to go free of tax and an incentive set up for people to buy "too much" housing and to adopt "too
often" the tax-favored form of tenureowner-ship. Although housing is the prime example, many others exist:
automobiles, boats, household durables, works of art, jewelry, and the like, besides the services of banks that
are paid for by interest forgone on checking accounts.
For a while the United Kingdom employed a system of imputing income to owner-occupied dwellings, but it
was discontinued. Although a variety of imputation schemes (for example, requiring evaluation of assets and
applying a standard assumed rate of return) can be imagined, the practical problems of taxing these direct
consumption yields are usually regarded as formidable enough to rule out the attempt. 19
For these forms of wealth, the consumption base would face exactly the same problems as the income base
were it not for the availability of an alternative accounting methodone that yields an equivalent tax base.20
Under the standard approach an investmentsay, in a houseis subtracted from cash receipts from whatever
source, whereas all return on the investment is included in the tax base. This approach would require the
unattractive imputation of a value to consumption services directly flowing from the asset. Under the
alternative approach the investment is not subtracted from the base, and none of the subsequent return flow is
included. Because this means that the actual tax liability occurs at the time of the investment (which is, in
effect, treated as part of consumption), Blueprints calls it the "tax prepayment" approach.21 Given
maximizing behavior by the investor, the two methods will produce a tax base of the same discounted
value.22
By obliging taxpayers to use the prepayment approach in accounting for assets yielding direct services, a
consumption tax can achieve at a stroke the neutrality between these investments and business investments
that eludes the income tax, while still maintaining a simple, cash-transaction-data-only information
requirement. Note, incidentally, the importance of timing. The tax prepayment approach is precisely the one
normally accorded consumer durables under an income tax, but it gives the wrong answer in that case. One
might describe the result as allowing the corresponding consumption to go tax free; yet the fact that the
implicit return goes tax free explains
Â
Page 12
why the same calculation gives the right answer under a consumption tax.
Making the System Inflation Proof
As is now widely understood, thanks to the Brookings conference on inflation and the income tax, two
problems need to be distinguished. 23 One is the tendency for taxpayers to be pushed up through the
progressive rate schedule by nominal income increases that exceed real income increases. The other is the
failure of income measurement rules that function satisfactorily under stable prices to continue to give a
reasonable approximation to real income when prices are changing. The first problem is relatively easy to
solve, by indexing the parameters of the tax calculationfor example, exemptions, marginal rate bracket
boundaries, credits. The same method applies to either an income or a consumption base. The second
20
21
problem is more difficult. As the discussion at the Brookings conference showed, there is no professional
consensus on how to carry out a full-fledged indexing of the measurement of an income base. Yet it is clear
that the rates of taxation of various forms of investment depend strongly on the rate of inflation, even when
the redistributions caused by changes in prevailing rates of inflation are ignored.
Problems of income measurement in a period of inflation arise because of the need to use market transactions
made at different times. In calculating gain from sale, for example, it is necessary to subtract the purchase
price from the current sales proceeds. This means subtracting apples from pears if the purchasing power of
the unit of account varies, as it may do substantially over a protracted period of inflation like the present one.
Inflation also produces an "inflation premium" in interest rates; a fully satisfactory real income accounting
calls for correcting these prices as well, a complex matter involving assumptions about the future rates of
change in the price level.24
As has been pointed out by Feldstein, in a flat-rate income tax system certain relatively simple corrections
could be madeindexing capital gains, depreciation, and inventoriesto obtain a satisfactory income base.25 In a
system allowing such adjustments, nonindexed financial yields would include an element of inflation
Â
Page 13
premium and an element of compensation for the tax on that premium. 26 In a system, like the American one,
that regulates the return on savings in certain formsespecially the Regulation Q restriction on savings account
interestthis response of financial yields to inflation may be prevented. In any case the ability of financial
markets to compensate for the incorrect income measurement no longer precisely holds for progressive
taxation, as high-bracket taxpayers would pay too much on their inflation premium and low-bracket
taxpayers too little.27
In contrast, inflation does not pose any problem for the consumption tax, because the calculation of the base
involves only current year transactions. The tax base and tax liabilities are always measured in consistent
units. To illustrate, consider a person who purchases a share of stock for $100 that grows in real value to
$110 by the next period, when it is sold and the proceeds are consumed. If his tax rate is 50 percent, under the
standard treatment he sacrifices $50 of consumption in the first period in return for $55 extra consumption in
the second period. If he chooses the tax prepayment approach, he sacrifices $100 of consumption in the first
period in return for $110 extra consumption in the second period. The same 10 percent return is obtained
under both treatments. If the price level doubles between the two periods, and the price of the stock with it,
the second period outcomes will be $110 and $220, respectively, under the two treatments, exactly the same
in real terms as when there is no inflation.
Averaging
The problem of averaging arises when a progressive rate structure is applied to a base that is calculated
periodically. Normally it is taken for granted that a person with a fluctuating income should not bear a
heavier burden of income tax than someone with the same average income experienced at a steady rate. It is
not immediately obvious how this objective should be made precise, that is, what the ideal averaging scheme
would be. One possibility would be to arrange matters so that income is taxed at a uniform rate over the
taxpayer's lifetime, with higher rates for those with ''better" income streams. Implementing this would be a
complicated matter under an income tax, and, as far as I know, no averaging scheme in use attempts to
achieve it.28
Â
21
22
Page 14
Interestingly, such an outcome would be exactly what one would expect under a consumption tax in which
the two methods of calculating the basestandard and tax prepaymentare simultaneously available, at least for
a reasonably broad class of investments. As shown above, the two methods lead to the same present value of
the tax base for any given stream of true consumption. 29 It follows that if the tax rate applied to the portion
of the base reported in a given year were higher than in other years, the taxpayer could reduce his burden in
present-value terms by rearranging his pattern of saving and dissaving in the two allowable forms. For
example, a taxpayer whose marginal rate this year is 20 percent and who expects next year to be subject to a
marginal rate of 50 percent will naturally use the tax prepayment method for his investments. But he will do
more than that: by borrowing according to the standard method this year, raising this year's tax base, and
investing on a tax prepaid basis, and then reversing the process next year, he can further reduce his tax
burden.
The gain from this profitable arbitrage would be limited by the progressive rate schedule. Thus, as the
arbitraging process continues, the present-period marginal tax rate rises, while next year's falls, and the
process stops when the two rates are just the same, somewhere between 20 and 50 percent.
In the simple world of a uniform and certain rate of return, with unlimited borrowing and lending capabilites,
this self-averaging system produces just the right result: the present value of the tax burden is a function of
the present value of lifetime consumption, regardless of the time pattern of earnings or the other features of
the lifetime plan. When borrowing and lending are limited, matters become less clear-cut; nevertheless, the
general principle would seem still to hold.30
Uncertainty introduces a
distinction between anticipations
and results that has interesting
implications. Since the outcome
of the investment is not known in
advance, the taxpayer cannot
assure himself of perfect
averaging ex post. Instead in
every period he will seek to
optimize ex ante. Roughly
speaking, the tax burden will be
related progressively to the
expected value of the taxpayer's
consumption stream, as viewed
from his perspective. It may be
debated whether this ex ante
progressivity is as desirable as ex
post progressivityI find it rather
appealingbut there is
unfortunately not the space to
pursue that question here.31
Â
22
23
Page 15
Defects of the U.S. Income Tax
It goes without saying that anyone's list of the defects of the U.S. tax system would include many items not
germane to the choice between income and consumption bases. However, some of the more egregious faults
are relevant for two reasons: (1) they arise from just those measurement problems noted in the previous
section that are very difficult to solve in an income tax but easy to solve in a consumption tax; and (2) they
reflect the apparent preference of Congress for a system that taxes savings more lightly than a comprehensive
income tax would.
Accruing Wealth Changes
The United States subjects ordinary business investments to an extremely complex and varied set of tax
treatments. The major features of the law deserving mention are as follows.
The investment tax credit pays for up to 10 percent 32 of the cost of new equipment (without any reduction in
the schedule of depreciation allowances).
Although no one knows how to

measure "true" economic
depreciation, the depreciation
deduction allowed under U.S.
tax laws is widely believed to be
too generous. This is especially
the case in certain industries,
notably real estate. Often the
readily visible tax shelter
"industry" provides evidence
that, indeed, the depreciation
allowances are excessive.
Certain forms of investment are written off immediately, notably outlays for research and development or
mineral exploration (not embodied in personal property), and selling expenses.
Equity investment in corporations is subject to a separate tax that is unrelated to the circumstances of the
investor; dividends are "double taxed."
Gains and losses from holding assets are not taxed as they accrue; gains and losses "realized" from the sale or
exchange of assetscapital gainsare subjected to the income tax according to special rules. The most notable of
the rules is the allowance of a deduction for 60 percent of the gain on assets held long enough. On the other
hand, severe limits are placed on the netting of realized losses
Â
Page 16
against other elements of income. It has been estimated that the need to distinguish capital gains from
ordinary income accounts for half the text of the individual income tax statute.
Many forms of retirement saving receive preferred tax treatment. Employer contributions to qualified
retirement plans are not included in the income of the employees, although the full amount of the ultimate
withdrawal on retirement is subjected to tax at that time. This is the classic method of implementing a
23
24
consumption tax; the effect is to exclude all the accruing yield on retirement wealth from income taxation.
Some other forms of retirement saving receive similar treatment, notably saving carried out through an
individual retirement account (for those with inadequate employer-provided plans) or a Keogh plan (for the
self-employed). The amounts of such tax-preferred saving allowed are, however, limited.
The buildup of life insurance wealth is also free of income tax. Earnings on the reserves of life insurance
companies are free of corporation income tax, and no allocation of accruing wealth to policyholders is
registered in the individual income tax calculations.
Interest from state and local government bonds is not included in the income calculation for tax purposes.
Reasons for the Provisions
The extraordinary variety of special provisions has no single explanation. But it is clear that a combination of
a policy preference for investment and sheer income measurement problems has played a central role. Of all
the features listed, the investment tax credit is most clearly designed as an incentive. No doubt the desire to
encourage investment has helped determine the rules for depreciation allowances and write-offs for research
and other activities, but the difficulty of measuring the asset value changes these allowances are designed to
represent makes them vulnerable to erosion through the political process. Businessmen claim they need more
realistic allowances; there is little evidence that they are wrong.
Whatever the origins of a separate tax on corporationsit predates by four years the twentieth century
incarnation of the federal tax on individual incomesit can be rationalized as an answer to the difficult problem
of allocating to individual equity owners the income accumulated within the institution of the corporation.
The separate tax has serious disadvantages, though, viewed as a compo-
Â
Page 17
nent of a system relating tax

liabilities to individual
circumstances. The anomalous
treatment of dividendswhich are
after all merely a change in
portfolio composition, to
cashespecially when compared
with the treatment of capital gains,
puts a premium on retentions or
tax-free routes for funds out of the
corporation. Indeed the scale of
dividend payments in the United
States is a continuing puzzle to
analysis, in view of the significant
tax disadvantages of that payout
method. In addition, there are the
limitations on the use of tax
credits, on loss carry-forward, and
so forth. All these rules taken
together have undoubtedly had a
marked effect on the financial
structure and corporate
organization of the United States.
24
25
The corporation income tax, in turn, provides one justification for the favorable treatment of long-term
capital gains. Though by no means all the capital gains realized and recorded on tax returns are earned on
corporate stock, the share is substantial, and a reduced rate on such gains compensates for the corporate
income tax paid on the underlying retentions. The special treatment of capital gains occurs in the first place
because of the difficulty of measuring accruals in the absence of cash transactions. Traditionally, the capital
gains preference has been justified as well by its incentive effect on investment, particularly risky investment.
Presumably, the nontaxation of

accruals of retirement benefits
and of wealth in the form of life
insurance has its origin in the
income measurement problems
noted above. That policy, too,
has no doubt had a profound
effect on the financial structure
of the United States, encouraging
the concentration of savings in
financial institutions.
Finally, the fact that state and local bond interest is tax free can be explained historically as a phenomenon of
U.S. constitutional law. Most legal commentators now regard that reason for the exemption as unsound; the
support for the exemption comes from state and local government officials and wealthy individualsfor
obvious reasons having nothing to do with either income measurement or any explicit policy to lighten the
tax burden on savings.
Shelters. Tax-exempt bonds provide a good example of the kind of problem that occurs because of the lack of
a consistent definition of income. In equilibrium, the interest on tax-exempt bonds falls below that on taxable
bonds, with the percentage difference representing the marginal tax rate at which it pays to switch savings
from one form to the other. Assuming for simplicity that it is possible to borrow at the taxable interest rate, it
would pay a person whose
Â
Page 18
marginal tax rate is above the boundary level, but who has no savings or who prefers to hold savings in other
forms, to borrow at the higher taxable interest and lend at the lower tax-free interest. In effect, a profit is
made on the tax system, with part of the benefit going to the state or local government borrower. The
operative mechanism is a "sheltering" of other incomeusually income from personal servicesby the interest
deduction. Such profitable arbitrage could continue until the individual's taxable income fell enough to drive
his marginal rate down to the level where the two returns do not differ.
It is debatable whether there is anything objectionable about this arbitrage. After all, what is accomplished is
to reduce the top marginal rate on earned income to the rate paid (implicitly) by owners of existing wealth on
tax-exempt bond interest. Congress, however, did not take that view and enacted a prohibition against
deducting interest on borrowing for the purpose of purchasing or holding tax-exempt bonds. Although to
economists it has an odd ring, the attempt is actually made to implement the policy through rules for "tracing"
the source of funds to purchase or hold tax exempts. Such rules are typical of the response of tax-law makers
to the pressures created by the inconsistent definition of income. Even when the income measurement
problem could be solved directly (as in this case, by taxing state and local bond interest), lawmakers tend to
enact special rules to counter the perceived "abuse." Thus, for example, the systematically excessive
allowances for depreciation in certain sectorsnotably real estateaccount for much of the complex anti-shelter
25
26
legislation.
To see how incorrect measurement rules may affect tax burdens, it may be helpful to look at the functioning
of a shelter that uses the depreciation allowances. Suppose that the rules are equivalent to allowing the
immediate expensing of one-half the investment in question, and suppose that the going interest rate is 10
percent. This means that by putting up $100 a taxpayer in the 70 percent bracket can buy roughly $154 worth
of the investment project, with the government putting in the extra $54 in tax relief (70 percent of the
deduction of 1/2 of $154). If the investment yields 10 percent, the high-bracket taxpayer can pay a deductible
$10 a year on $100 of borrowed money and pocket $1.62 a year ($5.40 Ã [1.0 - 0.7]), with no actual
personal investment at all. It is a case of pure arbitrage profit through the tax system; preventing it requires
burdensome
Â
Page 19
rulesfor example, appropriately limiting interest deductionswhich are bound to leave loopholes.
Imputed Consumption Yield
Under U.S. tax law no effort is made to capture the yield in the form of direct services from investment in
owner-occupied houses or other consumer durables such as paintings and jewelry, while the price of services
of this sort purchased on the market must support income tax. 33 The exemption of the yield on
owner-occupied housing, in particular, has long been the target of reformers.
Inflation
The sensitivity of the distribution of real tax burdens to the rate of inflation is also a problem in the current
U.S. system. The tendency of inflation to reduce the real value of depreciation allowances based on historical
costs has been a powerful argument of accelerated methods. Similarly, objections to the taxation of nominal,
inflation-caused increases in the value of assets, when real returns are small or negative, has much to do with
the political support for lightly taxing capital gains. That neither of these ways of dealing with inflation is at
all satisfactory has not prevented their adoption.
Until recently, there has been little interest in more systematic indexing. This lack is usually attributed to the
desire of Congress to claim credit for tax "cuts," made necessary by the fact that inflation pushes households
up the progressive rate schedule. However, as noted above, the mismeasurement of the basis for capital gains
and of depreciation allowances tends to raise the effective rate of tax on the return to savings.34 Present rules
also lead to capricious variations in the distribution of the tax burden. In analyzing data from tax returns
assembled by the Treasury, Feldstein and Slemrod found an example of this capriciousness by comparing real
and nominal capital gains reported.35 Although the aggregate gain realized from the sale of corporate stock
amounted to $4.6 billion in the year analyzed (1973), the real gain was approximately minus $0.9 billion (in
1973 dollars). Furthermore, the difference between real and reported gain varied systematically by income
class, with a close agreement between the two at the top of the income distribution and a wide divergence at
the bottom.36
Â
26
27
Page 20
Averaging
Few people regard averaging as a major problem of the U.S. income tax. But it has had an important bearing
on policy as an influential argument for the special treatment of capital gains and has probably helped prevent
the inclusion in the tax base of gifts and bequests received. Those elements of the Haig-Simons income
calculation require averaging provisions because of their typically large size relative to the rest of the income
flow. 37
Explicit provisions for averaging do exist in the individual income tax laws, and an extensive literature deals
with the best way to accomplish it. The present rules are simple enough for the taxpayer who does not change
his family status, but they are limited and useful only to even out increases in the individual base, not
decreases.38
Summary
I have identified three broad problem areas in the present U.S. income tax system: first, the extremely varied
taxation of the yields from ordinary investment, depending on the industrial sector, the form of business
organization, the type of financing, and the purpose of savings; second, the inconsistent taxation of the yield
from owner-occupied housing and other consumer durables; and third, the sensitivity of the measurement of
the real tax base to the rate of inflation. Correcting these defects fully to meet the Haig-Simons income ideal
is universally conceded to be impractical. Various studies, Blueprints among them, have suggested ways to
approximate this ideal. But there is little evidence that Congress wants to do so. What is clear is that the
present half-way status, with its chains of ad hoc remedies to problems arising from the basically inadequate
income measurement rules, is unsatisfactory.
Efficiency and the Choice Between an Income Base and a Consumption Base
Economic efficiency is commonly regarded as one of the principal reasons for preferring a consumption to an
income base for taxation. The disincentive to save resulting from the taxation of the returns to saving under
an income tax is eliminated under a consumption tax.39
Â
Page 21
However, all practical tax systems involve some disincentive effects, so the efficiency issue in the choice
between income and consumption taxation is the relative seriousness of the distortions under the two. As the
discussion just concluded makes clear, the actual income tax in the United States must be thought of as a
highly complicated array of different taxes on various transactions, but the analysis here must be confined to
simple models permitting only combinations of pure income and consumption taxes.
The efficient use of resources requires that the rates at which different desirable goods and services are traded
off for one another through changes in production just equal the corresponding tradeoffs in the preferences of
households. Otherwise a reallocation of resources is available that would produce a better outcome for all
households. In the absence of taxes, competitive forces bring about efficient resource use because firms and
households face the same prices. But when taxes must be levied to raise revenue, firms and households are
presented with different prices, and equilibrium is no longer efficient. The problem of tax policy is how to
make as close an approach as possible to efficient resource use and still raise the necessary revenue through
price-distorting taxes. This is typical of what has become known in the economics literature as a "problem of
the second best." 40
27
28
Second-Best Taxation in a Life-Cycle Saving Model
The essence of the problem is

captured by a model in which a
person lives for two periods,
working and consuming in the
first period and consuming
from the proceeds of his
first-period savings as a retiree
in the second period.41 The
budget constraint of this typical
taxpayer is then
where
C1 = consumption in period 1
C2 = consumption in period 2
L = labor supplied in period 1
r = interest rate
Â
Page 22
w = wage rate
tr = tax rate on interest
tw = tax rate on wages.
This budget constraint reflects a flat-rate income tax if tr = tw, and a pure consumption tax when tr = 0. (The
equivalence between a flat-rate tax on consumption and one on nonwealth receipts is immediately verified by
setting tr = 0 and dividing through 5 by 1 - tw. The result is the budget constraint for a person facing a flat
consumption tax at rate tc = tw/1 - tw. Government revenues are also the same in present-value terms under
either scheme.)
The analysis of the second-best problem in this case is made more transparent if the tax on interest is shown
as an equivalent tax on second-period consumption, so that the household's budget constraint becomes 42
where t2, the tax on second-period consumption, is related to tr by
The problem is to choose tw and t2 in such a way as to do the household as little harm as possible while
ensuring some specified total of tax revenue. In mathematical terms this is a straightforward problem, and it
28
29
can be shown that a solution will involve the following relationship between tw, and t2,
where the σ1j are parameters describing the household's preferences among the "commodities" of the
problem, first- and second-period consumption, and labor. Economists will recognize them as the
compensated elasticities of the demand for future consumption and the supply of labor.
It should not be surprising that this analysis will imply either pure income or pure consumption taxation only
by coincidence. The second-best optimum depends upon the elasticities of demand (including the demand for
leisure). This simple model, for example, would
Â
Page 23
imply a consumption base if the compensated supply elasticity of labor (necessarily nonnegative) were just
balanced by a positive responsiveness of the (compensated) labor supply to the price of retirement
consumption. Such a balancing relationship is not wholly implausible. It holds, for example, if individual
preferences imply that an increment of wealth will be used to purchase the three commodities in this problem
in proportion to their original quantities, as in the Cobb-Douglas utility function so familiar to economists. 43
Such a simple model can do no more than offer the roughest sort of assistance to policy, but it does illustrate
the severe limitations of unguided intuition in such a complex matter. The Cobb-Douglas utility function
offers a particularly striking example, since it is the usual form assumed in life-cycle models of savings, in
part because it allows analytical simplification, but also because it has seemed plausible to researchers.44 In
this two-period model it leads, as noted, to the optimality of a consumption base. This holds despite the fact
that the (uncompensated) interest elasticity of first-period consumption is zero. In other words, private saving
is totally insensitive to changes in the yield on savings, a condition normally thought to favor income
taxation.45
Choosing Between Income and Consumption
Earlier in this paper I argued that the taxation of capital in the United States is highly variable because of
uneasy compromises that have been made between cash flow accounting and Haig-Simons income
measurement. The record suggests that, whereas a precise measurement of Haig-Simons income may be
difficult, imposing a tax on the return to capital at a rate bearing any systematic relation (other than full
equality) to the tax rate on labor earnings is hopeless. It therefore might be argued that the proper question is
not what the optimal ratio between tax rates on labor income and capital income is, but what the choice
between a pure income tax and a pure consumption tax should be.
In this 1978 paper Feldstein analyzes that question by constructing an estimate of the welfare gain from
shifting from a tax of 40 percent on both capital and labor earnings to a tax of equal yield on labor earnings
alone. Feldstein employs a set of assumptions that roughly simulate the U.S. economy, including the
assumption that the (uncompensated) interest elasticity of savings is zero.46 He concludes
Â
29
30
Page 24
that the differential loss from the income base amounts to 18 percent of the revenue collected from the tax on
capital income. 47 In the Feldstein formulation, if savings are more responsive to the rate of return, the
efficiency cost will be higher.
Interestingly, as far as the efficiency cost is concerned, Feldstein's formulation does not give any points to the
consumption base for what is sometimes considered its chief attraction: encouraging capital formation, where
capital is viewed not as the aggregate of individual wealth but as the factor combined with labor to produce
output. It is characteristic of second-best optimal tax calculations that when, as in this case, all transactions
are potentially subject to tax, only properties of preferences (demand elasticities) enter the solution
formulas.48 Variations in assumptions about the production system will therefore not affect expression 8 for
optimal taxes. But since the welfare gain calculations just reviewed treat the before-tax rate of return and
wage rate as constant, the results can be interpreted as the efficiency gain when the capital per worker-hour
does not change (as in the case of a small open economy, where exchange with a larger system leads to the
constancy of factor prices). Usually the price changes that would follow from relaxing this assumption would
act to reduce the efficiency effects. (For example, if before-tax prices changed to keep after-tax prices
constant, there would be no efficiency gain.)
Various authors have also examined the effect of wage and interest taxes on the properties of steady-state
growth paths. Much of this work has been summarized and greatly extended by Atkinson and Sandmo.49
Their analysis explicitly addresses the problem of compromising between equity and efficiency objectives
and is therefore more suitably taken up in the next section, which deals with equity arguments.
Summary
It is hardly to be expected that the conclusions drawn from such simplified models will be persuasive when
applied to a complicated real policy issue. The analysis of efficiency just discussed considers a choice
between two nonexistent ideals. In particular, the present tax system in the United States deviates so
markedly from an income ideal that the more important efficiency effects may well be those
Â
Page 25
related to the misallocation of productive resources across uses and within the household, rather than those
related to the life-cycle patterns of consumption and labor. 50 On the other hand, since this volume is
concerned with the goal toward which tax policy should aim, the comparison of the pure forms is relevant.
Granting the necessary qualifications, I conclude that the efficiency analysis reinforces the administrative
considerations in support of a consumption base against an income base.
Equity
If neither practicability nor efficiency leads to a preference for an income base, the case for the superiority of
an income tax must be founded on justice or equity. Probably most laymen believe that income represents a
fairer basis for taxation than consumption. This has also been the position taken by a number of influential
expert commentators, including Goode, Musgrave and Musgrave, and Surrey.51 In general the preference for
income is based on the idea that the accretion of economic power during an accounting period, which may be
allocated to either consumption or savings, represents a better measure of ability to pay than consumption, a
mere component of total accretion (though the part may be larger than the whole, since savings may be
negative).52 The comparable opposing view is that consumption, a measure of what people take out of the
economic system, is a more appropriate basis for taxation than income, a measure of what they contribute to
30
31
the economic system in productive performance.
These arguments are commonly considered persuasive in themselves. They are not, for example, easy to
relate to the utilitarian theory that generally underlies welfare arguments in economics. Thus there is no ready
answer to the question of why income, which is a measure of change in economic power (however that might
be defined), should be preferred to some measure of the level of economic power.53 Nor is it obvious how the
consumption tax advocate should deal with the view that it is the power to consume, not its exercise, that is
the proper basis for taxation. (The miserly millionaire who consumes nothing is usually trotted out at this
point.) Furthermore, neither of these lines of argument takes into account the link between income and
consumption over the life cycle.
Â
Page 26
Utilitarian Analysis of the Choice of Tax Base
Utilitarian theory has been used to explore the choice of a tax base. I have already noted the paper by
Atkinson and Sandmo that surveys and extends one branch of the application of interpersonal utility
assumptions to the selection of separate rates of tax on payments to labor and capital. 54 Their analysis
focuses on the effect of wage and interest taxes on the equilibrium capital-labor ratio in steady-state growth.
They embed the problem of individual maximization subject to a budget constraint of the form of equation 6
in a model of growth in which the generations overlap, like the model introduced by Samuelson and extended
by Diamond and others.55 In such models the capital accumulated by one generation becomes the factor
cooperating with the next generation's labor in production. The properties of steady-state growth paths are of
prime interest; the usual goal analyzed is that of choosing the path to maximize the utility of a representative
individual (or of a representative ''effective individual" in the case of productivity growth through technical
change).
It is important to recognize that in adopting such a goal these studies move from an efficiency analysiswith
efficiency defined strictly according to Pareto optimalityto an explicit interpersonal utility comparison. This
issue may be easily understood by looking at the case in which no tax revenue need be raised at all. In that
case the second-best analysis discussed above implies zero tax rates, these necessary properties of Pareto
optimality being referred to by Atkinson and Sandmo as "static efficiency conditions."56 By the familiar
golden rule reasoning the resulting path will not be Pareto optimal, however, unless the rate of return on
capital exceeds the rate of population growth.57 When that condition is not satisfied, an excess of capital
exists, so that some of the existing stock can be consumed currently at no cost to the consumption of future
generations. Except in such a pathological case, however, the path satisfying "static efficiency" will be Pareto
optimal. This is true even though the utility of the representative individual will not be maximized. The
conditions of production and exchange necessarily characterizing an optimum in the latter sense are called by
Atkinson and Sandmo those of "dynamic efficiency,"58 a somewhat unfortunate term, given the usual
meaning of "efficiency" in economics.
When the capital-labor ratio in such a model diverges from the golden rule level, it will usually be possible to
raise steady-state
Â
31
32
Page 27
utility by a combination of wage and interest taxes that preserve budget balance. 59 In the case of a
capital-labor ratio below the golden rule level (rate of return above the rate of population growth), a
combination of a tax on interest and a subsidy on wages may raise steady-state utility by causing the
capital-labor ratio to rise. Such a result occurs when preferences over combinations of first-period
consumption, first-period leisure, and second-period consumption can be described by the Cobb-Douglas
form of utility function.60 With these preferences, labor supply is independent of the wage rate, and the
fraction of the after-subsidy wage devoted to capital accumulation is constant, that is, totally insensitive to the
after-tax return.
This conclusion, that even with no net government revenue requirement welfare could be improved by taxing
interest returns to subsidize wages, is in striking contrast to the efficiency analysis with the same preference
structure, as discussed above. In the absence of a need for net government revenue, efficiency called for no
taxes at all, whereas a positive government revenue constraint implied a tax on wage income only. In the
Atkinson and Sandmo model the pure efficiency analysis characterizes optimal taxes only in the special case
where the associated capital-labor ratio is at the golden rule level.
Welfare Analysis with Government Saving
Thus far I have been assuming that the only instruments available to the government intent on maximizing
welfare are wage and interest taxes. The picture is dramatically changed when the government can issue debt.
If so, that instrument alone suffices to attain the golden rule capital-labor ratio (one can imagine the
government standing ready to borrow or lend at a rate of interest equal to the growth rate of effective labor),
and taxes are zero in the absence of a net revenue requirement. Any net revenue requirement implied should
be raised through wage and interest taxes according to the second-best efficiency arguments already
summarized.61 In other words, roughly speaking, when capital accumulation (or decumulation) is the aim,
government deficits or surpluses constitute the preferred instrument; pure efficiency arguments determine the
structure of distorting taxes, which will be needed only to finance real government expenditures.
Introducing government debt into the Atkinson-Sandmo system is analytically equivalent to allowing an
element of lump sum taxation
Â
Page 28
in the second-best problem

discussed earlier. 62 In their paper
on the design of tax systems,
Atkinson and Stiglitz point out
that it is in general unrealistic to
rule out lump sum taxes, since
most tax systems have similar
elements (for example, the zero
bracket amount in the U.S.
income tax).63 Admitting such
taxes into the usual second-best
optimality problem leads to the
expected conclusion: because of
the efficiency advantages, lump
sum taxes will be used to the
32
33
fullest extent possible. This is

because the usual analysis in
effect assumes that all people are
alike. As Atkinson and Stiglitz
emphasize, it is the distributional
function, inherently deriving from
differences in people, that creates
the need for taxes on
transactions.64
While recognizing that a wide range of potentially observable signalssuch as incomeof differences among
people might be used for assessing different tax burdens (thus ruling out conclusions of any great generality),
Atkinson and Stiglitz carry out an illustrative analysis in which they assume that people differ only in labor
effectiveness, reflected in differences in the wage rate received. Otherwise people are assumed to be identical
(for example, they have identical utility functions), and the prices of all goods other than leisure are the same
for all.65 In this model world the government is equipped with a full set of flat-rate transaction taxes and a
potentially nonlinear tax schedule applicable to earnings from labor.66 When taxes are set to maximize a
social welfare function, the rather remarkable conclusion emerges that if individual utility functions are
weakly separable between labor and all consumption goods taken together, no taxes other than the tax on
labor earnings need be employed. This condition says roughly that the rate at which one is willing to trade off
future consumption for present consumption is independent of the amount of labor time one puts in. Though
such a condition is not empirically established, it does have a certain plausibility (again, Cobb-Douglas
preferences provide an illustration), and this suggests the possibility for a fully satisfactory redistributive tax
that exempts returns to savings.
Nonutilitarian Equity Arguments
The main contribution of the optimal tax literature has been to make precise certain ideas based on utilitarian
premises. The paper by Corlett and Hague, a precursor of the modern optimal commodity tax theory, is a
classic example.67 But outside professional economics
Â
Page 29
circles the debates about tax equity are carried on in terms difficult to analyze in a utilitarian framework.
Frequently they hinge on precisely those differences in peopleespecially differences in taste, health, and so
forththat cast most doubt on the systematic interpersonal comparisons needed for the utilitarian approach.
Concepts of "ability to pay" and "horizontal equity" carry great weight. To conclude this section on the equity
issues, then, I turn to certain of these less formal but perhaps more persuasive arguments. 68
The general problem is how to distribute the burden of taxation among people who differ in many
respectshealth, sex, family status, employment conditions, and so on. It is perfectly legitimate to take any of
these differences into account. However, the principle of horizontal equity says that people who are regarded
as similar should bear the same tax burden, while the principle of vertical equity says that those who are
relatively better situated should bear a relatively greater share of the tax burden. The central questions are
when people should be regarded as similar for this purpose (and when one person should be considered better
situated than another), and how the tax burden should be measured.
Although there is no obvious answer to these questions, I would suggest that two related postulates are
compelling: (1) events over a short period of time are not an adequate basis for determining the relative
"deservingness to pay tax" of two persons; and (2) tax payments over a short period of time are not an
adequate basis for comparing the relative tax burdens of two persons. These principles are the more
33
34
compelling the shorter the "short period of time." None would suggest comparisons based on the events or
tax payments of a day or week. A decade seems more reasonable. Yet there is no obvious stopping point
short of basing comparisons on a lifetime of circumstances and on a lifetime of tax payments.
To move from these general ideas to a concrete analysis and to isolate the issue of a choice between an
income and a consumption base, assume that the lifetime pattern of the labor supply of individuals is fixed,
implying for each one a fixed time path of what was called above "nonwealth receipts." People may differ in
initial wealth, in the time path of nonwealth receipts, and in the use made of these endowments. How should
they be compared? Since people have the same consumption possibilities, it seems compelling that those
having the same aggregate of initial wealth and present value of nonwealth receipts should bear the same tax
burden, measured in
Â
Page 30
present-value terms. Put another way, because of the lifetime budget constraint, the principle of horizontal
equity calls for equal reductions in the discounted value of consumption by people for whom this total would
be equal in the absence of tax. The discounted value of nonwealth receipts added to initial wealth might be
labeled "lifetime wealth," the operative constraint on lifetime consumption. The principle of horizontal equity
amounts to the idea that people should be taxed according to lifetime wealth. Vertical equity, in turn, calls for
positively relating tax burdens to lifetime wealth (and, one might add, the principle of progressivity requires
that the tax burden, as a fraction of lifetime wealth, be larger the larger is lifetime wealth).
These principles of horizontal and vertical equity are realized by a consumption tax implemented along the
lines described above, including the potential for self-averaging by individuals through the alternative
methods of accounting for saving. By such a device each person will confront a constant rate of tax on his
annual tax base, with the level of the rate a positive function of his lifetime wealth. Further, by the use of a
progressive schedule of taxes on the annual base, the burden, measured by the present value of taxes
discounted at the rate of interest (the same before and after tax), will be progressively related to lifetime
wealth.
By contrast, an annual income tax generates tax burdens that are haphazardly related to individual
circumstances, as measured by lifetime wealth. A flat-rate income tax, unlike a flat-rate consumption tax,
systematically biases the distribution of tax burdens. Given two persons with the same lifetime wealth, the
income tax imposes the lighter burden on the one who consumes early in life, the heavier burden on the one
who postpones consumption. Given two persons with the same lifetime wealth, the income tax imposes the
lighter burden on the one whose nonwealth receipts occur late in life. If the income tax is levied according to
a progressive schedule, the relation between lifetime wealth and lifetime tax burden will be even less
systematic. Not only do the timing of receipts and consumption enter the result in the way described but also
other characteristics of their time profile now influence relative tax burdens.
All this discussion is based on a highly simplified example. Other aspects of individual circumstances might
be considered relevant. For example, to make a distinction between nonwealth receipts due to labor and those
due to inheritance or gift might be desirable. Attention to family status is presumably in order. Imperfect
capital
Â
34
35
Page 31
markets and uncertainty complicate the picture. Yet none of these complications seems to me to point to a
preference for income over consumption taxation.
Conclusion
This paper has examined the choice between income and consumption taxes from many angles. In every case
a preference, strong or weak, for the consumption base has emerged. In principle, it is much more practical to
implement a consumption tax than an income tax. A large number of the most troublesome problems vanish
under a consumption tax, and no new problems are evident. This would be a poor basis for choosing a
taxafter all, there are even easier taxes to implementif there were a strong case on other grounds for
preferring the income base. Yet it is clear that the two bases are broadly similar and basically attractive. And
when efficiency and equity are the criteria by which they are measured, if anything the consumption base
proves superior.
In this complex and uncertain world, it is unlikely to find prescriptions for tax policy that will please
everyone. Weighty arguments, pro and con, are to be expected. For this reason I am surprised at just how
persuasive the case for the consumption base is, on criteria that I believe are broadly accepted, and I find
puzzling the persistent attraction of the income base.
Notes
I wish to thank W. D. Andrews, A. B. Atkinson, J. A. Kay, M. A. King, M. S. Feldstein, and J. E. Stiglitz for
the benefit of discussions on the subject of this paper.
1. U.S. Department of the Treasury, Blueprints for Basic Tax Reform (Government Printing Office, 1977)
(hereafter Blueprints); Institute for Fiscal Studies, The Structure and Reform of Direct Taxation, Report of a
Committee chaired by Professor J. E. Meade (London: Allen and Unwin, 1978) (hereafter Meade committee
report).
2. In this context the terms "expenditure" and "consumption" are usually interchangeable. I shall hereafter
stick to ''consumption." There have been other recent contributions to the literature, including a 1976 report
prepared by Sven-Olof Lodin for a government commission in Sweden, available in an English translation,
Progressive Expenditure Taxan Alternative? A Report of the 1972 Government Commission on Taxation
(Stockholm: LiberFÃ¶rlag, 1978); a paper by William D. Andrews, "A Consumption-Type or Cash Flow
Personal Income Tax," Harvard Law Review, vol. 87 (April 1974), pp. 1113-88; and two papers by Peter
Mieszkowski: "The Choice of Tax Base: Consumption versus Income Taxation," in Michael J. Boskin, ed.,
Federal Tax
Â
Page 32
Reform: Myths and Realities (San Francisco: Institute for Contemporary Studies, 1978), chap. 2, and "On
the Advisability and Feasibility of an Expenditure Tax System," in Henry J Aaron and Michael J. Boskin,
eds., The Economics of Taxation (Brookings Institution, 1980).
3. See, among others, Joseph A. Pechman, Federal Tax Policy, 3d ed. (Brookings Institution, 1977), and
Pechman, ed, Comprehensive Income Taxation (Brookings Institution, 1977); George F. Break and Joseph A.
Pechman, Federal Tax Reform: The Impossible Dream? (Brookings Institution, 1975); Report of the Royal
Commission on Taxation, vol. 3 (Ottawa: Queen's Printer, 1966), chap. 8; Studies of the Royal Commission
on Taxation, no. 25 (Ottawa: Queen's Printer, 1967), and ibid., nos 26, 28, 29 (1968); Arthur B. Willis, ed.,
35
36
Studies in Substantive Tax Reform (Chicago: American Bar Foundation, 1969). Bossons provides a
discussion of why, even when Haig-Simons income is the ideal, the impossibility of taxing some components
may lead to the omission of others. John Bossons, "The Value of a Comprehensive Tax Base as a Tax
Reform Goal," Journal of Law and Economics, vol. 13 (October 1970), pp. 327-63.
4. Pechman in 1971 observed' "The expenditure tax is not more widespread than it is primarily because of
difficulties of compliance and administration. ... It is generally agreed that the administrative and compliance
problems of an expenditure tax are formidable and that it would be very difficult for most countries to
enforce such a tax with the present state of administrative know-how." Significantly, perhaps, no similar view
is found in the next edition of Pechman's book. Joseph A. Pechman, Federal Tax Policy, rev. ed. (Brookings
Institution, 1971), pp. 164-65, and ibid., 3d ed., pp. 66-68, 197-99.
5. Two subjects readers might expect to find addressed have nevertheless been omitted here: the difference in
macroeconomic performance of an economy with a consumption tax from that of one with an income tax,
and problems of transition from the present system to a consumption base. The latter, in particular, is an
unfortunate omission. For even though, as I believe, the advantages of a consumption tax written "on a clean
slate" are becoming more widely appreciated, such a tax may still fail as a policy goal if there is no
acceptable way to effect the transition. However, an adequate analysis of the subject is dependent upon the
particular rules by which a consumption tax is to be realized and is thus beyond the scope of this paper. The
Treasury Department's Blueprints does provide a detailed plan and includes, in chapter 6, a discussion of the
problems of transition.
6. Pechman, ed., Comprehensive Income Taxation.
7. In an excellent brief survey of the history of the income concept, Richard Goode points out that the
Haig-Simons definition was anticipated by Georg von Schanz in 1896 and suggests that the term S-H-S
(Schanz-Haig-Simons) be used. Without meaning any disrespect for Shanz, I shall stick to the by now
conventional label. Richard Goode, "The Economic Definition of Income," in Pechman, ed., Comprehensive
Income Taxation, pp. 1-36.
8. The difference is emphatically not just a matter of convention, however, when income is the tax base. The
earlier the present value of anticipated receipts is recognized as part of wealth, the larger the tax base in
present-value terms. That is simply because the implied returns on wealth are included in the base only when
wealth is accounted for.
9. In the terminology of Blueprints I assume a "standard of living" concept of income and consumption.
Â
Page 33
10. For a good discussion of this and other aspects of income definition, see the Meade committee report,
chap. 3. The Meade committee considered this defect so serious that it took as its ideal income concept
(regarded as impractical for actual tax purposes) a definition that treats the taxpayer as having enjoyed no
income at all from the wealth gain in this case. (See also Goode's discussion of definitions stressing capital
maintenance in "Economic Definition of Income," pp. 3-5.) This may go too far in the other direction, since
the taxpayer can enjoy a higher consumption if he is not planning to consume at a steady rate in perpetuity.
The problem is one of a changing relative price of future consumption. A similar difficulty arises in a
transition from an income- to a consumption-based tax, when, in general, an increase in the price of current
relative to future consumption takes place. The appropriate treatment of accumulated wealth then depends
upon the time pattern of planned consumption.
11. The discussion in this section draws heavily from Andrews, "A Consumption-Type or Cash Flow
Personal Income Tax"; J. A. Kay and M. A. King, The British Tax System (Oxford: Oxford University Press,
36
37
1978); Meade committee report; and Blueprints.
12. After discussing these problems, Kaldor concluded it would be impossible to design a fair income tax.
Nicholas Kaldor, An Expenditure Tax (London: Allen and Unwin, 1955; Westport, Conn: Greenwood Press,
1977), chap. 1.
13. For a detailed discussion, see Blueprints, pp. 42-49.
14. Econometric techniques have been addressed to this problem, using "secondhand" asset market data, by
Frank C. Wykoff and Charles R. Hulten, "Economic Depreciation and the Taxation of Structures in U.S.
Manufacturing Industries: An Empirical Analysis," OTA Paper 28 (U.S. Department of the Treasury, Office
of Tax Analysis, 1977).
15. Although economists use the term to refer to value changes whether or not accompanied by cash flow, in
U.S. tax law "capital gain" refers to the gain from sale or exchange of a capital asset. It is thus explicitly a
realization concept, and the notion of "accruing" capital gains is, strictly speaking, nonsense. I stick to the
economists' usage here, even though there are some advantages in the other. In particular, when the
realization notion is used it becomes clear that the accounting idea of capital gains is essentially a correction
to past mismeasurement of income, as reflected in an incorrect basis (the amount subtracted from sale
proceeds in computing gain) of the asset under the tax rules.
16. Meade committee report, pp. 129-35.
17. Charles E. McLure, Jr., Must Corporate Income Be Taxed Twice? (Brookings Institution, 1979), chap. 7.
18. Perhaps this is because they are not designed for the kinds of explicit tax incentives for investment found
in the actual U.S. system. The bulk of the discussion at the Brookings conference on integration, for example,
concerned the preservation of the character of "preference" income in the hands of shareholders. See ibid. For
the proposals, see Blueprints, pp. 68-75, and Report of the Royal Commission on Taxation, vol. 4, pp. 19-30.
19. William F. Hellmuth, "Homeowner Preferences," in Pechman, ed., Comprehensive Income Taxation, pp.
163-203; Blueprints, pp. 7, 85-89.
20. Naturally there is no sure way to separate the investment and consumption aspects of an outlay for a
consumer durable. That it eliminates the necessity of even drawing this distinction is one of the attractions of
the alternative accounting method.
Â
Page 34
21. Blueprints spells out in detail how the distinction between the two treatments could be implemented in an
easily policed manner (pp. 123-27).
22. This statement requires modification to take into account uncertainty, as will be discussed in the text.
23. See Henry J. Aaron, ed., Inflation and the Income Tax (Brookings Institution, 1976).
24. See, for example, the comprehensive discussion in John B. Shoven and Jeremy I. Bulow, "Inflation
Accounting and Nonfinancial Corporate Profits: Physical Assets," Brookings Papers on Economic Activity,
3:1975, pp. 557-98; and Shoven and Bulow, "Inflation Accounting and Nonfinancial Corporate Profits:
Financial Assets and Liabilities," BPEA, 1:1976, pp. 15-57.
25. Martin Feldstein,

"Inflation, Income Taxes, and
the Rate of Interest: A
37
38
Theoretical Analysis,"
American Economic Review,
vol. 66 (December 1976), pp.
809-20.
26. For example, with a flat
50 percent tax rate both
borrowers and lenders will be
in the same position with an
interest rate of 5 percent and
no inflation or an interest rate
of 15 percent and an inflation
rate of 5 percent. The 7.5
percent after-tax nominal
return in the latter case
amounts to 2.5 percent in real
terms, the same as is realized
with no inflation.
27. This has been discussed in detail by Martin Feldstein, Jerry Green, and Eytan Sheshinski, "Inflation and
Taxes in a Growing Economy with Debt and Equity Finance," Journal of Political Economy, vol. 86 (April
1978), pt. 2, pp. S53-S70.
28. For discussion and a proposal for lifetime averaging under an income tax, see William Vickrey, Agenda
for Progressive Taxation (Ronald Press, 1947; Kelley, 1972), chap. 6.
29. This statement is precisely true in the simple case of a uniform and certain return on savings. As has been
noted above (note 22), uncertainty introduces complications, some of which are discussed in the text.
30. A possible objection that the proposed system puts an undesirable premium on knowledge and forward
planning does, however, acquire extra force as the world becomes more complicated. Because of the great
complexity of the rules and great differences in the tax treatment of different forms of saving, the present
system is perhaps even more vulnerable to this complaint.
31. I have carried the analysis further in comment on the Meade committee report in David F. Bradford, "The
Meade Report from a U.S. Perspective," presentation at the Social Science Research Council/Public Sector
Study Group Tax Reform Symposium, University College, London, May 26, 1978.
32. Eleven and a half percent if one counts the extra credit for employee stock ownership plans.
33. Insofar as a wealth accretion results from holding these assets, it also goes untaxed except as realized in
the form of capital gains. Even this is allowed to go substantially free of tax in the case of housing, through a
"rollover" rule coupled with tax exemption up to specified limits after the owner reaches age fifty-five.
34. For an estimate of the size of this effect, see Martin Feldstein and Lawrence Summers, "Inflation, Tax
Rules, and the Long-Term Interest Rate," BPEA, 1.1978, pp. 61-99.
Â
38
39
Page 35
35. Martin Feldstein and Joel Slemrod, "Inflation and the Excess Taxation of Capital Gains on Corporate
Stock," National Tax Journal, vol. 31 (June 1978), pp. 107-18.
36. Ironically, the political reaction to the effect of inflation on capital gains has most recently taken the form
of a reduction in rates on the gains experienced by high-income taxpayers. The tax bill initially approved by
the House of Representatives in 1978 did include explicit indexing of gains, starting in 1980, but this
provision was struck before the enactment of the Revenue Act of 1978.
37. David has shown that for

most people capital gains are
actually realized on a fairly
regular basis. It is no doubt
true, however, that very large
realized gains relative to other
income are not uncommon, as
in the sale of a house. Martin
David, Alternative Approaches
to Capital Gains Taxation
(Brookings Institution, 1968),
p. 107.
38. If the rules are intended to deal with variable income flows, they are not well directed, for Treasury
studies show that their main effect is to reduce the taxes of those with easily anticipated life-cycle increases
in income. See, for example, Eugene Steuerle, Richard McHugh, and Emil Sunley, "Income Averaging:
Evidence on Benefits and Utilization," in Compilation of OTA Papers, vol. 1 (GPO, 1978).
39. This assertion strictly holds only where tax rates do not vary over the taxpayer's lifetime. This is not a
negligible qualification when graduated rate schedules are used, and it is a point in favor of the self-averaging
aspect of the consumption tax system discussed in the text.
40. For an elementary discussion and introduction to the literature, see David F. Bradford and Harvey S.
Rosen, "The Optimal Taxation of Commodities and Income," American Economic Review, vol. 66 (May
1976, Papers and Proceedings, 1975), pp. 94-101. For a discussion specifically directed to the issue of the
tax treatment of savings, see David F. Bradford, "The Economics of Tax Policy toward Savings," in George
M. von Furstenberg, ed., The Government and Capital Formation (Ballinger, 1980). This paper also goes into
the question of how the choice of tax rules affects the accumulation of wealth.
41. See Paul A. Samuelson, "An Exact Consumption-Loan Model of Interest with or without the Social
Contrivance of Money," Journal of Political Economy, vol. 66 (December 1958), pp. 467-82; Peter A.
Diamond, "National Debt in a Neoclassical Growth Model," American Economic Review, vol. 55 (December
1965), pp. 1126-50; Martin Feldstein, ''The Welfare Cost of Capital Income Taxation," Journal of Political
Economy, vol. 86 (April 1978), pt. 2, pp. S29-S51.
42. I follow here the treatment in A. B. Atkinson and J. E. Stiglitz, Lectures on Public Economics
(McGraw-Hill, 1979).
43. Cobb-Douglas preferences for this case can be written
where In is the symbol for natural logarithms and H is the total time available for working in the first
period. We can think of the household as having (1 - tw)wH to spend on consumption in the two periods
and on leisure. The household with these preferences will spend a1(1 - tw)wH/(a1 + a2 + a3) on first-period
consumption; a2(1 - tw)wH/(a1 + a2 +a3) on claims on second-period consumption, at price (1 + t2)/(1 + r);
39
40
and a3(1 - tw)wH/(a1 + a2 +a3) on leisure, at price (1 - tw)w.

44. See, for example, Alan S. Blinder, Toward an Economic Theory of Income Distribution (MIT Press,
1974).
Â
Page 36
45. Feldstein explains why an insignificant responsiveness of saving to the rate of return does not imply a low
efficiency cost of capital income taxation. He points out that the necessarily nonnegative compensated
elasticity of demand for second-period consumption with respect to the rate of return (the basic parameter
related to efficiency) is consistent with a zero or even negative elasticity of the demand for saving, which is
the product of a quantity (second-period consumption) and a price (the present value of a second-period
dollar). Feldstein, "Welfare Cost."
46. Ibid.
47. Green and Sheshinski point out that in comparing both distorted equilibria to a nondistorted equilibrium,
Feldstein's approximation to the gain is unnecessarily crude. Their more refined calculations (with slightly
different assumptions) place the differential loss at about two-thirds of Feldstein's estimates. Jerry R. Green
and Eytan Sheshinski, "Approximating the Efficiency Gain of Tax Reforms," Discussion Paper 516 (Harvard
Institute of Economic Research, 1978).
48. See, for example, Peter A. Diamond and James A. Mirrlees, "Optimal Taxation and Public Production II:
Tax Rules," American Economic Review, vol. 61 (June 1971), pp. 261-78.
49. A. B. Atkinson and A. Sandmo, "The Taxation of Savings and Economic Efficiency," Discussion Paper
(Bergen, Norway: Norwegian School of Economics and Business Administration, 1977).
50. See, for example, the voluminous literature on the corporate income tax, running from Harberger in 1962
to Shoven in 1976. Arnold C. Harberger, "The Incidence of the Corporation Income Tax," Journal of
Political Economy, vol. 70 (June 1962), pp. 215-40; John B. Shoven, "The Incidence and Efficiency Effects
of Taxes on Income from Capital," ibid., vol. 84 (December 1976), pp. 1261-83.
51. Richard Goode, The Individual Income Tax, rev. ed. (Brookings Institution, 1976), pp. 21-25, and
Goode's paper in this volume; Richard A. Musgrave and Peggy B. Musgrave, Public Finance in Theory and
Practice (McGraw-Hill, 1973), pp. 205-06. More recently Musgrave has offered qualified support for a
consumption base that includes gifts and bequests, called an "ability to pay" consumption base in Blueprints,
p. 36. Richard A. Musgrave, "ET, OT and SBT," Journal of Public Economics, vol. 6 (July-August 1976), pp.
3-16.
52. In all of what follows I ignore the possible uses for accretion, such as medical expenses, other than
consumption or savings.
53. By taxing all changes, it is presumably possible over a long period to achieve a tax burden roughly related
to the level. It is argued below that precisely this effect can be achieved with greater consistency by a
consumption base. See also David F. Bradford and Eric Toder, "Consumption vs. Income Base Taxes: The
Argument on Grounds of Equity and Simplicity," in National Tax Association-Tax Institute of America,
Proceedings of the Sixty-ninth Annual Conference on Taxation, 1976 (Columbus, Ohio: NTA-TIA, 1977), pp.
25-31.
54. In addition to Atkinson and Sandmo, "Taxation of Savings," see Alan J. Auerbach, "The Optimal
Taxation of Heterogeneous Capital" (Harvard University, 1978).
40
41
55. Samuelson, "Exact Consumption-Loan Model"; Diamond, "National Debt"; and Diamond, "Taxation and
Public Production in a Growth Setting," in James A. Mirrlees and N. H. Stern, eds., Models of Economic
Growth (London: Macmillan, 1973), pp. 215-
Â
Page 37
35. See Atkinson and Sandmo, "Taxation of Savings," for a compact survey of the literature.
56. Atkinson and Sandmo, "Taxation of Savings."
57. Edmund S. Phelps, Golden Rules of Economic Growth (Norton, 1966).
58. Atkinson and Sandmo, "Taxation of Savings."
59. This paraphrases the portion of the Atkinson-Sandmo paper in which the capital-labor ratio is allowed to
vary. The sections taking the capital-labor ratio as fixed seem to me to accept an inexplicable and
unnecessary constraint.
60. See note 43.
61. The effect of a person buying a unit of debt in the first life period and selling it for (1 + r) in the second
(where r is the rate of return), is equivalent to a lump sum tax of one unit in the first period and a lump sum
subsidy of (1 + r) in the second In a growth setting, part of the second-period lump subsidy to the old can be
financed by the lump sum tax on the young, enough to return (1 + n), where n is the rate of population
growth. The rest must be raised through transaction taxes (or given away through transaction subsidies). If Dt,
Gt, and Tt are per capita government debt, government expenditure (other than interest payments), and tax
revenues, respectively, (1 + n)Dt+1 = (1 + rt)Dt + Gt - Tt. Letting unsubscripted variables denote steady-state
values, if r ℜ≠ n, D = (G - T)/(n -r). If r = n, the golden rule condition, there is no steady state unless G = T.
In this case, the steady-state debt per capita is indeterminate, since the interest payments are exactly covered
by the net new debt issue. Thus it is assured that whatever debt level is needed to bring the rate of return on
capital into line with population growth can be sustained, but the budget must be balanced for government
outlays other than interest payments.
62. See note 61.
63. A. B Atkinson and J. E. Stiglitz, "The Design of Tax Structure: Direct Versus Indirect Taxation," Journal
of Public Economics, vol. 6 (July-August 1976), pp. 55-75.
64. Such factors as the costs of administration might also lead to the use of transaction taxes, as in the case in
which a tax burden is placed on residents of other countries by a tariff, lump sum levies on these people being
impossible.
65. Note that, as in the preceding growth analysis, consumption in future periods can be included among the
goods distinguished by the model.
66. For example, this might consist of a uniform per capita credit plus a flat percentage of labor earnings.
67. W. J. Corlett and D. C. Hague, "Complementarity and the Excess Burden of Taxation," Review of
Economic Studies, vol. 21, no. 1 (1953-54), pp. 21-30.
68. This is not to say that those arguments could not be made more formally. A number of writers have tried
to integrate these ideas into conventional welfare economics, including Atkinson and Stiglitz, Lectures on
Public Economics, Musgrave, "ET, OT and SBT"; and Harvey S. Rosen, "An Approach to the Study of
Income, Utility, and Horizontal Equity," Quarterly Journal of Economics, vol. 92 (May 1978), pp. 307-22.
41
42
See also Martin Feldstein, "On the Theory of Tax Reform," Journal of Public Economics, vol. 6 (July-August
1976), pp. 77-104, who points out that when tastes are the same virtually any tax will qualify as horizontally
equitable.
Â
Page 38
Bibliography
Aaron, Henry J., ed. Inflation and the Income Tax. Washington: Brookings Institution, 1976.
Andrews, William D. "A Consumption-Type or Cash Flow Personal Income Tax," Harvard Law Review, vol.
87 (April 1974).
Atkinson, A. B, and A. Sandmo.

"The Taxation of Savings and
Economic Efficiency." Discussion
Paper. Bergen, Norway:
Norwegian School of Economics
and Business Administration, 1977
Atkinson, A. B., and J. E. Stiglitz. "The Design of Tax Structure: Direct Versus Indirect Taxation." Journal of
Public Economics, vol. 6 (July-August 1976).
. Lectures on Public Economics New York: McGraw-Hill, 1979.
Auerbach, Alan J. "The Optimal Taxation of Heterogeneous Capital." Cambridge: Harvard University, 1978.
Blinder, Alan S. Toward an Economic Theory of Income Distribution. Cambridge: MIT Press, 1974.
Boskin, Michael J. "Taxation, Saving, and the Rate of Interest," Journal of Political Economy, vol. 86 (April
1978), pt. 2.
Bossons, John. "The Value of a Comprehensive Tax Base as a Tax Reform Goal," Journal of Law and
Economics, vol. 13 (October 1970).
Bradford, David F. "The Economics of Tax Policy toward Savings," in George M. von Furstenberg, ed., The
Government and Capital Formation. Cambridge, Mass.: Ballinger, 1980.
. "The Meade Report from a U.S.

Perspective." Presentation at the
Social Science Research
Council/Public Sector Study
Group Tax Reform Symposium,
University College, London, May
26, 1978.
, and Harvey S. Rosen. "The Optimal Taxation of Commodities and Income," American Economic Review,
vol. 66 (May 1976, Papers and Proceedings, 1975).
, and Eric Toder. "Consumption vs. Income Base Taxes: The Argument on Grounds of Equity and
Simplicity," in National Tax Association-Tax Institute of America, Proceedings of the Sixty-ninth Annual
Conference on Taxation, 1976. Columbus, Ohio: NTA-TIA, 1977.
42
43
Break, George F., and Joseph A. Pechman. Federal Tax Reform: The Impossible Dream? Washington:
Brookings Institution, 1975.
Corlett, W. J., and D. C. Hague. "Complementarity and the Excess Burden of Taxation," Review of Economic
Studies, vol. 21, no. 1 (1953-54).
David, Martin. Alternative Approaches to Capital Gains Taxation. Washington: Brookings Institution, 1968.
Diamond, Peter A. "National Debt in a Neoclassical Growth Model," American Economic Review, vol. 55
(December 1965).
. "Taxation and Public Production in a Growth Setting," in James A. Mirrlees and N. H. Stern, eds., Models of
Economic Growth. London: Macmillan, 1973.
Â
Page 39
, and James A. Mirrlees. "Optimal Taxation and Public Production II: Tax Rules," American Economic
Review, vol. 61 (June 1971).
Feldstein, Martin. "Inflation, Income Taxes, and the Rate of Interest' A Theoretical Analysis," American
Economic Review, vol. 66 (December 1976)
. "On the Theory of Tax

Reform," Journal of Public
Economics, vol. 6
(July-August 1976).
. "The Welfare Cost of Capital Income Taxation," Journal of Political Economy, vol. 86 (April 1978), pt. 2.
, Jerry Green, and Eytan Sheshinski. "Inflation and Taxes in a Growing Economy with Debt and Equity
Finance," Journal of Political Economy, vol. 86 (April 1978), pt. 2.
, and Joel Slemrod. "Inflation and the Excess Taxation of Capital Gains on Corporate Stock," National Tax
Journal, vol. 31 (June 1978).
, and Lawrence Summers. "Inflation, Tax Rules, and the Long-Term Interest Rate." Brookings Papers on
Economic Activity, 1:1978.
Goode, Richard. "The Economic Definition of Income," in Joseph A. Pechman, ed., Comprehensive Income
Taxation. Washington: Brookings Institution, 1977.
. The Individual Income Tax. Rev. ed. Washington: Brookings Institution, 1976.
Green, Jerry R., and Eytan Sheshinski. "Approximating the Efficiency Gain of Tax Reforms." Discussion
Paper 516. Cambridge: Harvard Institute of Economic Research, 1978.
Harberger, Arnold C. "The Incidence of the Corporation Income Tax," Journal of Political Economy, vol. 70
(June 1962).
Institute for Fiscal Studies. The Structure and Reform of Direct Taxation Report of a Committee chaired by
Professor J. E. Meade. London: Allen and Unwin, 1978.
Kaldor, Nicholas. An Expenditure Tax. London: Allen and Unwin, 1955; Westport, Conn: Greenwood Press,
1977.
43
44
Kay, J. A., and M. A. King. The British Tax System. Oxford Oxford University Press, 1978.
Lodin, Sven-Olof. Progressive Expenditure Taxan Alternative? A Report of the 1972 Government
Commission on Taxation. Stockholm: LiberFÃ¶rlag, 1978 (originally published as Progressiv utgiftsskattett
alternativ? Stockholm: Statens Offentliga Utredningar 1976:62).
McLure, Charles E., Jr. Must Corporate Income Be Taxed Twice? Washington: Brookings Institution, 1979.
Meade, J. E. See Institute for Fiscal Studies.
Mieszkowski, Peter. "The Choice of Tax Base: Consumption versus Income Taxation," in Michael J. Boskin,
ed., Federal Tax Reform: Myths and Realities San Francisco: Institute for Contemporary Studies, 1978.
. "On the Advisability and Feasibility of an Expenditure Tax System," in Henry J. Aaron and Michael J.
Boskin, eds., The Economics of Taxation. Washington: Brookings Institution, 1980.
Â
Page 40
Musgrave, Richard A. "ET, OT

and SBT," Journal of Public
Economics, vol. 6 (July-August
1976).
, and Peggy B. Musgrave, Public Finance in Theory and Practice. New York: McGraw-Hill, 1973
Pechman, Joseph A. Federal Tax Policy Third ed. Washington: Brookings Institution, 1977.
, ed. Comprehensive Income Taxation Washington: Brookings Institution, 1977.
Phelps, Edmund S Golden Rules of Economic Growth Studies of Efficient and Optimal Investment. New
York: Norton, 1966.
Report of the Royal Commission on Taxation. 6 vols. Ottawa: Queen's Printer, 1966 (vols. 1, 2, 3, 5, 6), and
1967 (vol. 4).
Rosen, Harvey S. "An Approach to the Study of Income, Utility, and Horizontal Equity," Quarterly Journal
of Economics, vol. 92 (May 1978).
Samuelson, Paul A. "An Exact Consumption-Loan Model of Interest with or without the Social Contrivance
of Money," Journal of Political Economy, vol. 66 (December 1958).
Shoven, John B. "The Incidence and Efficiency Effects of Taxes on Income from Capital," Journal of
Political Economy, vol. 84 (December 1976).
, and Jeremy I. Bulow. "Inflation Accounting and Nonfinancial Corporate Profits: Financial Assets and
Liabilities," Brookings Papers on Economic Activity, 1.1976.
, and Jeremy I. Bulow. "Inflation

Accounting and Nonfinancial
Corporate Profits: Physical
Assets," Brookings Papers on
Economic Activity, 3:1975.
44
45
Steuerle, Eugene, Richard McHugh, and Emil Sunley. "Income Averaging: Evidence on Benefits and
Utilization," in Compilation of OTA Papers, vol. 1. Washington: Government Printing Office, 1978.
Studies of the Royal Commission on Taxation. 30 vols. Ottawa: Queen's Printer, 1966-68.
U.S. Department of the Treasury. Blueprints for Basic Tax Reform. Washington: Government Printing Office,
1977.
Vickrey, William. Agenda for Progressive Taxation. New York: Ronald Press, 1947; Clifton, N.J.: Kelley,
1972.
Willis, Arthur B., ed. Studies in Substantive Tax Reform. Chicago American Bar Foundation, 1969.
Wykoff, Frank C., and Charles R. Hulten "Economic Depreciation and the Taxation of Structures in U.S.
Manufacturing Industries: An Empirical Analysis," OTA Paper 28. Washington: U.S. Department of the
Treasury, Office of Tax Analysis, 1977.
Â
Page 41
2
The Choice Between Income and Consumption Taxes
Introduction
This chapter is intended to form the basis for discussion among tax practitioners of the potential advantages
of adopting a consumption basis, rather than an income basis, for the principal individual tax system. It
begins by describing the problem of individual tax base design in general terms. It then defines income and
consumption bases and discusses ways in which consumption appears a superior guide for tax policy. The
final section summarizes the actual data required to implement a practical consumption base. An appendix
elaborates on the definitions of income and consumption in a tax accounting framework.
The Problem
of Tax
Design
Under what may be called an individual tax system the tax authorities periodically (say, once per year) send a
bill to each individual. The bill requests payment of an amount of money determined by applying a tax
schedule to a number called the individual tax base. The schedule applied may depend upon characteristics of
the individual subject to tax (for example, age, family status, health). Conceivably, the procedures for
determining an individual's tax base may also depend upon these or other characteristics. However, the larger
the tax base of an individual with given characteristics, the larger his tax liability.
Although both the tax schedule applicable and the rules for calculating an individual's tax base may depend
upon characteristics such as family status, I shall for the most part ignore such distinctions in the following.
This means neglecting such interesting questions as
Â
45
46
Page 42
how to aggregate the tax burdens of family members. It will, however, enable us to focus on the issues of
primary concern if we look at the choice of individual tax base as though there were but one type of
individual taxpayer.
The impact of a fiscal system on individuals is only partly determined by the individual tax system. Other tax
instruments and expenditure programs may be equally or more important. The Value Added Tax and the
corporation income tax are two significant taxes affecting relative individual tax burdens, although these are
not individual taxes. In assessing the merits of one or another individual tax base it may be pertinent to
consider interactions with such other tax institutions.
We are accustomed to think of an income tax as the individual tax. However, there are others fitting my
definition to some degree. An estate or inheritance tax is an example. A payroll tax is another. Although the
individual typically does not get the bill, the payroll tax in the United States does recognize individual
circumstances. (Earnings are taxed only up to a specified maximum level in a given year.) More importantly,
the whole structure of social security taxes and benefits can be viewed as an individual tax system. An
interesting question, which I shall not pursue here, is how social security design is affected by the
characteristics of the primary individual tax.
Three Criteria
What are the criteria by which we should judge individual tax systems? Probably first and foremost is the
objective of fairness across taxpayers. Applied over and over through time, the rule defining the tax base
(together with a specification of tax schedules through time) will produce a series of tax assessments for each
individual. We would like it to be the case that people who deserve to bear less of a burden of tax experience
"better" (mainly lower) sequences of tax liabilities. Note that this description stresses not simply the effect of
applying the rule in a single period; it also takes into account its operation on individuals over time.
A second criterion is economic efficiency. Presumably, an ideal rule would assign tax burdens according to
what we might call exogenous features of an individual's circumstancesinnate intelligence, for example.
Practical rules must rely on endogenous featuresoutcomes resulting from individual choices under the
Â
Page 43
exogenously given features. For example, a practical tax would typically be positively related to labor
earnings, a quantity depending on innate earning power (arguably exogenous) and intensity of effort
(presumably endogenous). Such rules inevitably distort the individual's choices, and some potential value is
lost as a result. Other things being equal, these losses should be minimized.
A third important consideration is administrability. One could imagine a concept of relative

deservingness-to-pay tax based on elaborate psychological tests or close scrutiny of an individual's diet or
some such thing, wholly impractical to implement. What we call "income" for tax purposes is basically an
aggregate of market transactions. These transactions may be more or less easy for the tax authorities to
monitor. They may impose more or less difficult recordkeeping and calculation burdens on taxpayers. They
may require more or fewer arbitrary decisions or judgment calls.
Income and Consumption Tax Bases Described
This general description of the problem of tax design is the basis for a considerable and growing literature in
economics. It is not surprising that, depending upon the details of the formulation, formal models may imply
46
47
a wide variety of different tax systems. Instead of attempting to derive a tax base from first principles, it will
be more useful for us to review the properties of the two major contending ideals for practical systems: the
consumption and income bases. The following description borrows heavily from my paper, "The Case for a
Personal Consumption Tax" (1980a) [chap. 1, this vol.]. I have also provided as an appendix a somewhat
more elaborate discussion, emphasizing the accounting aspects.
Income is usually defined by tax theorists in terms of the uses to which the individual puts his resources
during the year. This definition looks like the familiar accounting identity
where ℜϒ, C and ∆W stand for income, consumption, and change in net worth over the accounting period
("savings"), respectively. To employ income as a tax base requires putting operational flesh on the concepts
of consumption and savings. For example, it must be determined whether outlays for medical treatment
constitute consumption. Similarly, it must be decided what sort of wealth will be
Â
Page 44
included in net worth; for instance, human capital is usually not included. It should be emphasized that the
terms "consumption" and "wealth" are not operationally defined a priori; they are defined in the process of
determining tax policy.
Practical calculation of either a consumption or an income tax base normally starts with the individual's
receipts rather than his outlays. This can readily be seen by starting with the assumption that the wealth
entering the definition of the tax base is an asset like a savings account. We think of a savings account as
having a readily identifiable current yield. The yield is simply the interest rate, r, times the account balance.
Let Wt stand for the wealth at the beginning of period t, and Et for the "nonwealth receipts" (wages, transfers,
and so forth), understood as occurring at the end of period t. The yield on wealth, rWt, and consumption, Ct,
are also regarded as occurring at that time. For these concepts to form a satisfactory accounting system, it is
necessary that
or
The left side of (2.3) is income

as we have defined it. The
right side, the sum of
nonwealth receipts and returns
on wealth, is the usual
calculation base. It is what the
layman thinks of as income.
The same approach is normally

taken to calculate a consumption
tax base. Savings are subtracted
from the sum of nonwealth
receipts and returns on wealth:
47
48
The difference between an income base and a consumption base lies entirely in the treatment of savings. A
common view is that a consumption tax would treat various nonsavings transactions differently than would
an income tax. For example, it is sometimes suggested that gifts and bequests received would necessarily be
treated differently in income and consumption accounting. This is not so. Assuming their anticipated value is
not counted in wealth, these receipts are simply included in Et in the appropriate period. Referring to
accounting relationships (2.3) and (2.4), one sees that such an increment in nonwealth receipts leads to the
same change in both income and consumption.
Â
Page 45
For another example, consider the treatment of gifts and bequests given. It may be argued that amounts given
away are not consumed. If this approach were taken, the corresponding amount would need to be deducted
from nonwealth receipts in the period in question. But the procedure follows whether it is income or
consumption we are trying to measure. Whether or not one regards gifts and bequests as consumption has
nothing inherently to do with the choice between income and consumption bases. In thinking about this
choice, all that should concern us is the treatment of savings.
Both income and consumption are reasonable choices as tax bases. Both seem likely to lead to relative tax
burdens among individuals that are broadly consistent with the degree of well-off-ness as it might be
independently specified. Consumption is typically a large fraction of income, and the ordering of taxpayers
by income and consumption is likely to be highly correlated. I have argued at various times that consumption
is the preferable base, however, and in the following sections I shall try to recapitulate those arguments
briefly.
Equity and the Choice Between Income and Consumption Bases
Let us suppose we could as easily implement either of these two taxes. Why favor the consumption base as
far as fairness or equity is concerned?
A part of the answer has to do with formal optimal tax theory. Consider, for example, a world in which all
individuals have the same preferences about consumption of goods at different times but have different
abilities. The ability differences are reflected in wage rates. Suppose that taxes can be levied on income or
consumption or both. Now pick tax schedules to raise a required amount of revenue and to maximize some
weighted sum of the utility levels obtained by individuals.
The solution to such a problem depends upon various parameters of the preferences and the distribution of
abilities. It is not unreasonable to describe the parameter values implying the consumption base as a kind of
midpoint of the plausible set. Other values imply more taxation of savings (in the direction of an income tax
or beyond) or less (that is, subsidy to savings). (Readers interested in a more extensive discussion may find
my 1980b paper helpful.)
Â
48
49
Page 46
To me, the more compelling part of the answer follows from reflection on the principle underlined earlier that
the difference between an income base and a consumption base lies entirely in the treatment of savings.
Imagine a situation without either tax and with two individuals we regard as similarly endowed. Then think
about raising revenue either by imposing an income tax or a consumption tax. In general, the results will not
be the same for our two individuals. One will be seen to be burdened more heavily by the income tax, the
other by the consumption tax. The one burdened the more heavily by the income tax will be the one who
saves more. (This may be because of a preference for postponing consumption or because the individual's
nonwealth receipts occur late in life.) One of the two tax bases must relatively burden the greater saver.
Forced to choose, I tilt toward the saver.
One reason for this tilt is an externality. Accumulation by individuals is socially desirable: I am better off
when my neighbor has a reserve against future needs. But even without this externality, it seems to me there
is something appealing about the neutrality toward savings characteristic of a consumption base.
Suppose, for example, we are comparing the results in a system in which the lifetime pattern of nonwealth
receipts is fixed for each individual. People may differ in initial wealth, in the time path of nonwealth
receipts, and in the use made of these endowments. How should they be compared? Since people having the
same aggregate of initial wealth and present value of nonwealth receipts have the same consumption
possibilities, it seems compelling that they should bear the same tax burden, measured in present-value terms.
Put another way, because of the lifetime budget constraint, the principle of horizontal equity calls for equal
reductions in the discounted value of consumption by people for whom this total would be equal in the
absence of tax. The discounted value of nonwealth receipts added to initial wealth might be labeled ''lifetime
wealth," the operative constraint on lifetime consumption. The principle of horizontal equity amounts to the
idea that people should be taxed according to lifetime wealth. (We can interpret the tax burden as the present
value of lifetime tax payments.) Vertical equity, in turn, calls for positively relating tax burdens to lifetime
wealth (and, one might add, the principle of progressivity requires that the tax burden as a fraction of lifetime
wealth be larger as lifetime wealth is larger).
Â
Page 47
Implementing Equity Principles by a Consumption Tax
These principles of horizontal

and vertical equity are readily
implemented by a consumption
tax along lines to be described
below. Under this scheme each
person will confront a constant
expected rate of tax on his
annual tax base, with the level of
the rate a positive function of his
lifetime wealth. Further, by the
use of a progressive schedule of
taxes on the annual base, the
burden, measured by the present
value of taxes discounted at the
rate of interest (the pre- and
posttax interest rates are the
49
50
same) will be progressively

related to lifetime wealth.
By contrast, an annual income tax generates tax burdens that are somewhat haphazardly related to individual
circumstances, as measured by lifetime wealth. A flat rate income tax, unlike a flat rate consumption tax,
systematically biases the distribution of tax burdens. Given two persons with the same lifetime pattern of
nonwealth receipts, the income tax imposes the lighter burden on the one who consumes early in life, the
heavier burden on the one who postpones consumption. Given two persons with the same lifetime wealth, the
income tax imposes the lighter burden on the one whose nonwealth receipts occur late in life. If the income
tax is levied according to a progressive schedule, the relationship between lifetime wealth and lifetime tax
burden will be even less systematic. Not only does the timing of receipts and consumption enter the result in
the way described, but also other characteristics of their time profile now influence relative tax burdens.
All this discussion is based on a simplified example. Other aspects of individual circumstances might be
considered relevant. For example, it might be desirable to make a distinction between non-wealth receipts
due to labor and those due to inheritance or gift. Attention to family status is presumably in order. Imperfect
capital markets and uncertainty complicate the picture. Yet none of these complications seem to point to a
preference for income over consumption taxation.
Efficiency and the Choice Between Income and Consumption Bases
One efficiency issue is commonly regarded as critical to the income versus consumption tax choice. Does
society save too little because
Â
Page 48
of the income tax? A great deal of effort has been devoted to attempting to measure the responsiveness of
aggregate private savings to the rate of return on saving. The conventional wisdom still seems to be that there
is not very much responsiveness. A commonly drawn implication is that we might as well tax income.
This is not a proper implication at all. At best the conventional wisdom concerning aggregate saving would
hold that consumption and income taxes are equivalent as far as the level of saving is concerned. But I would
go further and suggest that once government accumulation (or decumulation) is allowed, the link is broken
between any aggregate saving objective and the choice between income and consumption bases. (This
assumes government accumulation does not generate an equal and opposite reaction in private saving.)
A concern with aggregate saving is not readily justified by the usual analysis of efficient resource allocation.
Economists normally focus on losses due to the distortion of individual choices arising from the tax-caused
divergence between social and private tradeoffs. An example is the spread between the yield on savings and
the return to the saver under an income tax. Even in this context the responsiveness of saving to the rate of
return is not the key parameter. It is not difficult to construct examples in which a consumption tax leads to a
minimum of distortion even where savings are totally insensitive to the yield.
More important than any of this is the inference I draw from the following three propositions:
Unlike the ideal model, the real world treats very differently savings transactions that are economically very
similar. Real world income taxes generate large intersectoral and portfolio distortions.
Problems of implementation assure that any real world income tax will induce significant distortions of this
type.
A consumption type tax that is substantially free of these distortions would be easy to implement.
50
51
My guess (difficult to substantiate) is that efficiency losses associated with divergences from consistency in a
practical income tax system are unavoidable and large. Such losses may exceed those that might be at stake
in the choice between ideal income and consumption taxes. Because this view is tied up with problems of
implementation, let us turn to that subject now.
Â
Page 49
The Choice Between Income and Consumption Bases: Implementation Issues
For a time, conventional wisdom among many economists was that a consumption base might well be
preferable to an income base in principle but would be too difficult to put into practice. The currency of this
view is somewhat surprising since one of the early and influential advocates of a consumption tax, Nicholas
Kaldor (1955), stressed precisely the opposite. In my view, a consumption type tax would be easy to
implement and an equally consistent income type tax very difficult to implement. (Other recent studies
concur. See for example, Lodin [1978], the Meade Report [Institute for Fiscal Studies, 1978], Kay and King
[1978].)
The reason for this is that a great many of the most severe problems of measurement in the income tax fall
away in a consumption tax, while the latter adds virtually no new ones. The difficulties in the income tax
arise when we attempt to use data on transactions to attach numbers to the symbols in expression (2.3). Here
are three examples:
It is clear that an employee who receives wages has experienced a nonwealth receipt. But so also has an
employee who obtains a promise of future pension benefits from his employer. The first is easy to measure,
and the second difficult.
If an individual holds a savings account untouched during the year, it is clear and easily measured how much
has been his return from this part of his wealth (although correcting for inflation is not simple). But no
comparably simple data arise to allow us to measure the return in direct consumption services he enjoys from
ownership of a house, an automobile, a work of art, jewelry, and so forth.
Similarly, it is difficult to measure the return in direct accrual of wealth he experiences as a consequence of
holding ownership in a farm, a small business, even shares of stock in a corporation.
These three example illustrate three categories of measurement difficulties: (1) the problem of measuring the
value of receipt today of claims that will generate cash flows only in the possibly distant, possibly uncertain,
future; (2) the problem of measuring the value of consumption services rendered by assets, corresponding to
which no actual cash transaction ever occurs; and (3) the problem of measuring
Â
Page 50
accruing value of assets corresponding to which the cash transaction will occur in the possibly distant,
possibly uncertain, future.
These are not trivial economic phenomena. Mention of additions to retirement rights under category 1 and
owner occupied housing under 2 should suffice to demonstrate their significance. Category 3 accounts for the
51
52
need in the tax law for rules relating to depreciation of business structures and equipment and for rules
relating to capital gains. In the United States, these are very complex and contentious aspects of the law.
Category 3 also encompasses the taxation of accruing retirement value even in fully funded retirement plans
(untaxed in the United States) and the accruing value of life insurance policy claims (taxed in a very
complicated way in the United States). Finally, category 3 may be the intellectual basis for the taxation of
corporate profits, as a proxy for the otherwise untaxed accrual of wealth at the shareholder level. (I shall
return briefly to this point below.)
Measurement difficulties in categories 1 and 3 vanish if consumption is the desired tax base. Because what is
involved is both postponed cash receipt and postponed consumption, the failure to measure correctly
nonwealth receipts (Et) and returns on wealth (rWt) in (2.3) or (2.4) corresponds precisely to an equal and
opposite mismeasurement of savings (∆Wt). Under a consumption tax one could simply dispense with the
rules relating to depreciation, capital gains, special treatment of retirement savings and life insurance,
corporation income measurement and taxation, and innumerable complex procedures for dealing with
business transactions. Acquisition of an asset would result in a deduction on the right side of (2.4). Accruing
asset value changes or receipt of claims on future payments can simply be ignored until they actually
generate cash flow. The result is an extremely simple system of tax accounts for these problem categories.
Measurement difficulties in category 2, direct consumption services generated by wealth, are also easily dealt
with under a consumption base system by taking advantage of a technique based on the equivalence in
present value of the following: recognizing a deduction from the tax base for acquisition of an asset but
including in the tax base all return flows of value (the standard approach to implementing a consumption
base); and ignoring for tax purposes both the acquisition of an asset and the return flow of value. In-
Â
Page 51
tuitively, the second approach has the effect of paying in advance (by foregoing the deduction) the taxes that
would become due subsequently under the first approach. For this reason, it may be called the tax
prepayment method of accounting. For assets with ordinary cash yields, the choice between the two methods
is a matter of indifference (provided the applicable tax rate is the same). The tax prepayment method is
tailor-made, however, for investments generating hard-to-measure noncash yields.
Inflation and the Choice Between Income and Consumption Bases
One of the most serious current problems of income measurement is how to deal with inflation. My guess is
that inflation has been the main factor leading to the abandonment in the United States of any serious attempt
to tax on the basis of a consistent income concept.
The difficulty of measuring income in a time of inflation arises because of the need to use market transactions
made, actually or implicitly, at different times. In calculating the gain from sale of an asset, for example, it is
necessary to subtract the purchase price from the current sales proceeds. If substantial inflation has occurred
in the interim, adjustments are necessary to express both transactions in common units. Failure to adjust such
items as capital gain and depreciation calculations may enormously distort the tax base.
Inflation also produces an "inflation premium" in interest rates. Unfortunately, it is not easy to determine
what that premium is, since it depends upon hard-to-observe anticipations of future price level changes. But
failure to adjust interest payments and receipts can produce astounding distortions of incentives when the
inflation rate is large. To illustrate, take the case in which the interest rate in the absence of inflation is 2
percent, and suppose that with an inflation rate of 10 percent an interest rate of 12 percent prevails. Consider
now the price today for which a dollar of real purchasing power thirty years hence can be bought (or sold).
Table 2.1 shows how this price in column (1) varies with the tax rate on interest in the absence of inflation.
52
53
The same figures apply with inflation if interest payments and receipts are adjusted for tax purposes by
subtracting out the 10 percent inflation premium. Column (2) shows the price without inflation adjustment.
Prices in excess of 1 in column (2) for tax rates of 20 percent or more reflect the fact that, in
Â
Page 52
Table 2.1
Illustration of the effect of inflation on the price (via borrowing or lending) of a dollar of purchasing
power thirty years hencea
Individual tax (1) (2)
rate No inflation or inflation plus adjustment of 10 percent annual inflation and no
(percentage) interest for tax purposes adjustment
0 $0.55 $0.55
10% 0.59 0.79
20 0.62 1.13
30 0.66 1.62
40 0.70 2.34
50 0.74 3.40
a. Entries show the present amount one would have to pay to purchase a claim on one dollar (in present
purchasing power) thirty years from now. In the illustration the interest rate with no inflation is assumed
to be 2 percent; with inflation at 10 percent the interest rate is assumed to be 12 percent.
the absence of inflation adjustment, the after-tax rate of interest is negative under the illustrative conditions.
The example illustrates rather dramatically the strain imposed by inflation on an income tax system that has
no corrections for the inflation factor in interest payments.
Neither type of inflation correction would be needed under a consumption type tax. This is because the
calculation of the tax base involves only current year transactions. The tax base and tax liabilities are always
measured in consistent units. To illustrate, consider a person who purchases a share of stock for $100 that
grows in real value to $110 by the next period, when it is sold and the proceeds are consumed. (The same
reasoning would apply to the interest rate illustration in the previous table.) If the taxpayer's tax rate is 50
percent, under the standard treatment he or she sacrifices $50 of consumption in the first period in return for
$55 extra consumption in the second period. If the tax prepayment approach is chosen, the taxpayer sacrifices
$100 of consumption in the first period in return for $110 extra consumption in the second period. The same
10 percent return is obtained in both cases. If the price level doubles between the two periods, and the price
of the stock with it, the second period outcomes will be $110 and $220, respectively, under the two
treatments, exactly the same in real terms as when there is no inflation.
Â
53
54
Page 53
Averaging
Before turning to the question of how the choice of individual tax base bears on the taxation of corporations, I
would like to mention the neglected matter of tax averaging. The advantages of the sort of uniformity of
treatment accorded a wide range of investment or savings transactions by either a consistent income or
consumption base depend upon constancy over time of the applicable rate of tax. In a graduated rate income
tax system, the marginal rate is likely to vary over the individual's lifetime. (Averaging provisions may be
provided, as in U.S. law, but they typically do little to mitigate this pattern.)
Ideally, one would like to have a single flat rate applicable, or more precisely anticipated to be applicable, to
an individual's transactions, with a higher rate of tax applicable to better endowed individuals. Interestingly,
just this result follows from the availability in a consumption type system of the two methods of treating
savingsthe standard one involving current deduction followed by taxation of return flow, and the tax
prepayment method involving neither deduction of the amount saved nor taxation of the return. It is in the
interest of the taxpayer to choose the balance of the two approaches in just such a way as to maintain a
constant marginal tax rate over time. Furthermore, a progressively graduated schedule of rates means that the
constant rate attainable by a taxpayer will be higher, the greater his endowment.
The Individual Tax Base and the Taxation of Corporations
Corporations are taxed under a variety of different methods in different countries. Among the varying
features are the definition of taxable income, the treatment of dividends at both the corporate and shareholder
level, the deductibility of interest payments, the applicability of investment incentives, and the methods of
accounting for foreign earnings. It would take me far afield to attempt to discuss these aspects in any detail in
this paper.
On the other hand, certain general propositions about the systems of corporate taxation most readily
compatible with progressive individual income or consumption taxation seem reasonable:
First, in the context of a genuine accrual income tax at the individual level, any corporate level taxation of
incomedefined in the same way as individual income but with "distributions" replacing
Â
Page 54
"consumption"will distort the allocation of investment between corporate and noncorporate forms, the
structure of corporation finance, and the legal form of business organization. It seems rather difficult to
justify such a tax if the underlying choice of individual income is accepted as the correct measure of
deservingness-to-pay-tax. Under an accrual income tax, wealth changes associated with share ownership
would be most naturally measured by changes in market value of shares. The sum of dividends and share
value change would enter the calculation of rW in (2.3). There would be no need for corporate level income
accounting for tax purposes.
Second, corporate level accounting is needed if markets for shares are insufficiently active to permit regular
revaluation. The natural procedure in that case is to allocate income calculated at the corporate level to
individual shareholders. The practical problems of implementing methods to allocate corporate income to
shareholders are formidable in the context of actual tax systems. (I think the difficulties are largely
attributable to the fact that actual systems do not attempt to tax individuals on the basis of our definition [2.1]
of income.)
54
55
Third, under a system of individual income taxation that makes no attempt to follow the accrual of wealth
due to holding assets, a tax on income imposed at the corporate level may make sense. It is a kind of proxy
for the individual level tax. A flat rate tax, however, will be a bad approximation to the allocation of income
to the shareholder if there is substantial variation in individual marginal rates. Furthermore, the
approximation will tend to be the worse the higher the rate of return on corporate capital.
Fourth, the double taxation of dividends as practiced in the United States seems to me hard to justify. (But
then, U.S. corporations' practice of paying dividends is equally hard to explain.)
Finally, with a thoroughgoing consumption base at the individual level there would be no logic to a tax on
income at the corporate level. However, an extra flat rate tax on corporate cash flow to and from public
equity holders would be wholly compatible with an individual consumption base. Such a tax might be useful
as a method of capturing corporate wealth in the context of a transition from the present rules to a
consumption base.
How can an Individual Consumption Tax be Implemented?
As has been emphasized repeatedly here, the difference between a consumption and an income base resides
in the treatment of savings.
Â
Page 55
In other respects they are, or can be, the same. As we have also already emphasized, two basic methods of
accounting, the standard and the tax prepayment approaches, produce the effect of exempting savings from
tax. Implementation of an individual consumption base, then, requires developing methods for keeping track
of savings given the standard treatment (deduction), so that the returns will be taxed, and isolating savings
undertaken on a tax prepaid basis, so that the associated returns will be exempt from tax.
The following simple rules will accomplish this:
All assets purchased for direct personal use (house, automobile, and so forth), must be treated on a tax
prepaid basis (no deduction). This treatment is currently followed in the income tax. It avoids the problem of
measuring the returns taking the form of direct services.
All assets purchased for use in closely held business enterprises must be given standard treatment
(immediately expensed). In other words, unincorporated business enterprises must be accounted for on a cash
flow basis. This procedure precludes the necessity of separating out the earnings of capital and labor.
Financial assets may be purchased or sold via qualified accounts. A qualified account is supervised by a
fiduciary institution, such as a bank. Net deposits to a qualified account during the accounting period are
reported to the tax authorities and deducted from the depositor's tax base. Net withdrawals in an accounting
period are similarly reported and are added to the depositor's tax base. It is permitted to borrow via a qualified
account, in which case the proceeds of the loan are added to the tax base; repayments are deducted. Nothing
transpiring inside the qualified account has direct tax consequences, only deposits and withdrawals.
Transactions with respect to financial assets other than via qualified account, including payment and receipt
of interest, receipt of dividends, purchases and sales, have no consequences for the tax base.
Table 2.2, which in a sense represents the summing up and conclusion of this paper, shows the data that
might be used to implement a consumption type base, called a Cash Flow Tax (CFT). For comparison, I have
also included in the table an indication of the data needed to implement a reasonably consistent individual
income base called a Comprehensive Income Tax (CIT). As expected, many of the data items are common to
55
56
the two taxes. Items preceded by a
Â
Page 56
Table 2.2
Information on tax returns for illustrative comprehensive income and cash flow taxes
(Items preceded by a bullet are treated differently under the comprehensive income tax [CIT] and the cash flow
tax [CFT].)
An element of
Household receipts and deductions C.I.T. C.F.T.
Receipts:
R-1 Wages, salaries, tips, royalties, and so forth, subject to taxa Yes Yes
R-2 Receipts of pensions, annuities, disability compensation, workman's compensation, and Yes Yes
sick payb
R-3 Gifts, inheritances, trust distributions and life insurance death benefits receivedc No Yes
R-4 Interest received on financial assets, adjusted for inflation Yes Nod
R-5 Dividends received on corporate earnings Noe Nod, f
R-6 Allocated share of inflation corrected corporate earnings Yes No
R-7 Policyholder claim on earnings from life insurance, annuity, and pension plan reserves, Yes No
adjusted for inflation
R-8 Increase in value of the claim on a trust, beyond allocated share of amounts given or Yes No
bequeathed to the trust, adjusted for inflation
R-9 Proceeds from the sale, exchange or distribution of capital assets Yes Nod, f
R-10 Imputed service value attributable to owner-occupied housing and other household Yes No
durables
R-11 Gross receipts from unincorporated business enterprises Yes Yes
R-12 Withdrawals from qualified accounts (including withdrawals of borrowed funds)g No Yes
R-13 Total receipts (sum of included items R-I-R-12) Deductions:
Deductions:
D-1 Special items as a matter of policy (e.g., charitable contributions, medical expenses) Yes Yes
D-2 Contributions to qualified retirement plans Yes Yes
D-3 Gifts and bequests made to an identified taxpayer or trust with eligible beneficiary (cf. No Yes
receipts item R-3)
D-4 Interest paid on indebtedness (including interest on home mortgages) adjusted for Yes Noh
inflation
D-5 Net life insurance premiums No Yes
D-6 Employee business expense (includes qualified travel expenses, union and professional Yes Yes
association dues, tools, materials and qualified educational expenses)
D-7 Basis of assets sold, exchanged or distributed (cf. receipts item R-9), adjusted for Yesi No
inflation
D-8 Current expenses associated with unincorporated business enterprises Yes Yes
56
57
(table continued on next page)
Â
Page 57
Table 2.2
(continued)
(Items preceded by a bullet are treated differently under the comprehensive income tax [CIT] and the cash flow
tax [CFT].)
An element of
Household receipts and deductions C.I.T. C.F.T.
D-9 Capital outlay associated with unincorporated business enterprises No Yes
D-10 Depreciation allowances for current and past capital outlays associated with Yes No
unincorporated business enterprises, adjusted for inflation
D-11 Deposits to qualified accounts (including repayment of borrowed funds) No Yes
D-12 Total deductions (sum of included items D-I-D-11)
Tax Base:
B-1 Total receipts less total deductions (R-13 minus D-12)
a. The definition of ''wages subject to tax" could incorporate differential rules according to individual
characteristics (for example, marital status might be used to mitigate the marriage tax problem). This item
would exclude social security taxes attributable to retirement benefits and contributions to retirement plans. It
would include employer-paid health and life insurance premiums and similar employee benefit outlays.
b. This item would probably include social security benefits of all types. Pensions and annuities for which there
has been no exclusion under R-1 or deduction under D-2 or D-11 would be excluded under R-2.
c. The exclusion of gifts and inheritances received under the illustrative Comprehensive Income Tax may be
rationalized by the difficulty in measuring accruing wealth from anticipated inheritance. Correspondingly, the
giver, bequeather, or life insurer receives no deduction from the Comprehensive Income Tax Base (deduction
items D-3 and D-5). The same system could be applied to the Cash Flow Tax. Alternatively, with appropriate
modification of the treatment of life insurance and trusts, the Comprehensive Income Tax could follow the
illustrative Cash Flow Tax usage, and vice versa.
d. Under the illustrative Cash Flow Tax, a significant distinction is made between assets owned on a tax prepaid
basis, the return on which is excluded from the base, and assets owned via qualified accounts, the return on
which is ultimately included on the base when withdrawn. See footnote g.
e. The exclusion of dividends under the Comprehensive Income Tax is a corollary of the allocation of all
corporate income to shareholders. Dividends result in a reduction in the basis of the shares for purposes of
calculating gain from sale or exchange (capital gain).
f. Under the Cash Flow Tax, qualified account treatment would be obligatory for closely held corporations. See
footnote g.
g. "Qualified accounts," which are similar to IRAs and H. R. 10 accounts in current U.S. income tax law, play a
critical role in the Cash Flow Tax. All inflows are deducted from the tax base, and all outflows are included.
Nothing transpiring inside the account has direct tax consequences.
h. Borrowing via qualified account (cf. fn. g) is included in the tax base and all consequent return payments into
the account, whether of principal or interest, are deducted. However, neither interest on nor repayment of
borrowing outside of a qualified account has tax consequences under the Cash Flow Tax.
i. The deduction of adjusted basis would have to be limited, as at present, in relation to sales proceeds in R-9.
Unused deduction of net losses could be carried forward.
57
58
Â
Page 58
bullet are included in one and excluded in the other. Rather than provide a point-by-point commentary, I urge
the reader to work through the illustrative tax return information. It will be readily apparent that those
bulleted items presenting significant problems (requiring inflation adjustment, for example) are required for
the income base and not for the consumption base. In some cases, for example, in the treatment of pensions
or gifts and bequests, alternative rules might be employed, but the consistent alternatives would not simplify
the CIT. The items included in the consumption base and not in the income base are typically readily
measured cash flows.
Appendix 1
Income and Consumption: Operational Definitions
An operational tax base of the income tax type is not a quantity like water in a closed hydraulic system,
wherein the total remains constant regardless of how it is directed by valves and pumps. Rather, it is an
aggregation of transactions, usually voluntary, although sometimes implicit. The transactions that take place
will depend in part upon how they are treated by the tax system. The choice of a tax base is a choice about
how to tax certain transactions.
An operational tax base is necessarily defined by a set of accounting rules that specifies the use of actual and
implicit transactions in reaching the total to which a tax schedule is applied to determine the taxpayer's
liability. The concept of income generally used in discussion of tax reform has been called an accretion
concept. It is supposed to measure the command over resources acquired by the taxpayer during the
accounting period, that command having been either exercised in the form of consumption or held as
potential for future consumption in the form of an addition to the taxpayer's wealth. Hence, the apparently
paradoxical practice of defining income by an outlay or uses conceptconsumption plus change in net worth.
Everyday usage, on the other hand, tends to associate income with the sources side of the accounts. Thus, one
speaks of income "from labor," such as wages, or income "from capital," or "from proprietorships," such as
interest and profits. Because sources and uses must be equal in a double entry accounting system, the result
should be the same whichever side is taken for purposes of measurement, provided that all uses are regarded
as appropriate for inclusion in the tax base.
Definitions of Income and Consumption
To illustrate, it may be helpful to outline a rudimentary classification of transactions to define income and
consumption. The accounts considered
Â
Page 59
first are those of a wage earner whose only sources of funds are his wages and his accumulated balance in a
savings account.
In the simplest case, the possible applications he can make of these funds may be divided into the purchase of
goods and services for his immediate use and additions to or subtractions from his accumulation of savings.
Thus, an account of his situation for the year might be the following:
58
59
Sources Uses
Wages Rent
Interest Clothing
Balance in savings account at beginning of Food

period
Recreation
Balance in savings account at end of

period
The two sides of this account are, of course, required to balance. Of the uses, the first four are generally
lumped under the concept of consumption. The last constitutes the net worth of the household. Thus, the
accounts may be schematically written as:
Sources Uses
Wages Consumption
Interest
Net worth at beginning of period Net worth at end of period
The concept of income concerns the additions to sources and their application during the accounting period.
These can be found simply by subtracting the accumulated savings (net worth) at the beginning of the period
from both sides, to give:
Addition to sources Uses of addition to sources
Wages Consumption
Interest Savings (equals increase in net worth over the period)a
a
Note savings may be negative.
Income may be defined as the sum of consumption and increase in net worth. Note carefully that this involves
a uses definition as a measure of differences in individual circumstances. This approach to the concept of
income has substantial advantages as a device for organizing thinking on particular policy issues. With this
uses definition of income, the situation of the illustrative individual may be represented by:
59
60
Wages Income
Interest
Â
Page 60
The last version of the accounts makes clear the way in which information about sources is used to determine
the individual's income. To calculate his income for the year, this individual obviously would not add up his
outlays for rent, clothing, food, recreation, and increase in savings account balance. Rather, he would simply
add together his wages and interest and take advantage of the accounting identity between this sum and
income.
This classification of uses into consumption and increase in net worth is not sufficient, however, to
accommodate distinctions commonly made by tax policy. It will be helpful, therefore, to refine the accounts
to the following:
Wages Consumption
Interest Cost of earnings
Certain other outlays
Increase in net worth
An individual's outlay for special work clothes needed for his profession requires the category "cost of
earnings." These are netted out in defining income. Note that the decision about which outlays to include in
this category is a social or political one. Thus, under U.S. income tax law, outlays for specialized work
clothes are deductible, but commuting expenses are not. There is no independent standard to which one can
appeal to determine whether such outlays are consumption (and, hence, a part of income) or work expenses
(and, hence, out of income).
Similarly, a judgment may be made that some outlays, while not costs of earning a living, are also not
properly classified as consumption. The category of "other outlays" is introduced for want of a better label for
such transactions. For example, in everyday usage, charitable contributions would not be an application of
funds appropriately labeled "personal consumption," much less "increase in net worth." Thus, using the
definition of income as the sum of consumption and the increase in net worth, we now have:
60
61
Earnings (Wages + Interest) Income (Consumption + Increase in net worth)
Cost of earnings
Again, to calculate income it is generally convenient to work from the left (sources) side of the accounting
relationship described above. In this case,
Income = Earnings
minus
Cost of earnings
minus
Certain other outlays.
Â
Page 61
Similarly, consumption may be calculated by starting with source data:
Consumption = Earnings
minus
Cost of earnings
minus

minus
Increase in net worth.
One further addition to the accounting scheme is needed at this point: the item "gifts and bequests given."
This is a use of funds that some would regard as consumption, but I reserve the term consumption, without
modifier, for the narrower notion of goods and services of direct benefit to the individual in question. The
accounts now have the following structure:
61
62
Wages Consumption
Interest Gifts and bequests given
Gifts and bequests received Cost of earnings
Increase in net worth
It must be decided whether gifts and bequests given are to be regarded as income, that is, as a component of
the total by which taxpayers are to be compared for assigning burdens. The term "ability-to-pay" is used to
describe the income concept that considers income to be the sum of consumption plus gifts and bequests
given plus increase in net worth, because it is within the taxpayer's ability to choose among these uses and,
hence, all three measure taxpaying potential equally. The label "ability-to-pay" is intended to be suggestive
only. There is no agreed upon measure of the idea of a taxpayer's ability to pay. Because of this, I employ
quotation marks when I use the term "ability-to-pay" in its role as a label for an income or consumption
concept.
"Ability-to-pay" income or consumption would also generally be calculated by starting on the sources side:
"Ability-to-pay" income = Earnings

plus
Gifts and bequests received

minus
Cost of earnings
minus
Certain other outlays.
"Ability-to-pay" consumption = Earnings

plus
Gifts and bequests received

minus
Â
62
63
Page 62
Cost of earnings
minus

minus
The difference between consumption and income is the savings or increase in net worth over the period.
Thus, equivalently:
"Ability-to-pay" consumption = "Ability-to-pay" income

minus
Finally, there is the pair of income and consumption concepts that exclude gifts and bequests given from the
category of uses by which tax burdens are to be apportioned. These are given the label "standard-of-living"
because they are confined to outlays for the taxpayer's direct benefit. As with the term "ability-to-pay," this
label is intended to be suggestive only. The "ability-to-pay" and "standard-of-living" concepts are related as
follows:
"Standard-of-living" income = "Ability-to-pay" income

minus
Gifts and bequests given,
"Standard-of-living" consumption = "Standard-of-living" income

minus
This discussion leads to a four-way classification of tax bases:
Gifts Given
Included Excluded
Increase in net Included "Ability-to-pay" "Standard-of-living" income

worth income
Excluded "Ability-to-pay" Standard-of-living"

consumption consumption
63
64
Note
1. This Appendix is adapted from U.S. Treasury Department, Blueprints for Basic Tax Reform (Washington,
D.C.: Government Printing Office, January 1977). An earlier version of the essay appeared in Tax Notes 16
(August 23, 1982): 715-723.
Bibliography
Bradford, David F. "The Case for a Personal Consumption Tax." In Joseph A. Pechman, ed, What Should Be
Taxed: Income or Expenditure? Washington, D.C.: The Brookings Institution, 1980a.
Â
Page 63
. "The Economics of Tax Policy Towards Savings." In George M. Von Furstenberg, ed., The Government and
Capital Formation. Cambridge, Mass Â· Ballinger Publishing Company, 1980b.
Institute for Fiscal Studies. The Structure and Reform of Direct Taxation: Report of a Committee Chaired by
Professor J. E. Meade. London: George Allen and Unwin, 1978.
Kaldor, Nicholas. An Expenditure Tax. London: Allen and Unwin, 1955; Westport, Conn.: Greenwood Press,
1977.
Kay, John A., and King, Mervyn A. The British Tax System. Oxford: Oxford University Press, 1978.
Lodin, Sven-Olof Progressive Expenditure TaxAn Alternative? A Report of the 1972 Government
Commission on Taxation. Stockholm: LiberForlag, 1978 (originally published as Progressiv utgiftssakattett
alternativ? Stockholm: Statens Offentliga Utredningar 1976: 62).
U.S. Treasury. Blueprints for Basic Tax Reform. Washington, D C Government Printing Office, January
1977.
Â
Page 65
3
On the Incidence of Consumption Taxes
Tax debates are usually plagued by ignorance, and the debate about consumption taxes is no exception.
Participants often mean different things by the term consumption tax. Even when they mean the same thing,
they often fail to understand the different ways that same thing may be implemented. When the discussion
turns to who will bear the tax, the critically important details of how consumption tax rules might be
introduced is typically overlooked. Thus, confusion about exactly what is under consideration is added to the
already very difficult problem of determining the incidence of even a well-defined tax structure. The purpose
of this chapter is to lay out the incidence issues in a nontechnical manner.
64
65
To illustrate the complexity of the subject, consider the case of introducing the Accelerated Cost Recovery
System (ACRS) to the income tax system in 1981. By shortening the period over which the cost of newly
constructed equipment and structures could be deducted from a company's revenues in calculating income
subject to tax, ACRS reduced the tax liabilities of businesses. Most people probably regarded this step as
regressive, that is, as relatively favoring the well-to-do. People liked the new policy who (a) thought the tax
system was too progressive to begin with or (b) thought the stimulus to economic prosperity would buy
general benefits that offset the distributional change.
However, economic analysis suggests a surprising twist to the story. 1 Increasing the depreciation allowances
on newly constructed capital makes new assets cheaper than old. Since new and old assets must command the
same prices, the introduction of ACRS presumably imposed a windfall loss on owners of existing assets, that
is, on the wealthy, particularly compared with the alternative of lower rates of tax. Conversely, the sort of
changes now in train, lengthening
Â
Page 66
depreciation lives and lowering tax rates, may be expected to generate windfall gains for owners of existing
assets, in contrast to the conventional view that these changes increase the burden on wealth owners.
I shall return to these ideas below. Before that, though, we must deal with matters of definition and of
specifying the assumptions underlying the incidence analysis.
What are Consumption Taxes?
Although the term consumption tax is widely used, there is very little uniformity of view about exactly what
is meant by it. Most people seem to have in mind a tax like the sales taxes familiar in the states' fiscal
armories. Followers of tax policy would add to the list value-added taxes (VATs) of the sort widely employed
in other countries, along with the business transfer taxes (BTTs) that have been under discussion here and in
Canada. The real aficionados recognize the possibility of a consumption tax that looks much like the existing
income tax, variously referred to as a cash flow income tax or consumed income tax. 2
I associate two properties with taxes based on consumption. Property 1 may seem obvious, namely, that a
consumption tax relates tax liabilities to a measure of a household's or individual's consumption. This
contrasts with the theoretical idea of an income tax, which relates liabilities to a measure of the algebraic sum
of a household's consumption and saving during a year (since saving may be negative, income may be either
larger or smaller than consumption). A tax on consumption might be levied at a flat rate or at graduated rates.
It might be based on an annual aggregate of a household's consumption, or it might be based on a discounted
flow of such expenditures.
Property 2 of a consumption tax is derived from the idea that consumption tax burdens should not be
influenced by the level of saving. In other words, the reward to saving obtained by the saver should be equal
to the payoff society obtains by investing the saved amount. Thus, a second defining characteristic of a
consumption tax is equality between the "before-tax" and "after-tax" rates of return on saving.
Both of these properties are often but not always present in taxes commonly thought of as consumption taxes.
Thus, a flat-rate tax on
Â
65
66
Page 67
some annual measure of consumption will also imply a zero tax on the normal return to saving. However, a
graduated-rate tax on an annual measure of consumption will generally not do so (because amounts set aside
in a time of low consumption, and hence low marginal tax rate, may pay off in a future period of high
consumption, and hence high marginal tax rate). Furthermore, some taxes (such as a tax on consumer
purchases of gasoline) that do not affect the return to saving and are often thought of as consumption taxes do
not attempt anything like a comprehensive measurement of consumption.
The X-Tax as Typical Consumption Tax
Rather than picking any one of the existing possible consumption taxes to analyze here, I would like to
emphasize the range of institutional arrangements that would constitute introduction of a broad-based
consumption tax. To draw attention to the elements that link and distinguish the various possibilities, I shall
describe here a hypothetical consumption tax, which I call the X-Tax. In addition to having properties that
might make it interesting as an actual policy option, the X-Tax provides a convenient framework for analysis.
The Basic Mechanics Explained
The X-Tax can be viewed as a variant of a value-added tax (VAT) with adjustment for vertical distribution; it
is also a close relative of the Simple Flat Tax that has been promoted by Robert Hall and Alvin Rabushka. 3
The basic X-Tax is a system with two components: a business tax (paid by all businesses, whether corporate,
proprietorship, or partnership) and a compensation tax (paid by all who receive compensation for services as
employees or the equivalent). All businesses pay taxto be specific, let us say at a rate of 7 percenton a base
consisting of the receipt from sales of all types (including sales out of inventory or sales of other existing
assets) less the outlays for purchases from other businesses and less payments to workers, whether for
current, past, or future services. All workers pay tax on the amount received from businesses (or payments of
the same character from nontaxpaying entities such as governments). Payments from more than one employer
are added together. The re-suiting total is taxed at graduated rates, with an exempt amount and marginal rates
of, say, 3, 5, and 7 percent on successively higher
Â
Page 68
levels of compensation. No other receipts of the workers (such as interest or dividends) are included in the
compensation tax base. The top rate of compensation tax is the same as the single flat rate of business tax.
The X-Tax would be administered in conjunction with the existing income tax; virtually all of the
information required is either used in the existing tax or is necessarily required by the taxpayer to derive
information used on the existing tax return. The left-hand panel of Table 3.1 describes the operation of the
X-Tax in a simple two-firm, three-worker economy. The table indicates how, under the assumed schedule of
rates, profits are taxed at a flat 7 percent and worker compensation is taxed at graduated rates ranging from 0
percent for the first $10,000 to 7 percent on amounts exceeding $50,000.
Policy Choices in Consumption Taxes
The X-Tax is a consumption tax. To see why this is so in both senses I have described above, it will help to
back up and consider an X-Tax in which the compensation tax component is not on a graduated rate basis,
but instead is assessed at the same 7 percent rate applicable to business ''income." Such a tax would be
exactly the same as one levied only on businesses but with no deduction at the business level for payments to
employees. Because all transactions among businesses are netted out (what one business includes as a receipt,
the paying business deducts as a business expense), the result would be a flat tax on sales to nonbusinesses.
66
67
We can see, therefore, that the X-Tax amounts to an annual tax on the aggregate of a household's purchases
from firms, combined with a graduated relief from that tax based on the year's earnings from employment.
The X-Tax is thus a combination of a consumption tax in the first sense with a subsidy to employment for
relatively low earners. Furthermore, because businesses immediately deduct the cost of their purchases on
capital account, the rate of return received by the investor is the same before and after tax. In effect, the
government is a full partner in the investment, sharing, via the deduction, in 7 percent of all costs and, via the
tax on receipts, in 7 percent of all returns. The X-Tax is thus also a consumption tax in the second sense
mentioned above.
It may be asked, however, whether the aggregate of sales to household is what we really mean by
consumption. As I have emphasized elsewhere, there is no scientific answer to this question. 4
Â
Page 69
Table 3.1
The X-Tax illustrated in a simple economy
Basic X-Tax (Employee pay deducted) Alternative X-Tax (Earned income credit)
Firm A Firm A
Receipts from sales less 100,000 Receipts from sales less 100,000
Purchases from firm B 20,000 Purchases from firm B 20,000
Salary to worker 1 15,000 Salary to worker 1 13,950
Business tax base 30,000 Business tax base 80,000
Tax (@ 7%) 2,100 Tax (@ 7%) 5,600
Profits after tax 27,900 Profits after tax 27,900

Firm B Firm B
Receipts from sales less 80,000 Receipts from sales less 80,000
Business tax base 5,000 Business tax base 80,000
Tax (@ 7%) 350 Tax (@ 7%) 5,600
Profits after tax 4,650 Profits after tax 4,650

Total business tax 2,450 Total business tax 11,200
Worker I Worker 1
Salary 15,000 Salary 13,950
Tax Tax credit
7% of amount over 50,000 0 7.53% of amount below 9,300 700
5% of amount over 25,000 4.30% of amount over 9,300
and less than 50,000 0 and less than 23,250 200
67
68
Total tax 150 Total credit 900
Disposable income 14,850 Disposable income 14,850
Worker 2 Worker 2
Tax Tax credit
7% of amount over 50,000 0 7.53% of amount below 9,300 700
Total tax 950 Total credit 1,500
Â
Page 70
Table 3.1
(continued)
Basic X-Tax (Employee pay deducted) Alternative X-Tax (Earned income credit)
Worker 3 Worker 3
Tax Tax credit
7% of amount over 50,000 1,750 7.53% of amount below 9,300 700
and less than 50,000 1,250 and less than 23,250 600
Total tax 3,450 Total credit 1,800
Total compensation tax 4,550 Total credit 4,200

Total of compensation and business tax 7,000 Total business tax less credit 7,000
What we mean by consumption in

this context (as in the income tax
context: income is the sum of
consumption and saving) is
68
69
necessarily a policy choice, made

to effect the discriminations
among taxpayers that we deem
desirable as a matter of equity in
sharing the aggregate tax burden.
There are many policy issues hidden in the question of how consumption ought to be defined for tax
purposes, and the simple system described above adopts implicit positions on many debatable points (for
example, the taxation of such institutions as universities). To pursue all of these points would take us too far
afield, and many of the particulars are not critical to the subject at hand. However, three issues, the treatment
of consumer durables, inheritances, and transfer payments merit mention.
In including all sales of newly constructed consumer durables, a category within which I include
owner-occuped housing, the aggregate of sales to households clearly diverges from consumption as we
usually use the term. We would not normally include in a household's annual consumption the amount paid
for a new house in a given year. Instead, we would impute to the owner-occupied house a flow of services
over time. The X-Tax applies what has come to be called the "tax prepayment" approach to these outlays. 5 In
effect, the tax paid on the acquisition of a newly constructed house or automobile constitutes payment in
advance of the expected present value of taxes that would otherwise be collected over time on the flow of
Â
Page 71
services if they were actually measured. This characteristic of the X-Tax provides an administratively simple
solution to the problem of applying the same rate of tax to housing services as to other forms of consumption.
The X-Tax ignores inheritances and bequests. It may be described as taxing amounts inherited when they are
consumed. This characteristic is a contentious property of many consumption taxes. One way to think about
the matter is to ask whether amounts given away should be regarded as consumed by the donor. If so,
bequests and other gifts would need to be added to the X-Tax base, presumably by levying a flat 7 percent tax
on them. 6
Not only does the X-Tax ignore private transfers (gifts and bequests), it ignores public transfers as well (such
as unemployment compensation and welfare benefits). There is nothing that says such transfers could not be
included in the compensation tax schedule. More probably, policymakers would be concerned not about the
undertaxation of transfers but about their overtaxation. Intuitively, imposition of a VAT might be thought of
as imposing a burden on transfer recipients. However, since both sides of the transfergiver and receiverare
affected alike, they can presumably adjust. In the case of public transfers, this means holding benefits
constant in real terms as the price level may vary under the influence of the tax. It is in this sense that the
critical element of a correction for vertical distributional effects in the tax system is with respect to labor
earnings. Private transfers will take care of themselves, and public transfers are subject to explicit public
policy choice.
Finally, a word is in order on the distinction between a consumption tax and a tax on labor earnings,
sometimes called a wage tax. If we think of the way people acquire claims to goods as divided into payments
for working (labor earnings) and payments for providing capital services (capital income), then the
observation that the X-Tax, like most consumption taxes, has the effect of eliminating the difference between
the yield on investment and the reward to the saver makes it natural to describe it as a tax on earnings. In the
formal sense just described (putting aside public and private transfers as other sources of claims to goods) the
characterization is surely correct. However, it may also be misleading in conjuring up a tax imposed only on
the ordinary wage-earner. In fact, the X-Tax would apply to such unconventional sources as new
technological inventions, discovery of new mineral deposits, increased rental value of
69
70
Â
Page 72
urban land, and many other types of increase in market value we not normally have in mind when we
describe the world as divided into labor earning a wage, w, and capital earning a return, r.
The economics literature makes a second distinction between a wage tax and a consumption tax which has
nothing to do with the question just discussed. 7 This difference turns out to hinge critically and nonobviously
on the manner in which a consumption-type tax is introduced. Intuitively, a wage tax is levied on payments to
labor, whereas a consumption tax is one levied on the purchases of consumption goods by the household,
whether the source of funds is labor earnings or yield from capital. We know, though, that the two taxes,
which sound very different, once in place have the same effect on the household's options over time. That is,
the household's budget constraint over time makes the two types of tax into the same thing. Since ownership
of capital is obtained by saving out of labor earnings, the household that must pay a flat 25 percent of its
earnings in tax will face the same opportunities as does the household that pays no tax on its earnings, but a
flat 25 percent tax on outlays for consumption (the outlays understood as including the tax itself).
In spite of this equivalence via the budget constraint, there is a useful distinction that we can associate with
the labels "consumption" tax and "wage" or "earnings" tax as they are employed in the technical economics
literature. That distinction is a matter of transitionthat is, a matter of the way the new budget constraint is
introduced. A wage tax can be said to result when the return flow from capital existing at the time of
introduction of the tax is exempted from tax (to simplify, this discussion assumes there is no existing tax); a
consumption tax can be said to result when the return flow from capital existing at the time of introduction is
included in the tax base. It is probably fair to say that most taxes of the consumption type that are discussed
in the policy literature would also be called consumption taxes in the theoretical economics literature. The
transition rules by which a consumption tax is introduced are of critical importance, because the incidence in
the course of transition to the new policy can differ greatly between two taxes that amount to the same thing
once in place.
An Alternative Way to Offset Regressivity
Equivalences and near-equivalences abound in the world of consumption taxes and distinctions without a
difference in economic terms may be very important in political terms. An illustration is the
Â
Page 73
following equivalent to the X-Tax, a system that apparently levies much higher taxes on business and
includes much more liberal treatment of workers: Instead of allowing businesses to deduct their payments to
workers under the X-Tax, oblige them to pay a flat tax of 7 percent on the entire amount of the difference
between their receipts from sales and their purchases from other businesses. Instead of the tax on
compensation, provide workers an earned income credit of 7.53 percent (for the lowest earners), with the
credit reduced to 4.30 percent on earnings in excess of the level at which the 3 percent compensation tax
bracket was reached under the original plan, to 2.15 percent on earnings in the next bracket, and with no
credit at the margin for earnings in the top bracket of the original compensation tax. (The odd percentages
result from the necessity to base credits on what amounts to before-tax earnings.) This-tax-plus-credit system
is exactly equivalent to the original X-Tax combination of business and compensation taxes, but it looks very
70
71
different. The second panel of Table 3.1 puts the earned-income credit system side by side with the basic
X-Tax, illustrating the way in which the two systems produce the same outcome for both workers and owners
of firms.
Relationship of Well-Known Alternatives to the X-Tax
Value-Added Tax
Subtraction Versus Invoice Method
In introducing the X-Tax I described it as a variant of a value-added tax. More precisely, it is a variant of
what we would normally call a value-added tax of the consumption type (because capital outlays are
immediately expensed) administered by the subtraction method, coupled with an employment subsidy to
modify the distributional effects of the flat-rate tax. The more familiar European style VATs differ from the
hypothetical X-Tax in many details, including importantly the method of administration. To emphasize the
connection with the familiar income tax and the potential for simplified administration through being
piggy-backed on the income tax, the firm's base under the X-Tax is calculated simply by adding together all
sales and subtracting all purchases from other firms. (Payments to employees are treated differently under the
two alternative approaches.) The European VATs employ the so-called invoice method, whereby the firm
claims a rebate on purchases from other firms not on the basis of amounts
Â
Page 74
paid to those firms but on the

basis of VAT identified on the
purchase invoices. 8 If a single
rate of tax is employed, the
invoice and subtraction
methods are evidently
equivalent, so the economic
analyses of the flat-rate VAT
and the X-Tax (putting to one
side the graduated earnings
offset) will also be identical.
The invoice method facilitates levying different tax rates on different commodities. In theory, one could also
employ different rates of tax in the subtraction method as well (just as one could oblige firms to include
different fractions of the receipts from sales of specific commodities and subtract different fractions of the
purchases from other firms in calculating income subject to tax). Our interest, however, is in incidence and
not administration. I shall comment below on the effect on the incidence effects of differentially taxing
various commodities. As far as terminology is concerned, once multiple rates and exclusions become part of
the system, the tax ceases to be a consumption tax in the first sense I described, namely, one based on a
concept of annual consumption (it might be described as a multiple set of such taxes), although it typically
continues to be a consumption tax in the second sense, in preserving the equality between the yield on
investment with the rate of return earned by savers.
Origin Versus Destination Basis
The description of the X-Tax implies that the firm will be taxed on the proceeds of all sales and may deduct
all purchases from other firms. Since either the sale or purchase transactions might be with foreign residents
or firms, the X-Tax would be regarded as on an origin basis. It would, however, be perfectly feasible to
specify that sales to foreigners would be excluded from tax and purchases from foreigners disallowed as
71
72
deductions. That would place the X-Tax on the destination basis common to most VATs. Despite the great
political importance attached to the distinction, there is little reason to expect there to be much economic
difference.9 Consequently, the incidence analysis of the basic X-Tax will apply to a VAT on a destination
basis.
Business Transfer Tax
Various versions of business transfer tax have been under consideration recently in the United States, and a
BTT is under active discussion in Canada.10 The term generally refers to a VAT administered by the
subtraction method, in other words, a very close relative of the X-Tax. Indeed, the X-Tax can be described as
a BTT with employment subsidy. A BTT is normally conceived of on a destination basis.
Â
Page 75
Some versions of BTT differ from the X-Tax in drawing the line around the taxpaying firms more narrowly,
stopping at the wholesale stage, for instance. The significance of such characteristics for incidence is
probably minor.
Retail Sales
A retail sales tax differs from the business tax component of the X-Tax in being administered only at the
point of sale to the public. The two taxes would be identical in economic effect if the definition of firms and
final sales subject to tax were the same. In practice, it seems that retail sales taxes often exclude professional
services and new housing construction. Also, just as a typical VAT applies lower rates for commodities
believed to be particularly important in the budgets of poor people, such commodities may be exempt from
retail sales tax.
Others
A broad-based consumption tax can be thought of as composed of a series of separate taxes on the various
individual commodities or services embodied in the relevant consumption concept. We have noted that a
VAT or retail sales tax may provide for lower or zero rates on particular commodities or services. It is
obviously a small step to any arbitrary collection of excise taxes on particular commodities.
A specific commodity tax that is frequently mentioned as a revenue source is one on energy, sometimes more
narrowly targeted at imported petroleum. To the extent these taxes are organized so as to impinge only on
final sales to consumers, they fall within the general class of consumption taxes, and the analysis is a simple
subset of the cases just described. If producer uses of energy are included, these taxes pose more complicated
issues of incidence (and efficiency) analysis, issues that go beyond the scope of this survey.
Steady-State Incidence Effects
Preliminaries: Basic Ideas of Incidence
Nominal Versus Effective
Incidence analysis concerns the real burden of taxes. For this purpose, the person or institution that sends the
check to the Treasury is of little or no relevance. Just as, in the familiar textbook analysis of a
Â
72
73
Page 76
commodity tax, it matters not whether the tax is levied on the buyer or seller side of the market, it is of no
economic importance whether taxes are nominally paid by individuals or firms. The two versions of the
X-Tax, economically identical in spite of dramatically different nominal incidence, provide a good
illustration. However, there may be very different political feelings about economically identical taxes. Those
who would choose a form of tax because it is "hidden" and apparently fools the public and those who look for
taxes that are most obvious to the public, and therefore resisted, agree that form, as well as substance
according to the usual models of incidence, matters. If they are right, the usual models of incidence may be
wrong. Apparently equivalent taxes may have different real effects. My further remarks reflect the
economist's usual skepticism on this point, but one should recognize the alternative possibilities. (For an
exploration of the issue in the context of labor supply, see Harvey Rosen's "Tax Illusion and the Labor
Supply of Married Women." 11
Differential Incidence
In the example of the incidence of introducing ACRS, discussed at the beginning of this chapter, there was an
implicit assumption that the reduced tax receipts due to shortening depreciation lives were made up in
increased income tax rates. Actually, the receipts were made up through increased issue of debt. Because the
government must operate on a budget constraint over time, it is not possible to change one tax without
changing the deficit or spending or another tax. Just which other instrument is varied will have a bearing on
the incidence of burdens.
Among the more likely uses of a consumption tax in the present situation is to reduce the outstanding
government debt. Ideally, we would like to analyze the incidence of the policy option "consumption-tax cum
reduced deficits" relative to the option "no reduced deficits." However, since I have very little to say about
the incidence of deficit financing, I shall take as the base case here the assumption that the consumption tax
revenue will be used to finance a wasteful increase in government expenditure. Obviously, I do not regard
this as an interesting policy, but it can be taken as a convenient standard against which to compare
alternatives.
Lifetime Perspective
A more fundamental matter in incidence analysis is the choice of time perspective. Most readily available
information about the dis-
Â
Page 77
tribution of tax burdens does something like attempt to allocate to existing individuals the equivalent of a
current year's installment on their tax burdens. Treasury data showing tax burdens on households classified
by current year's income are the standard fare of workaday incidence analysis.
Data of this kind are often less than satisfactory for two reasons. First, they frequently incorporate
over-simple conceptions of incidence. For example, tables purporting to show the beneficiaries of tax
expenditures typically equate reduced payments to the Treasury with reduced tax burdens (thus committing
the fallacy just mentioned). Second, even where greater attention is given to matters of economic modeling of
incidence, 12 an annual "snapshot" may give a misleading impression of both the tax burdens and the
economic positions of households. Tax burdens may be misrepresented because of the difficulty of taking
into account currently such phenomena as the income taxes payable at the time of retirement on savings set
aside in a tax-sheltered account. Economic position may be misrepresented because of the life-cycle
relationship among earnings, return to savings, transfer payments, and age. Thus, for example, those whose
73
74
income is currently low include a preponderance of the young (just embarking on their career of earnings)
and the old (in their retirement years). But the same person is once young, once middle-aged, and once old,
and presumably some measure that takes into account this fact is needed for incidence analysis. This issue is
of particular importance in the context of analysis of consumption taxes, a point to which I shall return
shortly.
Transition Incidence
It is desirable not only to take a long view about the incidence of a tax in place but also to recognize the
important incidence effects of the introduction of a tax, sometimes called "transition incidence." Actually, a
full description of a tax policy includes not only the rates and base but also the time path and conditions with
which the rate and base are introduced. Although we tend to take up transition incidence as a separate
phenomenon because the analysis is otherwise simply too complex, it would be preferable to conceive of tax
policy in terms of variation in the time paths of instruments such as rates and deductions. A stronger tradition
of thinking in dynamic terms would perhaps lead us to pay more attention than we customarily do, for
example, to the difference in investment incentives generated by a permanent and an on-again-off-again
investment credit.13
Â
Page 78
In any case, the phenomenon of

transition incidence is important,
as may be illustrated by
reference to the wage tax versus
consumption tax distinction. A
wage tax and a consumption tax,
as distinguished earlier in this
chapter, have the same
steady-state properties.
However, introduction of a
consumption tax imposes a
windfall loss on owners of
capital (on average, the older
generations), whereas
introduction of a wage tax
bestows a windfall gain on them.
Intuitively, the consumption tax
applies to the accumulated
capital, and in a sense
confiscates a fraction of it. The
wage tax instead imposes the
burden only on those in the
earning phase of the life-cycle.
Exactly this phenomenon is
referred to in the opening section
of this chapter.
Vertical Distributional Effects
In this section I focus on steady-state comparisons, to the neglect of wealth redistributions that may take
place upon transition.
74
75
Probably most interest centers on the vertical distribution of consumption tax burdens. In this connection, the
contrast between the ''snapshot" and the long-run or lifetime perspective is of particular importance, as is the
related question of how we classify people according to level of economic well-being. A proportional tax on
a broad-based measure of consumption will be regressive measured against a single year's income. That is,
the ratio of consumption tax to income will be a declining function of the amount of income.
The picture changes if the standard of well-being is lifetime resources, understood to be the discounted value
of a person's labor earnings and transfers received. If individuals are uniform in the ratio of bequests they
leave to lifetime resources, a proportional consumption tax will also be in the same proportion to lifetime
resources for all households. Although we know remarkably little about the bequest behavior of U.S.
families, 14 it is likely that the well-to-do bequeath on average a larger fraction of lifetime resources than do
poorly endowed families. In that sense, a fiat consumption tax will also be regressive in some degree when
measured against lifetime resources, although not when measured against the resources of the sequence of
individuals in a bequest chain. Davies, St-Hillaire, and Whalley who explored these issues in the context of a
simulation model based on Canadian data, found surprisingly little increase in the ratio of bequests to lifetime
resources with increasing resources.15
Â
Page 79
Their study also supports the intuitive expectation that lifetime incidence calculations indicate more
progressivity of the tax system as a whole than do annual snapshots, and the calculations are much less
sensitive to variations in the assumptions made about the incidence of the major taxes.
The vertical distribution can also be affected by selectively reducing the rate of tax applicable to commodities
making up a relatively large portion of the expenditures of the poor. However, since virtually no broad
category of consumption is wholly absent from their expenditures, there are distinct limits to what can be
done to modify the vertical distribution through these means. Davies, again referring to Canadian data, found
that the usual sorts of special rates (on food, clothing, and shelter) do rather little to change the vertical
distribution of burdens, with an exemption for clothing actually reducing progressivity because it forms a
larger fraction of the budgets of the rich than of the poor. 16 By contrast, the application of graduated rates to
a consumption base or to an earnings base, as in the X-Tax, permits wide latitude to vary the vertical
distribution of burdens.
Horizontal Distributional Effects
All taxes discriminate among individuals with different characteristics. For example, an earnings or income
tax imposes a larger burden on individuals with high earning power or a taste for working long hours than
would a uniform lump-sum tax. A consumption tax shares this property of an income tax.
Taste for Saving
The most obvious way in which a consumption tax differs from an income tax is in the variation of burdens
among people with different tastes or necessities to save. In the context of the assumption that people with
the same lifetime resources as defined above have access to the same consumption possibilities ("perfect
capital markets"), a consumption tax that satisfies the second property (no tax on savings) is neutral among
equally endowed individuals. By contrast, an income tax places a relatively heavy burden on those who save,
or rather on those whose lifetime resources are paid to them relatively early and on those whose tastes favor
later consumption.
Â
75
76
Page 80
Taste for Goods
In much the same way, consumption taxes that differ in their inclusion of different commodities create
differences in burdens among individuals according to their taste for the goods that are relatively heavily
taxed. A nonsmoker, I find the burden of the tax on tobacco quite bearable, but I would welcome relief from
the tax on wine.
Specialized Production Factors
The discussion thus far has stressed the commodity consumption side of the story. But people differ in their
ownership of productive factors as well. Anyone owning resources with specialized application to particular
commodities will naturally suffer relatively heavily from taxes on those commodities. The taxation of energy,
involving significant amounts of rent obtained from ownership of resources of little value in alternative uses,
is a particular case in point.
Transitional Incidence
Age/Generation Differences: The Case of Directly Owned Capital
Differences in ownership of specialized resources is of particular importance in connection with the short-run
transition, when careers and other long-lived commitments are difficult to change. In the long run, resources
of talent and labor supply are quite flexible in their uses, and capital can be directed to a variety of
applications. However, the length of the adjustment period is doubtless very long in some cases and in the
short run human, intellectual, and physical capital are fixed, and transition incidence is a significant
phenomenon.
The most-studied such effect is the tendency, mentioned above, of the introduction of rules that favor new
capital over old to induce a loss in value of existing assets. In the simplest case, with highly durable capital
and no costs of adjusting the level of the capital stock, introduction of a consumption tax at some flat
fractional rate effectively expropriates the same fraction of the existing capital stock. (This accounts for the
finding in several studies that introduction of a flat-rate consumption tax leads to higher long-run living
standards than does introduction of a flat-rate wage tax. The wealth expropriated by the consumption tax is
used to finance lower tax rates and higher consumption for future generations.) 17 This effect is
Â
Page 81
moderated if adjustment costs give existing capital owners an advantage in exploiting the newly profitable
investment opportunities.
Because it imposes an implicit windfall tax on wealth, introducing a consumption tax also has
intergenerational incidence effects. It is usual to model the economy as though the older generation owns the
existing stock of capital, planning to sell it to the younger generations to finance retirement consumption. To
the extent this model is accurate, the transition effect of introducing a consumption tax results in a
redistribution from the old to the young and future generations.
Portfolio Differences
However, the story is complicated by the availability of financial assets. It is quite possible for the young to
76
77
own the real assets, having issued debt to the old. In this case, introducing a consumption tax will not affect
the consumption of the old; the young will bear the cost.
One may think that owners of debt will bear the consumption tax by virtue of price inflation that its
introduction will induce. However, this confuses two things: the effect of the tax and the determinants of the
terms of borrowing and lending. If appropriate inflationary expectations are built into the terms of the loan,
creditors will not lose from introduction of the tax. To put the matter more simply, if the lending is carried
out in real purchasing power terms (indexed bonds would be the obvious mode), lenders will be unaffected
by the introduction of the tax.
Specialized Production Factors
Finally, we should note that the differential effect on owners of specialized resources applies in particular to
the introduction of different rates of tax on different commodities. In an elegant extension of general
equilibrium modeling techniques, Goulder and Summers have put plausible quantitative dimensions on the
differential impact on the value of firms in broad industry categories of introducing a variety of tax
alternatives. 18 Their figures show substantial differences in capital value changes across sectors, with
considerable sensitivity to the degree to which the policy change is anticipated in capital markets.
Â
Page 82
Concluding Comment
This chapter has three themes: First, there are several quite different methods of implementing a consumption
tax. Economically these approaches are very similar, if not identical, in their effects, but they appear very
different to many of those involved in the policy debate. Once the connection among the different methods is
understood (for example, the two forms of the illustrative X-Tax discussed in the chapter), it becomes clear
how policy decisions in one (for example, the treatment of owner-occupied housing) translate into the same
decisions in the others. Analysis suggests that some politically contentious choices such as that between an
origin and a destination basis for a value-added tax may be of little economic significance. Similarly, some
hopes raised by consumption taxes, such as the expectation of dealing with the underground economy, are not
supported by economics. Second, because they are so similar economically, in their flat-rate form all of these
taxes spread the burden of taxation similarly, namely, in proportion to a household's discounted lifetime
consumption. The most commonly employed approach to introducing progressivity to these taxes, namely,
exemption of purchases of particular commodities or services from tax, is very limited in its power to alter
the vertical distribution of burdens. However, alternative methods are available to introduce any degree of
progressivity desired by policymakers. Third, although the various consumption taxes are similar in their
long-run incidence, they may differ significantly in "transition incidence," the effective taxation of wealth
implied by their introduction. Intuitively, we can think of a choice between taxing or not taxing consumption
funded out of past saving. Some care is required, however, to determine the transition incidence of a
particular tax, and such matters as the choice of financial portfolio by a household (between stocks and
bonds, for example) may make a critical difference.
Policy positions on the various forms of consumption tax are often based on inadequate models of incidence,
and public discussion suffers from confusion about the alternative approaches that might be taken to taxation
based on consumption. A greater appreciation in the policy debate of the basic similarity of the alternative
approaches to taxation based on consumption would improve the chances of avoiding complex rules that ill
serve the objectives they are introduced to achieve.
Â
77
78
Page 83
Notes
1. See, for example, the following references: Alan J. Auerbach, "Corporate Taxation in the U.S." Brookings
Papers on Economic Activity (Washington, D.C.: The Brookings Institution, 1983), p. 2. Alan J. Auerbach
and James R. Hines, "Tax Reform, Investment, and the Value of the Firm," National Bureau of Economic
Research Working Paper No. 1803, Cambridge, Mass., Jan. 1986. Alan J. Auerbach and Lawrence Kotlikoff,
"Investment versus Savings Incentives: The Size of the Bang for the Buck and the Potential for
Self-Financing Tax Cuts," in Laurence H. Meyer, ed., The Economic Consequences of Government Deficits
(Boston: Kluwer-Nijhoff, 1983), pp. 123-149. J. Alan Auerbach and Lawrence Kotlikoff, Dynamic Fiscal
Policy (New York: Cambridge University Press, 1987).
2. See, for example, the following references: Henry J. Aaron and Harvey Galper, "A Tax on Consumption,
Gifts, and Bequests and Other Strategies for Tax Reform," Options for Tax Reform (Washington, D.C.: The
Brookings Institution, 1984), pp 106-146. David F. Bradford, Untangling the Income Tax (Cambridge, Mass.
Harvard University Press, April 1986). David F. Bradford and the U.S. Treasury Tax Policy Staff, Blueprints
for Basic Tax Reform, 2nd ed., reg. (Washington, D.C.: Tax Analysts, 1984; originally published by U.S
Treasury, 1977). Nicholas Kaldor, An Expenditure Tax (London: Allen and Unwin, 1955; Westport, Conn.:
Greenwood Press, 1978). U.S. General Accounting Office, "Tax Policy, Choosing Among Consumption
Taxes," Staff Study, August 20, 1986. U.S. Treasury Department, Tax Reform for Fairness, Simplicity, and
Economic Growth Vol. 1, Overview; Vol. 2, General Explanation of the Treasury Department Proposals;
Vol. 3, Value-Added Tax (Washington, D.C.: U.S. Government Printing Office, November 1984).
3. Robert E. Hall and Alvin

Rabushka, Low Tax, Simple Tax,
Flat Tax (New York:
McGraw-Hill, 1983) and The
Flat Tax (Stanford, Calif..
Hoover Institution Press, 1985).
4. Bradford, Untangling the Income Tax.
5. Bradford et al., Blueprints for Basic Tax Reform.
6. For more extended discussion of this issue, see ibid., esp. ch. 8.
7. See, for example, Auerbach and Kotlikoff, "Investment versus Savings Incentives," and Dynamic Fiscal
Policy, as well as James B. Davies and France St-Hillaire, "Reforming Capital Income Taxation in Canada:
Efficiency and Distributional Effects of Alternative Options," mimeo, Department of Economics, University
of Western Ontario, London, Canada, 1986.
8. For a good discussion see Henry Aaron, ed., The Value Added Tax: Lessons from Europe (Washington,
D.C. The Brookings Institution, 1981), or Charles E. McLure, Jr., The Value-Added Tax: Key to Deficit
Reduction? (Washington, D.C.: American Enterprise Institute, 1987).
9. See, for example, Bradford, Untangling the Income Tax, pp. 328-329, and Gene M. Grossman, "Border
Tax Adjustments: Do They Distort Trade?" Journal of International Economics 10 (1980): 117-128.
10. James B. Davies, "Manufacturers' Sales Tax, Value-Added Tax, and Effective Tax Incidence," Canadian
Tax Foundation, Thirty-Seventh Tax Conference, Toronto, 1986.
Â
78
79
Page 84
11. Harvey S Rosen, "Tax Illusion and the Labor Supply of Married Women," Review of Economics and
Statistics 57-2 (May 1976) 167-172.
12. As, for example, in Joseph E. Pechman, Federal Tax Policy, 4th ed. (Washington, D C.: The Brookings
Institution, 1983) and Who Paid the Taxes, 1966-85? (Washington, D.C.: The Brookings Institution, 1985),
and in Joseph E. Pechman and Benjamin A. Okner, Who Bears the Tax Burden? (Washington, D.C.: The
Brookings Institution, 1974).
13. See Auerbach and Hines, "Tax Reform and the Value of the Firm," and David F. Bradford and Charles
Stuart, "Issues in the Measurement and Interpretation of Effective Tax Rates," National Bureau of Economic
Research Working Paper No. 1975, Cambridge, Mass., July 1986.
14. Bradford et al., Blueprints for Basic Tax Reform, pp. 169-173.
15. James B. Davies, France

St-Hillaire, and John Whalley,
"Some Calculations of Lifetime
Tax Incidence," American
Economic Review 74
(September 1984) 633-669.
16. Davies, "Manufacturers' Sales Tax, Value-Added Tax, and Effective Tax Incidence."
17. See Auerbach and Kotlikoff, "Investment versus Savings Incentives," and especially Dynamic Fiscal
Policy.
18. Lawrence H. Goulder and Lawrence H. Summers, "Tax Policy, Asset Prices, and Growth A General
Equilibrium Analysis," National Bureau of Economic Research Summer Institute Paper, Cambridge, Mass,
July 1986.
Â
Page 85
4
Fundamental Issues in Consumption Taxation
1
Introduction
Motivated by a desire to simplify compliance and improve incentives, many are interested once again in
restructuring the U.S. tax system. Most of the proposed reforms would move toward a system based on
consumption, rather than income. My purpose in this study is to draw attention to the similarities and
differences between the two approaches and to lay out important problems of transition to the
consumption-based taxes at the heart of the reform plans.
Four consumption-based reform plans are currently under active consideration:
79
80
a retail sales tax
a "flat" tax
the Unlimited Savings Allowance (USA) tax
a value-added tax
All four systems are conceived of as replacements for the federal income tax, both corporate and individual.
In addition, the USA tax integrates the social security taxes through a system of credits.
As far as I know, no actual legislative proposal for a federal sales tax has yet been offered, but Representative
Bill Archer, chairman [in 1996] of the Ways and Means Committee of the House of Representatives, and
Senator Richard Lugar have been notable supporters. A number of value-added tax proposals have been
made, including a highly detailed plan introduced by Representative Samuel Gibbons. Representative
Richard Armey is a particularly well-known advocate of a flat tax, although several others have advanced
similar proposals. All of these are modeled on the flat tax developed by
Â
Page 86
Robert Hall and Alvin Rabushka

(1983, 1995), which I take as
representative of the breed.
Senators Pete Domenici and Sam
Nunn developed the USA tax and
introduced it in April 1995. 1
All four approaches would bring about a shift of the U.S. tax system to a consumption base. In terms of
implementation, the four have certain features in common. The first approach, a retail sales tax, is a
proportional tax paid by businesses. The second and third, the flat tax and the USA tax, both consist of
integrated systems. Each uses a proportional tax paid by businesses, more or less similar to a value-added tax,
and a personal-level tax, along the lines of the existing individual income tax. In the case of the flat tax, the
individual tax is imposed only on a person's compensation, such as wages and salary, which is deducted from
the base of the business tax. The individual-level tax in the USA system is an example of what has come to
be called a "consumed income" tax. That is to say, it is based on something like the present taxable income
with a deduction for net saving (and inclusion of net dissaving). In the USA system, there is no deduction
from the business tax base for payments to individuals (although there is a coordinated set of credits for
payroll taxes).
My purpose here is not to provide a detailed description of any of these reforms. Instead, I use uniform
(single-rate) income and consumption taxes,
to show that the two approaches are, in principle, much more similar than is generally understood
to point out how it has proved difficult, in practice, to implement an income tax that treats investment
uniformly and that is not affected by inflation
to explain how the consumption approach is naturally inflation proof and naturally treats different kinds of
investment uniformly
to indicate the somewhat subtle ways in which shifting from an income to a consumption base may impose a
one-time tax on "old savings" or "old capital"
80
81
to indicate the tradeoffs that must be confronted in dealing with this phenomenon
to show how price-level changes that may or may not accompany a transition affect the distribution of gains
and losses
to sketch out how a transition might affect interest rates and asset prices (including owner-occupied housing)
Â
Page 87
to explore the case in equity for protecting the tax-free recovery of old savings
to emphasize the incentive problems that arise if savers and investors anticipate a change in the tax rate in a
consumption-based system (A transition from zero to some positive rate upon introduction is a particularly
important instance.)
By now, a considerable body of literature addresses the effects of time-varying tax policy, including
transitions of the sort considered here. These analyses are complementary to my undertaking, which attempts
both less, in raising but leaving open the answers to questions about the quantitative effects of policy
changes, and more, in addressing aspects of transition on which the models to date are largely silent. For
excellent examples of the technically more detailed models, see Auerbach and Kotlikoff (1983, 1987), Howitt
and Sinn (1989), Keuschnigg (1991), and, especially, Sarkar and Zodrow (1993).
The study is organized as follows: In section 2, I explain how uniform consumption and income taxes can be
implemented at the level of the business firm and develop the key differences between the two approaches.
Section 3 describes the main transition issues, and section 4 addresses possible ways of dealing with them. In
section 5, I touch on some of the arguments that go to the merits of seeking to moderate transition effects.
Section 6 contains brief concluding remarks.
2
Key Concepts in Consumption and Income Taxation
As I have suggested, all four reform models bring about a switch to a consumption base. Just why this is so
will be made clear by consideration of how one might administer either an income- or a consumption-based
tax in the form of a tax paid by businesses.
Many people are familiar with the idea that a consumption tax can be administered in this way. Indeed, a
retail-sales or value-added tax is often identified with "consumption tax." The idea that an income tax could
be administered in much the same way is less familiar. It becomes obvious, however, if we start with a
description of what is called in the jargon of the tax trade a subtraction-method value-added tax of the
consumption type. The subtraction method is not the approach usually employed in value-added taxes, but it
is the
Â
81
82
Page 88
approach generally advocated in the United States. More important for my purpose here, it has a clear
connection with the familiar income tax paid by businesses.
A Value-Added Tax of the Consumption Type
The building block of a value-added tax, regardless of the method used, is the business firm subject to tax.
Policy makers have quite a bit of scope in deciding just what is a ''business firm" for this purpose. But I take
for granted that no distinction is made between businesses according to legal form. In particular, there would
normally be no difference between the treatment of corporations and other businesses, such as partnerships or
proprietorships. So business firm is not the same thing as corporation, and business tax should not be
identified with corporation tax.
Under a "subtraction method" of implementing a value-added tax, the tax base of a business consists of the
difference between the payments it receives for sales of goods and services of any kind (including sales of
assets, such as a building) and the purchases of goods and services from other firms. This total is then taxed
at some predetermined fixed rate. That is it. In the ordinary case, financial transactionssuch as borrowing and
lending, issue and repurchase of stock, payments and receipts of dividends, and the likedo not enter the
calculation of the taxable base. 2
Because what is sold by one business to another results in an increase in the tax base of the selling firm and a
deduction at the same time and in exactly the same amount by the purchasing firm, and since both are subject
to tax at the same rate, transactions between businesses give rise to no net tax liability to the government.3
The only circumstance under which a net tax liability is created is when there is a sale by a business to "the
public," or more precisely, to a person or organization that is not a business firm subject to tax. At that point
there is no deduction taken to offset the tax paid by the seller.
The aggregate business tax base is thus the sum of all sales by business to nonbusiness, which is a measure of
aggregate consumption, one reason this is aptly called a consumption tax. (Another reason for calling it a
consumption tax is the way saving and investment are treated, a matter I address later.)
Â
Page 89
How Does the Value-Added Tax Relate to Other Reforms?
Retail Sales Tax
The subtraction-method value-added tax of the consumption type can be helpfully related to a variety of
apparently different tax regimes. For example, because of this netting of the taxes on interbusiness
transactions, a consumption-based subtraction-method value-added tax would result in the same revenue flow
to the government as a tax on business-to-nonbusiness sales. In other words, a consumption-based
subtraction-method value-added tax is fundamentally the same economically as a retail sales tax, even though
it may look different and even though the two might be administered differently. 4
Invoice-and-Credit Value-Added Taxes
There is also a simple equivalence between a subtraction-method value-added tax and the invoice-and-credit
form of value-added tax typically employed in other countries. European tax systems, for example, use the
invoice-and-credit type of value-added tax. Under this method of taxation, the selling firm pays a tax on all
sales, with no deductions, but claims a credit for taxes paid by the seller on purchases from other firms. The
amount of the credit is the amount shown in the invoice from the selling firm, hence the name. If the tax rate
82
83
of both businesses is the same, the credit obtained by the buying firm exactly equals the value of the
deduction that would be claimed under the subtraction method. Thus, under both methods, a
business-to-business sale results in a simultaneous, equal tax liability for the seller and tax credit for the
buyer. No tax is paid to the government until the product or service is sold to a buyer who is not a taxable
firm. This type of tax results in the same flow of revenues to the government as the subtraction-method
value-added tax, or a retail sales tax for that matter, with the proviso that the same goods and services are
subject to tax at the same rate.
Flat Tax
As has been mentioned, the flat tax consists of a coordinated pair of taxes, one levied on businesses, the other
on individuals. The business tax component of the flat tax is a subtraction-method value-added tax of the
consumption type with an important additional deduction for
Â
Page 90
the business firm: payments to workers. 5 In turn, workers pay tax only on what they are paid by employers.
The key insight is that if the tax on what workers receive is at the same rate as applies to employers, the net
effect is the same as though there had been no deduction of the payments by the employer and no tax paid by
the worker. In other words, the system would produce the same result as a subtraction-method value-added
tax.
Under the flat tax, the rate of tax is the same for workers and employers. But for workers, the flat tax applies
that rate only to the excess over an exempt amount that depends on the worker's family size. So the flat tax
can be understood as (a) a subtraction-method consumption value-added tax, coupled with (b) a scheme
based on workers' earnings designed to make the system progressive.
USA Tax
The USA tax also consists of a coordinated pair of taxes, one levied on businesses, the other on individuals.
The business tax component of the USA tax is a subtraction-method value-added tax of the consumption
type. To a first approximation (I neglect many fine points), the individual component of the USA tax can be
understood as levied on a family's total consumption. The way it works is that the individual or married
couple is taxed on all receipts of any kind, including receipts that result from sale of assets, with a deduction
for any saved amounts, that is, for amounts used to acquire assets. The net amount, receipts not devoted to
savings, must be used either for the purchase of consumption or for the payment of taxes. Under the
individual component of the USA, the family is taxed on this measure of consumption according to graduated
rates, just as in the existing individual income tax.6
What Kind of Beast is a Consumption-Type VAT?
A Stage of Production Tax?
A value-added tax is sometimes described as a tax levied at each successive stage of production. As applied
to a uniform tax, this description is somewhat misleading because a sale at any intermediate stage in the
production process that is subject to tax gives rise to an immediate offsetting deduction from a firm's tax base
at the next stage of production. So no net tax is paid until the final stage of production, when the product is
sold to the public.
Â
83
84
Page 91
A Transaction Tax?
Sales and invoice-and-credit type value-added taxes are said to be transaction taxes. Some people draw a
contrast between these types of taxes and income or subtraction-method value-added taxes, believing that the
former are accounted for on a transaction-by-transaction basis, while the latter are aggregated on the
company's books. In some contexts (such as regulatory accounting), the difference is of some significance,
but it is a matter of legal form rather than economic substance. To figure out a company's income (and audit
its tax accounts), we must add up individual amounts received (or payable) and subtract individual payments
made (or payable). Under either a "book income" or "transaction tax" regime, it is necessary to keep track of
and monitor individual transactions.
Under either regime, there is an enforcement advantage if the amount deducted from the tax base of (or
amount of credit claimed by) one business should be matched exactly by an amount included in the tax base
of (or amount of tax paid by) another. This will hold for consumption-type systems using either the
subtraction or invoice-and-credit approaches. Under an income tax, however, this exact symmetry breaks
down, because some of the deductions of the business buying capital equipment are spread over time in forms
such as depreciation allowances. 7
A Tax on Profits?
Some assert that a value-added tax is the same as a tax on all the payments to factors of productionwages,
salaries, and profits. This depends on how terms are defined.
As defined by accountants (economists sometimes use a different definition), profit includes the "normal"
return entrepreneurs receive for waiting and taking risks plus any deviation, positive or negative, that results
from the varying but statistically unpredictable fortunes of the firm. The return for risk taking, both its normal
and its unpredictable component, is taxed identically under both an income tax and a consumption-type
value-added tax. By contrast, because of differences in how a business firm's capital purchases are treated,
the normal return for waiting is subjected to tax in an income tax but not in a consumption-based value-added
tax.
The point is worth emphasizing. Like an income tax, a consumption-based value-added tax subjects to tax
much of what is ordinarily understood as profit. If, for instance, a business discovers oil on its
Â
Page 92
property, all the payoff is subject to tax. So is the reward for an innovation, such as the development of a
successful software product, or the return on intangible property such as a trademark established through a
successful advertising campaign. With a consumption-based value-added tax, the general public becomes a
proportional shareholder in all enterprises. 8 If profits exceed the normal rate of return (risk adjusted), the
general public shares in the good fortune. If they fall short, the general public, having "invested" via the
deduction for investment outlays, shares in the shortfall.
Investments and Capital Income
With the value-added tax, business outlays for investment purposesadditions to inventory, to the stock of
buildings, or to fixed equipment, for exampleare deducted immediately. A successful investment will
generate future tax liabilities that outweigh losses in revenues to the government due to current deductions.
For a marginal investment, the cash flows consisting of the combination of the current tax saving and the
84
85
anticipated future taxes due on the payoff to the investment will have a value of zero in the capital market.
(More conventionally, if less precisely put, the expected net present discounted value of the combination is
zero.) In this sense, a consumption-based value-added tax exempts income from capital. For what economists
call an "extramarginal" investmentan investment opportunity that will beat the marketthe tax is positive.
(More precisely, the anticipated distribution of cash flows to the government will have a positive value in the
capital market.)
In contrast, under an income-based tax, the deduction of the investment outlay is postponed. Conceptually, it
is recovered as the value of the asset is depleted. This timing difference means that the profile of cash flow to
the government associated with an investment that is a barely breakeven proposition will have a positive
market value.
What is exempt from tax in a consumption-type tax system can be expressed in a variety of ways. One way is
as the yield on a riskless investment, typically taken to be the Treasury bill rate. A consumption-based
value-added tax allows investors to receive the Treasury bill rate on riskless investments, free of tax. A
riskless rate of return in excess of this rate is taxed. A riskless rate of return below the market rate is
compensated, in effect, through reduced tax. In contrast, under a true income tax, returns in excess of zero are
taxed. To put this
Â
Page 93
in quantitative perspective, the real riskless rate of return available to a tax-exempt investor has historically
been below 1 percent per year. 9
Is Risk Taking Taxed?
The treatment of the reward to bearing risk requires some analysis. Most assets do not have a risk-free return.
Even Treasury bills suffer from the risk of unpredictable inflation. The average real rate of return available to
a tax-exempt investor in the stock market has historically been about 9 percent per year.10 This higher
average return is bought at the price of substantial risk. Can one also say that under a consumption-type tax,
the investor obtains the reward to risk taking free of tax?
The answer is a qualified yes, but this is also the answer to the same question with regard to an income tax.
This point is, I think, not widely understood probably because we are accustomed to seeing risk and waiting
intertwined.11 Under a uniform proportional income tax, a pure bet that does not involve time is shared with
the tax collector on fair terms, through the allowance for losses.12 If there is a risk premium, a uniform tax
will have positive expected revenue, but it will not impose a burden on the investor. The positive expected
revenue is the risk premium collected by the government for assuming part of the risk, via the tax system.
The same result is obtained under a consumption-type tax, except that the cash flow to the government may
not be simultaneous with the outcome of risk taking, so that the argument needs to take into account
discounting for differences in timing. Essentially, an income tax and a consumption tax treat risk the same.
Essentially, under both systems, any positive expected revenue is the market-determined reward for the risk
that the government takes and is not a burden to the taxed investor.
Is a Value-Added Tax a Wage Tax?
It is often said that a value-added tax of the consumption type is equivalent to a tax on wages or a tax on
labor income. Again, whether this is so depends on one's definition of the terms wages or labor income. The
payoff to the oil gusher or the successful advertising campaign or the development of Microsoft DOS might
not be regarded as wages or labor income in ordinary parlance. But these are subject to a value-added tax of
the consumption type.
85
86
Â
Page 94
As the discussion above of the effective taxation of risk bearing suggests, in either case the tax may or may
not represent a burden. We may think of the activity of an individual entrepreneur or firm as one of selection
from "ideas" that arise, perhaps randomly or perhaps as a result of investment of funds and effort. The
unprofitable ideas are rejected. Those that are accepted are believed to be, at least, breakeven propositions in
the sense that the associated distribution of cash flows will have a market value of at least zero.
Both consumption- and income-tax systems will place a burden on profitable ideas. The resulting incentive
effects will depend on the details of the process by which ideas are generated. 13 To the extent that oil gushers
and the like represent just the upside of investments that were expected to produce the risk-adjusted market
rate of return, the positive revenue in a consumption-type system is just the market-determined reward to the
government (that is, other taxpayers) for taking on risk. Under a true income tax there is, in any case, a
burden put on postponement of consumption, as reflected in the risk-free real interest rate; in the same sense,
a disinvestment project is subsidized.
Is Owner-Occupied Housing Taxed?
The income tax in the United States exempts the yield from investment in household capital, by which I mean
durables such as automobiles, boats, washing machines, electronic equipment, and the like. Quantitatively,
the most important example is owner-occupied housing (including second homes). To capture such capital in
an income tax would involve treating the household as in the business of selling to itself the services of these
investments. Because the income tax does not attempt to do this, it puts such forms of investment at an
advantage, relative to ordinary business capital.
One of the features of consumption-based taxes that makes them appeal to economists is that they could, in a
simple way, tax household capital on an even plane with business capital. (See, for example, Jorgenson
1995.) This result would obtain not by setting up the household as a business under a value-added tax of the
subtraction type. Rather, these implicit businesses are left out of the set of firms subject to tax. The advantage
for these implicit firms is that they are not subject to tax on their implicit sales (for example, the sale of
housing services by owner-occupiers to themselves). The disadvantage is that they are not eligible to deduct
(in the case of the invoice-
Â
Page 95
and-credit approach, obtain credit for the taxes paid on) their purchases. With a taxable business, the general
public shares in the investment and payoffs in proportion to the tax rate. In making investment decisions, the
taxable firm considers its share. For a tax-exempt household "business," the general public does not share in
the investment or the return. The investment decision is based on the full cost and the full return. The
breakeven requirement will be the same for a taxable and tax-exempt business.
If the sale of housing is treated like all other sales under a subtraction-method value-added tax, it will remove
the tax advantage that owner-occupied housing enjoys under the income tax. 14 A question of some interest is
what this change will imply for the value of existing houses. I take up this question below.
A Value-Added Tax Could be an Income Tax
86
87
It may not be well known that a value-added tax could be used to implement an income tax. To do so
requires, essentially, using conventional income accounting for the firm's income, instead of the cash-flow
accounting used in the consumption-type tax, although to do it "right" would require making systematic
corrections for inflation.15 An income statement should report the change in the firm's net worth during the
year. When a firm acquires a unit of capital equipment, it exchanges one asset (money in the bank) for
another (the purchased equipment). If the investment is wise and expected in the capital market, the value of
the firm does not change. That is the logic behind the fact that accountants "capitalize" such outlays for
purposes of income accounting, which means they are not deducted currently. Over time, however,
equipment typically declines in value. This decline must be accounted for in the income statement, even
though it does not correspond to current cash flow. That is why accountants take depreciation charges in their
measure of current income.
Similarly, when goods are acquired to augment inventory, the cash outlay for purchases does not result in a
change in the business's net worth. It is added to the stock of inventory assets. The accountants register the
cost of producing or acquiring inventory later, in the form of a deduction for the cost of goods sold.
Associating past outlays for inventory with current receivables is one of many problems related to pinning
down accounting concepts for income purposes
Â
Page 96
that do not arise in the cash-flow accounting system appropriate for a consumption-based tax system.
The essential difference between income accounting and consumption-based accounting is timing. A tax on a
firm's income would be based on the difference between amounts received from sales and a measure of the
firm's costs, including recovery of past outlays for equipment, inventories, and such. When a
business-to-business sale occurs, the purchasing firm deducts the amount paid from its income calculation.
Generally, that deduction is less than the invoiced price because the purchaser must capitalize some of the
outlay, to be recovered in future income calculations.
There is a sense in which a true income tax, by contrast with a subtraction-method value-added tax, would
effect a tax at each stage of production. This point is most easily understood by assuming that each stage of
production is carried out by a different firm. 16 The aggregate amount subject to tax would equal the
difference between the value of the goods and services sold and the discounted value of the future deductions
by purchasing firms in the form of depreciation allowances and other forms of cost recovery. The aggregate
of the taxable incomes of all the firms would equal the sum of the sales to the public and the net increase in
the stock of capitalized past outlays in the hands of the firms. As described above, the former is aggregate
consumption, and the latter (which could be negative) is a measure of the increase in aggregate net worth, or
saving.17 (Theoretically, the tax calculation for a vertically integrated company that combines several stages
of production should lead to the same result.)
This exercise highlights four important aspects of consumption and income as tax bases:
First, it emphasizes the role of timing. The distinction between an income and a consumption tax is
essentially a matter of timingthe time value of money. (In both systems the costs of production are recovered,
but costs are recovered earlier under a consumption tax.)
Second, it demonstrates that an indirect tax can be employed to implement a uniform income tax, just as one
is accustomed to thinking one can use such a system to impose a uniform consumption tax.18
Third, it highlights important differences in accounting. Income accounting is more difficult than cash-flow
accounting. That diffi-
87
88
Â
Page 97
culty is responsible for much of the complexity in the current income tax system. (This is a major theme of
Bradford 1986.)
Fourth, it explains why, at any given time under an income tax a company has a stock of past outlays that
may be thought of as an accumulation of tax-prepaid claims on future consumption. These claims,
represented by "basis" in the company's assets, belong to the company's owners. This assumes great
significance in any transition from an income- to a consumption-based tax system.
Deductions for
Wages and
Interest and the
VAT
With the subtraction-method value-added tax regime, little would change if companies were permitted to
deduct payments to employees, provided those payments were subject to tax, at the same rate. (That is, the
tax saving to the employer would exactly equal the extra payments to employees necessary to leave them in
the same place after tax). The government would receive the same cash flow, and the economic effects
should be identical. More generally, payments to workers might be subject to tax at a different rate, or on a
graduated schedule, without changing the essential economics. As we have noted, if the schedule were to
consist of an exempt amount plus a single rate for amounts above the exempt amount (the same single rate
applied to firms), the result would be what we have identified as a flat tax.
The same idea extends to interest payments in a value-added tax regime. Little of economic substance would
change if companies were allowed to deduct interest paid to creditors, as well as payments to employees if
the tax rate for these income recipients was the same as the business-level tax. Apart from graduated rates,
this is a rough approximation of how the combination of corporate and individual income taxes currently
operates. If the same tax rate applies to the deduction by the firm and inclusion in taxable income of
bondholders, the result should be the same as a value-added tax with no such deductions. 19
The concept of deducting interest and taxing it at the bondholder level is introduced here because the
commitment to make interest payments according to a fixed contract creates an important problem of
transition to a regime with no interest deduction. Bondholders can be expected to require compensation for
assuming tax obligations; thus the stated interest rate will be different when interest can be
Â
Page 98
deducted from when it is not. When no tax is incurred at the bondholder level, the bondholder receives an
after-tax interest rate; with the other treatment of interest, the stated rate will be the before-tax rate.
3
Transition Issues
We often describe the distribution of tax burdens in terms of the progressivity of a continuing systemthe
relative tax burdens imposed on people at different levels of income. But a major change in the tax system
88
89
causes one-time gains and losses that may be only loosely related to the progressivity of the system either
before or after the change. Introduction of a value-added tax of the consumption type provides an excellent
and highly relevant instance of this point.
Imagine that we have been operating under a proportional value-added tax of the income type and a decision
is made to switch to a consumption-type tax (basically by substituting expensing of investment outlays for
depreciation allowances over time). Under both these tax regimes, broadly speaking, burdens are spread
proportionally to income levels. In the course of a switch from one to the other, however, there may be
one-time burdens that fall much more heavily on owners of business capital, on those who have saved in the
past, and on older citizens than on others. Indeed, the ''others" may be significant gainers. One may like or not
like this result. Depending on what one wants to emphasize, one may say the transition effect is highly
progressive or very unfair to those who have accumulated savings in the past. But it is unlikely the effect will
bear any systematic relationship to the usual standards by which we judge tax distributions.
Whether such effects will be significant will depend on the allowance made in the rules for transition to the
new system. By examining the particularly simple case of uniform income and consumption taxes, we can
highlight the essential problems. The existing system is approximated by a value-added tax of the income
type and the reforms as a switch to a value-added tax of the consumption type. This switch can be analyzed
by a two-step procedure. First, consider the effect of introducing the value-added tax of the consumption
type, and then consider the offsetting effect of eliminating the tax of the income type. To start with, I focus on
the cases where there is only a
Â
Page 99
business tax, with no deduction of wages or interest and no taxation at the individual level.
One-Time Asset Tax
One concrete example captures the essence of the problem. It is the case of a retail store, whose owners have
purchased, on the day before the value-added tax is introduced, a stock of canned goods for $10,000. They
sell the goods the day after the value-added tax is introduced, say at a rate of 20 percent. In calculating the
business's value-added tax liability, the proceeds from the sale of the canned goods are on the inclusion side
of the ledger. In the normal course of events, there would have been a deduction in the past for the purchase
of those goods. But at the moment of introduction of the new tax, the purchase of the goods is already a thing
of the past. Applying the new rules, there is a tax of $2,000 on the sale of the canned goods from inventory
but no offsetting deduction. Barring special rules, the effect of introducing the new system is to impose a
onetime tax, at the value-added tax rate, on the stock of inventory at the time of transition.
This story, writ large, captures the essence of the way introducing a consumption-type tax, with no special
transition rules, would impose a one-time tax on the stock of wealth in the economy, so-called old capital or
old savings. In the example, the extra tax on the stock of inventory was imposed immediately, because the
goods were sold to the public on day one of the new system. In actuality, the tax payments that give effect to
the one-time tax would take place over time, for example, over the lifetime of a piece of fixed equipment. But
the discounted present value of the extra tax imposed is the same as if the assets were sold immediately to the
public (provided the tax rate is constant). So the effect is the same, even though the payments are spread out
in time.
Most people, of course, do not own retail stores. Instead, their ownership interest in businesses is indirect,
through stock in corporations. Even the ownership of stock may be indirect. It may, for example, be through a
claim on a defined-contribution pension plan that in turn owns stock. But the one-time tax effect will carry
through to these indirect owners. Imposing a value-added tax of the consumption type would be predicted to
cause a fall in the market value of stock commensurate with the extra one-time tax liability. I use the
89
90
Â
Page 100
term commensurate here because one critical aspect of the adjustment to the new system that I have not yet
discussed will have a large effect on the transition incidence. That is the impact of the policy change on the
general level of prices.
Price Level Changes
Most people assume that companies will "pass forward" a newly introduced value-added tax in the form of
higher prices. It is important to recognize that whether or not this occurs (1) is not a matter that is settled by
well-developed theory and (2) does not affect the one-time wealth tax due to introducing the new tax,
although it may affect the way the burden of that tax is distributed. In a competitive economy, a value-added
tax of the consumption type must be extracted from the difference between the value of goods sold by
companies and what they pay to noncompany suppliers, which we may here take to be workers. That much is
clear. Whether, however, a newly introduced tax leads to an increase in the prices of things sold or a decrease
in the wages of workers is not well determined. It will depend on the institutions of wage and price setting
and on monetary policy. It is commonly believed, however, that introducing a value-added tax of the
consumption type will bring with it a monetary policy adjustment that results in a one-time increase in the
price level (not a change in the rate of inflation) and no change in payments to workers in nominal terms (so
that, before taking into account any offsetting reduction in income taxation, nominal wages are unchanged
but real wages of workers decline by the amount of the tax). By contrast, it is generally thought that
introducing a tax levied on the earnings of workers will lead to a decrease in their take-home pay and no
change in the prices charged by companies. The real result for workers is the same.
If there is a change in the price level and if this change in price level has not been anticipated and therefore
built into transactions expressed in dollar terms (for example, through an adjustment in interest rates), then
introducing the value-added tax of the consumption type will bring about a redistribution of wealth from
lenders to borrowers, through a decline of the real value of the dollar.
Leverage and the Wealth Tax Effect
That the matter is of some importance, and that policy makers might want to encourage such an
"accommodating" monetary policy, is
Â
Page 101
suggested by considering the situation of the business owners who have financed the acquisition of business
assets by issuing debt with fixed nominal terms. In my example, suppose the store's owners have borrowed
the $10,000 to buy the inventory of canned goods, expecting to sell the goods the next day for a little more
than $10,000 and repay the loan. (Interest on the loan is not important in this very short-term transaction.) If,
in the meantime, the value-added tax has been introduced and there is no change in the price level, the
owners of the inventory will suffer a loss equal to $2,000the tax rate times $10,000. They will suffer this loss
even though they have no net wealth at all, since their assets in the form of inventory are just balanced by
their liabilities in the form of the loan. The one in this picture who does have some wealth, the lender who
holds the $10,000 note, will experience no loss from the new tax. He or she will still be able to take the
proceeds of the loan repayment and buy the same goods and services as if there were no new tax.
90
91
By contrast, if the price level increases by the amount of the tax, the owners of the inventory will suffer no
loss; they will have $10,000 left after tax with which to repay the loan. Instead, the transition loss will be
borne by the lender, via the erosion of the purchasing power of the $10,000. 20
Translating this story to the stock market, if the transition brings with it an unexpected one-time increase in
the price level, the transition loss, measured in real purchasing power, will be spread evenly across the debt
and equity holders. Equities would keep their market value but lose in real value a proportion equal to the tax
rate, and the same would be true of nominally denominated assets and liabilities. At the other polar case of no
adjustment in the price level (actually, polar case is a somewhat misleading term here; anything could
happen), nominally denominated assets and liabilities would keep their real value and the entire loss, of the
tax rate times the sum of equity and debt value, would be borne by equity holders. As in the case of the store
owner who paid for the inventory with borrowed money, there is a leveraged effect on the equity holders that
could be substantial.21
A Tax on the Elderly?
An important question is who are the owners of assets who would bear the cost of transition to a
consumption-type tax (barring special provisions to mitigate the effect)? As has just been emphasized, the
Â
Page 102
impact of the transition will depend both on the extent to which it is accompanied by price-level changes and
on the composition of people's portfolios (especially the division between nominal and real assets and
liabilities). It is generally assumed that the effect will be roughly proportional to wealth (that is, either wealth
owners have similar portfolios or there is an unanticipated price-level adjustment). Apart from knowing that
the distribution of wealth ownership is highly skewed, with a large fraction of wealth owned by a relatively
small fraction of the population, we also know that, owing to life-cycle factors in the process of
accumulation, wealth is also correlated with age. This means that the policy regarding transition to a
consumption-type tax should be thought about in the framework of intergenerational distribution.
Many commentators (in particular, Kotlikoff 1992 and Auerbach, Gokhale, and Kotlikoff 1993) have noted
the tendency of fiscal policy in the United States over the past thirty years to shift the net burden of financing
the government away from older and toward younger and future generations. To the extent that this tendency
describes fundamental political factors, it would suggest we should expect to see any transition to a
consumption-type tax accompanied by rules that would protect the interests of older generations. To the
extent that there is a movement toward readjusting the fiscal balance (arguably, the proper economic
interpretation of "deficit-reduction"), we might expect to see a transition to a consumption-type tax taken as
an opportunity to lighten the projected burdens of young and future generations.
The Effect on the Analysis of Taxing Wages at the Level of the Worker
If we modify the example to permit companies a deduction for wages, there will be a change in the locus of
tax paymentsthe bulk of tax collections will be from workers, rather than businessesbut, from a formal point
of view, none of the issues discussed above will be affected. I use the word formal advisedly, since it is
commonly believed that workers are likely to resist changes in their before-tax wages. If before-tax wages are
fixed and wages are not allowed as a deduction, then the price level must increase to accommodate the tax.
If wages are allowed as a deduction (and taxed at the worker level), holding constant the level of wages
before tax means a reduction in wages after tax. In that case, there would be no adjust-
91
92

Page 103
ment in the price level necessary to establish equilibrium in the relationship between prices and wages.
Effects from Eliminating an Income Tax
Our idealized tax reform involves introduction of a consumption-type tax and elimination of an income-type
tax, represented in the present analysis by a value-added tax of the income type. Under a value-added tax of
the income type, the yield on business investment is taxed, at the value-added tax rate, at the business level.
In the pure case, there is no tax at the individual level. Interest (along with other forms of reward to owners of
capital) is "pre-taxed" at the company level. 22 In an equilibrium with a value-added tax of the income type,
the interest rate on financial assets would equal the after-tax rate of return on investment. This is in contrast
to the result when interest is deducted by business borrowers and taxed in the hands of the individual. In that
case, the interest rate tends to equality with the before-tax rate of return on business investment.
Under a value-added tax of the consumption type, changes in the tax rate (for example, from zero to a
positive rate at the time of introduction of the tax) produce changes in the value of a company along the lines
just discussed. By contrast, under an income-type tax, with proper accounting rules, changes in the rate of tax
do not produce changes in the market value of assets. (The qualification, "with proper accounting rules," is
important. Technically, what is required is that the basis of assets be equal to their market value. With
accelerated depreciation, for example, basis will be less than market value. In that case, a decline in the
income tax rate produces an increase in a company's value, because of the reduction in the tax liability that
will come due when the difference between the asset's basis and its market value is realized in future
transactions.)
Consequently, the transition issues raised by a switch from a pure value-added tax of the income type to a
pure value-added tax of the consumption type are the same as those involved in introducing value-added tax
from a situation with no tax.
Effect of a Regime Shift on Interest and Assets
Thus far I have focused on transition effects apart from effects on relative prices. Among the most important
of the latter are the
Â
Page 104
interest rate and its correlative, asset prices. Changes of policy of the magnitude under consideration here
might be expected to produce significant changes in interest rates and asset prices. Unfortunately, asset price
effects are hard to predict; yet they may constitute an important part of the transition story. Since my
objective is to lay out issues, not necessarily to resolve them, I sketch out here the implications of what might
be taken to be the polar possibilities.
The effect of the policy shift on asset prices will depend on two major dimensions of the economy's response.
One dimension is, in economists' jargon, the relative demand and supply elasticities of capital. At one
extreme of this spectrum is the "infinite elasticity of demand for capital" assumption that the opportunities to
invest at the going rate of return are unlimited within the relevant range. This might be a good description of
the situation of a small country well connected to world capital markets. The before-tax rate of interest is
unaffected by policy changes in the small country. At the other extreme of this spectrum is the "infinite
elasticity of supply of capital" assumption that the amount of wealth people are willing to hold at the going
92
93
rate of return to savers is unlimited within the relevant range. This assumption can be justified on the basis of
very long lifecycle or dynastic purposes of saving (Summers 1981). Under this assumption, a country's tax
policy has no effect on the after-tax rate of return received by savers, since they will spend down their wealth
if the after-tax rate of return is lower and accumulate if the after-tax rate of return is higher.
The second dimension of response that is important in determining the effect of the reform is the cost of
adjustment. To illustrate, suppose we introduced a tax credit for new investment in equipment. For a
company with a given investment opportunity, this would appear to offer extra profit. But if other companies
can rush to take advantage of the same opportunity, the extra profit will be dissipated in lower prices of the
extra output or higher prices of the equipment in question. Where the balance will be struck will depend on
the extent to which the incumbent company has a cost advantage in undertaking the new investment.
The higher costs of other companies are what is referred to as the cost of adjustment. And the effect of the
rules that make investment more attractive at given interest rates will be influenced by the extent to which
owners of existing capital have a temporary advantage over potential competitors for investment
opportunities when interest
Â
Page 105
rates fall. When the temporary advantage is large, the incumbents, those who already own business assets,
can earn supernormal profits during an adjustment period during which new assets are constructed. At the
other extreme of this dimension of response, if new assets can be put in place instantaneously, with no extra
cost, the price of existing assets will equal their replacement cost, regardless of interest rate changes. So the
new rules will have no effect on the prices of reproducible assets. If, as is certainly the case to some degree,
expanding the capital stock quickly brings with it extra costs, owners of existing assets will reap a capital
gain on a change that leads to a higher level of the capital stock.
The opposite effect applies to owners of assets for which adjustment requires a fall in the stock. This might
be the case for assets that are favored by the existing tax law. Under a true, uniform income tax, all assets are
treated alike. We would then expect a shift to a new regime to call for expansion or contraction of all asset
stocks. But under the actual income tax, some assets are more lightly taxed than others, and a shift to a
uniform consumption-type tax that might call forth an expansion of regular business investment might
involve contraction in the stocks of formerly tax-favored assets. Adjustment cost in this case refers to the
time required to work off an excess in the supply of an asset type, relative to that justified by the costs of
reproduction. The practical case of such an asset of most importance is owner-occupied housing, which I
discuss further below. 23
In view of the influence of such assumptions on the results, the analysis here can indicate only the nature of
transition problems. The magnitudes and even the directions of change cannot be inferred from general
principles.
Interest Rates
If adjustment costs were zero, eliminating a uniform business tax on accretion income would not change
business asset prices, except insofar as they reflected any transition wealth tax. The policy change could,
however, lead to a change in the rate of interest received by savers. At the one pole is an unchanged rate of
return on investment before tax. Since the tax being eliminated was paid at the business level, competition
would drive up the rate of interest by the amount of the former tax. The interest rate net of tax would
therefore rise by the amount of the former income tax rate, providing wealth owners with a higher yield. At
the other pole is an unchanged rate of interest
93
94
Â
Page 106
net of tax, the case of "infinitely elastic" supply of capital. In this case the market rate of interest would be
unchanged, but expansion of the capital stock would result in driving down the rate of return on investment
by the amount of the former tax. Wealth holders would see no change in the yield on their holdings.
If interest were taxed at the level of the recipient, rather than at the level of the paying firm, the outcomes
would be the same, but since the market rate of interest would then be a before-tax rate, the description would
need to be revised accordingly.
Concentrating on the case of no change in the before-tax yield on investment, we see that the higher rate of
return works to compensate wealth holders for the loss imposed on transition to the consumption-type tax.
Whether a particular wealth holder gains or loses depends critically on the planned timing of consumption.
For people who are planning to draw down their wealth in the near future, even a very large increase in the
rate of return will not compensate for the one-time loss. For people who are planning to defer consumption
for a long time, the one-time wealth loss will be more than made up by the increased yield. 24
The analysis of these polar cases establishes several points:
In general, imposition of a consumption-type tax will cause a one-time loss to owners of certain assets. The
loss will be spread over all wealth owners to the extent the transition is accompanied by an unanticipated
increase in the price level.
Substitution of a consumption-type tax for an income-type tax may bring with it a higher after-tax yield to
holders of wealth.
A higher yield is a compensating

factor for a transition wealth loss,
but the offset depends on the
consumption plans of the holders.
Those planning long
postponement will gain. Those
planning near-term consumption
will lose. If there is no increase in
the yield on wealth, there is no
compensating offset to the loss of
wealth on transition.
There is thus a considerable range of outcomes possible, depending in particular on the response of the rate of
return on saving to the change in policy and on the distribution of preferences in the wealth-holding
population.
Empirical research will be required to draw more specific conclusions.
Â
94
95
Page 107
Owner-Occupied Housing
The effect of reform on the value of owner-occupied housing is a matter of great political moment. Just as in
the case of other assets, the effect of a transition to a consumption-based tax on owner-occupied housing has
two elements. First is any tendency for the transition to effect a one-time tax on this form of real asset. The
good news is that real assets in the hands of households, including consumer durables and inventories of
goods, as well as houses for personal use, are not subject to the one-time tax. We may contrast the situation
of a person in the business of building houses for sale, who would confront the wealth tax in transition,
because the sale of a house in inventory would be subject to the consumption tax. That house would compete
in the market with houses already owned by households and would therefore sell for the same price.
The second element is the case of adjustment costs. In connection with the discussion of business capital
above, the case of no adjustment costs was highlighted. In that case, the stock of business capital is assumed
to adjust so that all forms of investment have the same yield at the margin (the going interest rate). The
corresponding assumption for owner-occupied housing would lead to the same result: the value of the stock
of housing would not be affected by any change in the interest rate brought about by the shift in tax regime. 25
To illustrate the effect of positive adjustment costs, suppose the stock of housing in the income tax era is
adjusted to yield the after-tax risk-free interest rate of 1.5 percent when the market risk-free interest rate,
before tax, is 2 percent. Owner-occupied housing will attract investment until its yield, which is exempt from
tax, is 1.5 percent. The regime shift produces a decline in the market interest rate to, say, 1.75 percent as a
result of an expansion of saving and capital accumulation. To get to a yield of 1.75 percent on the margin in
housing investment requires a downward adjustment in the stock of housing. We might expect such a
downward adjustment to take time, which is to say to involve adjustment costs. In this case, the regime shift
would have a depressing effect on housing prices and result in losses to owners of housing capital.
Interest at the Individual Level
A simplified version of a shift to a consumption-type tax from something like the present tax system would
involve a change from
Â
Page 108
an income-type tax with a deduction at the level of the firm for interest payments and wage payments (and
taxation, at the same rate, at the level of wage and interest recipients) to a consumption-type tax at the
business level with no deduction for interest (or taxation of interest received) and possibly no deduction for
wages (and in that case, no taxation of wages received). I have argued that the equilibrium results obtained
should not depend on the level (payer or recipient) at which interest or wages are taxed. The alternative tax
treatments will be reflected in compensating differences in the terms of the transactions between payers and
recipients. There is, however, a potential transition problem posed by contracts entered into in one tax regime
that are to be carried out in the other.
Starting with a system that taxes interest and wage payments at the level of the individual implies the
existence of commitments, fixed for some period of time, that will incorporate the expectations of the parties
about the tax treatment of the payments. Consider first the case in which the shift to a consumption-type tax
is not accompanied by a change in the general price level. Then, during the period covered by the contractual
commitment, the interest recipient or wage recipient under a contract made before the change will gain, and
the payer will lose. The gains and losses will be simply the amount of the tax.
95
96
Alternatively, if there is a one-time price level change in connection with introduction of the
consumption-type tax, the purchasing power of the recipient who no longer pays tax will be the same as
before the change. The transition problem is thus linked to the price level determination.
It may be reasonable to single out interest commitments in this regard, as being fixed for a longer term than is
likely in other contracts and, perhaps, as being less amenable to renegotiation. (In a sense, most employment
contracts are in a continual process of renegotiation.) But the difference is one of degree.
4
Moderating Transition Effects
At the business level, two points at which special transition rules would seem likely are basis in assets and
liabilities and commitments to interest with a presumption of deductibility established before the shift in
regimes. The more generous the allowances, the less the potential loss from transition, at a cost in revenue
that must be made up in a higher tax rate in the new system.
Â
Page 109
Basis in Business Assets
For the simple case of substitution, overnight, of a uniform subtraction-method value-added tax of the
consumption type for a uniform income tax, a ''cold turkey" transition would effect a one-time tax on the
holders of real assets, perhaps reallocated to the holders of wealth more generally, through price-level effects.
26 To eliminate any one-time tax on existing real assets would require allowing immediate deduction of the
company's basis. As for past investment, immediate deduction of the company's basis would put previous
investment on the income-tax system up to the time of the switch and on the consumption-tax system from
then on. We can describe allowing such a deduction as a policy of fully protecting "old capital."27
It is often said that protecting old capital in this way would unacceptably increase the government's budget
deficit. Dealing with the deficit consequences is, however, a matter of structuring the rules to achieve the
desired cash flow. Thus, an economically equivalent policy to permitting immediate write-off of existing
basis would be to permit write-off over a period of time, as long as desired, provided there were an allowance
of interest earnings on as yet undeducted basis. The effect would be to provide taxpayers with the same
discounted value of tax savings as immediate deduction, but the cash flow to the government would be very
different.
The revenue effect of taxing or not taxing existing stocks of assets is, however, not a matter of detail. In their
modeling of such transitions, for example, Auerbach and Kotlikoff (1987) conclude that a switch from an
income tax to a consumption tax, while fully protecting old capital (a policy they refer to as a wage tax)
might well generate an effective reduction in national welfare, whereas a switch that provided no protection
to old savings would generate large gains in every case. These consequences stem from the lump-sum
character of the one-time tax on wealth (in their analysis, they treat the change as unanticipated), which
allows a lower tax rate and thereby enhances future efficiency.
Such particular conclusions are a function of the specifics of the modeling of the economy and the policies.
(For example, Auerbach and Kotlikoff do not incorporate the smoothing of revenues that might be
accomplished by delaying the recovery of basis along the lines discussed above.) The generic point is,
however, unavoidable,
Â
96
97
Page 110
and it is a centrally important element of policy choice. A one-time tax on old savings can generate a great
deal of revenue, and if it is really unanticipated, it can do so with no efficiency cost at all, just the
distributional consequences discussed above. I return below to both the efficiency and the distributional
questions.
As an alternative to fully protecting old capital, permitting businesses to continue to take as deductions
against the value-added tax of the consumption type the depreciation allowances to which they would have
been entitled under the income tax might seem a reasonable policy. It would appear to fulfill the expectations
of businesses that had made investments with certain assumptions about the tax treatment of the transactions.
As the discussion thus far should make clear, however, this approach is not the same as fully protecting
existing assets from the one-time tax that would result from a simple switch to consumption-tax rules. The
pattern of incidence is also somewhat curious: owners of assets with short remaining lives would be closest to
fully protected, while those owning assets with long remaining depreciable lives, such as recently constructed
buildings, would come closest to incurring the full one-time tax. It is not clear such a pattern of incidence has
anything to offer in terms of either efficiency or equity.
Preexisting Commitments to Pay Interest
As I have suggested, the commitment to pay interest would seem particularly prominent among the various
preexisting commitments that might attract special transitional treatment. I have also suggested that
adjustment for preexisting commitments will occur automatically if there is a one-time price level change. In
the alternative case of no change in the price level, it would appear feasible in the framework of uniform
systems to adopt a grandfathering approach. With regard to those commitments that predate the transition to
the new system, retain the old tax rules. The deduction by one taxpayer and the inclusion by another taxpayer
at the same rate of tax have no allocational significance, but if the applicable tax rate is the same in the
income and the consumption-tax systems, both borrowers and lenders will be unaffected by the change in
rules. This observation applies to deduction by mortgage borrowers of interest on preexisting contracts, just
as it does to businesses.
Â
Page 111
The neat symmetry and simplicity of this transition rule would be considerably complicated by the presence
of multiple rates of tax (for example, both taxable and tax-exempt entities) in the preexisting system or in the
new system. In some way, preserving preexisting expectations with regard to interest payments and receipts
may, however, provide a way to moderate a transition effect that has no obvious policy merit. (That effect is
imposing an unexpected burden on borrowers and providing an unexpected windfall on lenders by virtue of
the change in treatment of payments labeled interest.)
The issue of the treatment of prechange commitments to pay and receive interest needs to be distinguished
from the issue of the effect of the change in the rate of interest as a relative price. If the after-tax interest rate
increases as a consequence of the regime shift, net creditors will gain and net debtors will lose. This
phenomenon is taken up in the next section.
5
Should Old Capital be Taxed?
97
98
Thus far I have concentrated on the burden-shifting effects of a switch between a uniform income tax and a
uniform consumption tax. There are two important incentive-related aspects of the transition.
Effects of Transition on Saving and Investing
First, as has been discussed, the one-time wealth tax potentially provides the revenue to permit a relatively
low tax in the new system. Economists describe such a tax as "lump-sum" because there is no behavioral
change by which a person can avoid its impact. (The classic textbook lump-sum tax is a poll tax.) The
advantage of a lump-sum tax is that, since it cannot be avoided through behavioral changes, it does not distort
economic choices. If it is true that the transition is in the nature of a lump-sum tax, it would provide revenue
to reduce taxes that do distort behavior, such as taxes on earnings from work or saving. As demonstrated by
Auerbach and Kotlikoff (1987 and especially 1983), the gains from lower tax rates could actually permit
everyone, including those on whom the onetime wealth tax is imposed, to gain from the transition. 28
One can, however, make too much of the supposed lump-sum character of a cold-turkey transition. To be a
lump-sum tax, the tax
Â
Page 112
on wealth must be unanticipated. Otherwise people will take steps to avoid the tax by consuming more
rapidly, saving and investing less. This is the second important efficiency aspect of transition. An anticipated
introduction of a consumption tax, or an anticipated increase in its rate, for which no compensating transition
rule is provided, will discourage saving and investing and encourage current consumption. Assets held at the
time of an increase in rate suffer a one-time percentage tax equal to the change in the consumption-tax rate.
There is an incentive to convert assets to consumption before the rate change. An investment project will
result in a deduction today at the lower rate, but its future payoff will be taxed at the higher rate. There is an
incentive to postpone investment projects. It is easy to see that these incentive effects could be very strong.
Concern about incentive effects of this sort leads policy makers to establish an "effective date" on the day of
first serious consideration of programs, such as an investment tax credit, designed to encourage investment. It
is recognized that an anticipation that investment will be more favorably treated in the future than in the
present will lead to a postponement, typically the very opposite of the desired effect.
There is, finally, the reputational effect of imposing a one-time tax on transition. If as a result, taxpayers are
wary of future such "onetime" changes, their incentives for forward looking investment are no longer
captured by whatever purport to be the current rules. Instead, investors will behave as though there were a
tax, since they are at risk of a change in the rules and lack any assurance that transition impacts of such a
change will be ameliorated through appropriate rules. This is the problem of time consistency.
Political Economy Arguments
There is a certain contradiction in the idea that imposition of a consumption-type tax is unexpected and
therefore is effectively a lump-sum tax, with no disincentive consequences, whereas an increase in the rate of
a consumption tax may be anticipated, discouraging investment. The logic is, "We'll only do this just this
once." The potential that a government may introduce a consumption-type tax without compensating
transition rules presumably has disincentive effects as soon as it is felt to be operative (as perhaps in the
present time). In any case, once a consumption-type tax is in place, the consequences of anticipated changes
need to be taken into account.
Â
98
99
Page 113
There are basically two directions to take toward dealing with this problem on a more systematic basis. The
first is to design into the systems mechanisms that limit the variability in tax rates. An example would be
enacting special majority rules for tax rate increases. Another example is the self-averaging feature of the
cash flow tax described in the U.S. Treasury's Blueprints for Basic Tax Reform (U.S. Treasury 1977; also
Bradford et al. 1984). In briefest sketch, the Blueprints cash-flow tax would operate wholly at the level of the
individual (rather than the firm), producing a consumption-type base by allowing deduction of net deposits to
"qualified accounts" and otherwise leaving interest and similar flows out of the tax base. The resulting base
would be taxed at graduated rates. In such a system it would be in the individual taxpayer's interest to
"self-average" to maintain a constant tax rate over time. 29 Self-averaging is generally thought of as a way to
deal with variation in the individual's tax base, due to life cycle or other sources of changing economic
circumstances. But self-averaging could also deal with anticipated changes in legislation, leading taxpayers to
engineer a current tax increase for themselves in anticipation of an upward shift in the tax rate schedule to be
brought about through legislation or a tax decrease in anticipation of a downward shift in the tax rate
schedule. In the process, adverse incentive effects on investment would be eliminated.
The second approach is to establish a principle or mechanism to ensure that changes in rates will be
accompanied by measures to reduce or eliminate the effective wealth levies, for example, by grandfathering
provisions. (Another example of such a device is the common practice in the legislative process, when
investment incentives are likely to be affected by rule changes, of announcing an effective date after which
transactions will come under the new rules if enacted.) It seems easier to imagine grandfathering rules to
protect taxpayers from having their wealth taken by a rate increase than ones that would extract the wealth
gain from a tax rate cut. A practical instance of such a gain was effected by the Tax Reform Act of 1986,
which resulted in substantial reductions in the taxes on pension benefits that had been deducted earlier at
higher marginal rates.
There is thus an interaction between a general practice of protecting investors from the sort of wealth tax that
we have been discussing and the degree to which tax rate changes need to be inhibited.
Â
Page 114
Equity Arguments
Early and Late Consumers
There are two main strands of equity arguments in connection with the transitions to a consumption-type tax.
30 The first one focuses on the fact that, under a consumption-type tax, owners of wealth will obtain their
normal yield free of further tax. This applies to wealth accumulated after the transition and is an aspect of the
argument in favor of the consumption approach. But for those who accumulated their wealth before the
transition, goes the argument, the new rules may effect an unexpected, if not undeserved, gain.
As discussed above, any one-time tax on wealth in the transition is, to greater or lesser degree, compensated
if the rate of return to owners of wealth increases in the process (a likely, although not necessary outcome).
For a wealth owner who plans to consume immediately after the change in regimes, an increase in the rate of
return is of no value. There is no compensating gain. For one who plans to postpone consumption for a long
time (for example, by passing wealth along to heirs), the gain in rate of return may more than compensate for
a one-time tax.
99
100
If it were desired as a matter of policy to achieve rough neutrality in the transitionthat is, generate neither
gainers nor losersit would be necessary to develop a way of discriminating among wealth owners according
to their likely consumption horizons. If the distinction is based on behavior (that is, on when people actually
consume), giving greater effective protection from the wealth tax on transition to those who consume
(dissave) earlier, a price will be paid in the form of both equity and efficiency consequences rather at
variance with the philosophical underpinnings of consumption taxation. Any attempt to discriminate with
precision among people according to the timing of their planned consumption is likely to introduce such
incentives, precisely because of the need to refer to people's behavior to determine their preferences.31
Consistent Application of Consumption-Tax Philosophy
Many people advocate a shift away from income and toward consumption taxation on grounds of efficiency.
In particular, they seek the neutrality of a consumption tax (at least a uniform consumption tax) with respect
to the decision to save. I have elsewhere (1986) suggested that a principal argument in favor of a
consumption
Â
Page 115
approach is, rather, one of equity. In brief, if two people are otherwise similarly situated but differ in their
preferences in the timing of consumption, a consumption tax will impose the same burdens on them, but an
income tax will discriminate in favor of the one who prefers to consume earlier. The same argument that
suggests it would be unfair to discriminate between people according to their preferences for clothes of
different color would imply it would be unfair to discriminate on the basis of differences in preferences for
the timing of consumption.
A sketch of the way this line of argument might be carried over to policy toward transition goes as follows.
Consider the two people who are similarly situated except that one prefers to postpone consumption more
than the other. By the transition date, the late-consumer has a larger stock of savings than the early-consumer
and will have paid more in taxes. After the transition, the two will pay the same amount of tax except for any
wealth tax effect of the transition itself, which will work to the relative disadvantage of the late-consumer.
The argument for neutrality of treatment according to preferences about the timing of consumption would
seem to imply, at least, protecting the wealth in the transition. Viewed from the perspective of the lifetime
treatment at the hands of the tax system, fully protecting wealth in transition leaves the discrimination against
the late-consumer to the extent of the duration of the income-tax regime.
6
Concluding Remarks
There are potentially large gains to be had from pursuing what I call a consumption "strategy" of taxation.
The consumption approach readily solves a variety of pesky problems in the income tax, permitting much
simpler and more transparent rules in the process. I hope this study will assist in clarifying what is involved.
In particular, I hope I have succeeded in showing there is less at the conceptual level to the difference from
the income approach than seems often to be claimed by both advocates and opponents of the consumption
approach. The difference boils down to the tax treatment of the risk-free rate of return to saving, a rate that
has historically been below 1 percent per year. The rest of what is generally thought of as "capital income" is
taxed alike in conceptually pure income and consumption taxes.
Â
100
101
Page 116
The existing income tax is far from "conceptually pure," however. For example, the purely inflationary
element in capital gains is subjected to tax, as is the inflation premium in ordinary interest received. By the
same token, the deduction for interest paid is too large by the amount of the inflation premium. There are, in
addition, dozens of peculiarities in the taxation of saving and investment. The controversy over capital gains,
the complex regulations governing tax-advantaged retirement savings, the double-taxation of dividends, and
many other elements of the tax picture come to mind. Thus those who would defend the status quo might best
divorce their arguments from the theoretical concept of income.
At the same time, the transition to the consistent consumption-type system that is within practical reach does
have much in common with a shift between idealized income and consumption taxes. As described above,
there is much merit in the view that the transition poses serious challenges. There seem to be two main
attitudes toward those challenges. One approach is to minimize them, in the interest of moving ahead to
achieve the reformers' objective. The other is to become intimidated with the problems of transition, so that
they form a roadblock to change. Major tax changes have taken place (good examples in the United States
include the tax reforms in 1981 and 1986) that have presumably had significant transition effects but have
somehow been carried out anyway. I hope the analysis presented here will assist policy makers both to put in
perspective problems of transition to consumption-based taxes and to address those that are important.
Notes
1. For useful discussions of current proposals, see U.S. Congress, Joint Committee on Taxation (1995) and
Arthur Andersen (1995). A detailed description of the USA Tax, prepared by Alliance USA, was published
as a special supplement by Tax Notes, March 10, 1995. The actual legislative proposal was reproduced as a
special supplement by the Bureau of National Affairs, April 26, 1995.
2. In the helpful terminology of the Meade Committee Report (Institute for Fiscal Studies, 1978), this is an
R-base ("real" transactions, as opposed to "financial") tax. There are some interesting questions about how to
distinguish a financial from a real transaction. For example, is the purchase of a piece of paper giving rights
to use a trademark a financial or a real transaction? Financial institutions also present special problems that I
have treated elsewhere (1996).
3. Throughout, the assumption is made that deductions (and credits) can be used. This would apply if losses
(and net credits) were carried forward with interest, for example, or were refundable.
Â
Page 117
4. Economic differences in practice may arise because of differences in the definition of sales subject to
taxfor instance, sales of clothing or medical services might be treated as taxable in one system and not in
another.
5. Amounts put aside for worker retirement would also be deducted, that is, treated essentially the same as
under the present income tax.
6. Particularly at higher rates of

tax, it can be important to keep
track of whether the tax itself is
included in the calculation. So,
for example, a rate of tax of 20
percent on earnings of $50,000
101
102
would produce a tax of

$10,000. That amount would be
25 percent of the $40,000
consumed out of the salary by a
worker who added nothing to
savings during the year.
7. It is sometimes argued that the invoice-and-credit method is more easily enforced because the buyer can be
required to show the tax paid by the seller; however, it would seem equally easy to trace the purchases
deducted by the buyer to the tax returns of the sellers in a subtraction-method system. In contrast with the
case of an income tax, there should be an inclusion on a selling business's tax return for every deduction by
the buyer.
8. The general public's claim is on a share of the assets of the company, which it has helped finance through
the tax deduction. The general public does not share in the reallocation of the claim through debt finance.
9. Under the U.S. income tax, nominal interest payments are included in the base of the recipient and
deducted by the payer Nominal interest rates include compensation for anticipated inflation and, thus,
generally exceed real interest rates. For a discussion of the consequences, taking into account the presence of
multiple marginal tax rates, see Bradford (1981). One of the important effects of a shift to a
consumption-type tax would be automatic indexing for inflation.
10. For details on historical rates of return, see R. G. Ibbotsen Associates (1995).
11. For a related discussion, see Gordon (1985).
12. This statement applies to a tax based on accruing income, so that losses and gains are reflected
immediately in the tax due. The way the deductibility of losses balances the taxation of gains was emphasized
in a famous treatment by Domar and Musgrave (1944) of the effect of taxation on risk taking. Apart from
problems relating to inflation, U.S. business income taxation is reasonably close to an accrual basis, although
there are many exceptions. In the individual income tax, however, a "realization-based" approach complicates
the story considerably. For a discussion of the complex effects of current realization-based income tax rules
on incentives for risk taking, see Scar-borough (1993).
13. For elaboration of this point, see Kaplow (1994).
14. It is an open question whether an actual consumption-type system would treat housing consistently with
other forms of consumption.
15. Alan Tait (1991, 11-12) reports that some countries allow less than full write-off of certain capital
acquisition expenses.
16. Strictly speaking, the flow of revenue to the government from a true income tax should not depend on the
timing of transactions. The income calculation would be based on accruing gains and losses.
Â
Page 118
17. Even a uniform tax on a real (inflation-corrected) market-value basis has substantial implementation
problems. For a discussion, see Bradford (1986). For a discussion of similar issues that arise in national
income accounting, see Bradford (1991).
18. The distinction between "direct" taxes (such as income, including corporate profits, wealth, and property
taxes) and "indirect" taxes, such as sales and value-added taxes, is a matter of custom and not based on any
102
103
fundamental economic difference.

19. The implied measure of income of the creditor fails to satisfy the strict concept of economic incomeit
does not, for example, require distinguishing between interest payments and accured interest income. It is a
simple reassignment of responsibility for tax payments from the company to bondholders. For an extended
discussion of the concept of economic income and why measures based on cash payments, such as interest,
break down, see Bradford (1986).
20. There would be, in addition, the usual distributional effects of inflation, on fixed-income recipients, for
example, or on holders of currency. See Browning (1978).
21. For a company for which equity constitutes a fraction e of the value of the company's assets, the implied
percentage decline in value of the shares would be t/e, where t is the value-added tax rate. For a sufficiently
leveraged firm (e < t), the transition incidence would imply bankruptcy (Howitt and Sinn 1989). (Note that if
everyone holds the market portfolio of all financial instruments, the effect of price level change is
neutralized.)
22. This is the scheme outlined in the U.S. Treasury's Comprehensive Business Income Tax plan 1992.
23. Auerbach (1989) studies the dependence of the effects of investment incentives on the cost of adjustment.
See also the discussion and references in Sarkar and Zodrow (1993).
24. For further discussion of the trade-off, see Bradford et al. (1984, 180-84).
25. The same reasoning extends to the land on which owner-occupied housing sits, provided there is no tax
levied on its sale. Note that I have not tried in this analysis to deal with the elimination of the property tax
deduction, which is not related to the choice between income- and consumption-tax regimes.
26. Hall and Rabushka (1995) advocate this form of transition to the flat tax.
27. Although I do not develop the point here, it might be noted that a similar line of argument leads to the
conclusion that to protect old savings from the one-time tax on transition to a consumed-income tax at the
personal level would involve allowing immediate recovery of basis.
28. To establish whether these conditions are fulfilled in fact, and whether a particular transition plan would
effect such a general gain, would require more detailed empirical analysis than has been carried out to date.
29. Taking into account risk leads to an interesting modification of this story.
30. I have particularly benefited from discussions with Louis Kaplow on the subject of this section. Some of
his ideas about transitions in general are set out in Kaplow (1986) and about transition to a consumption-type
tax in Kaplow (1995).
Â
Page 119
31. This point is emphasized by Kaplow (1995) in his analysis of the USA tax. The USA tax proposal
incorporates a system for individuals to recover the basis in their wealth at the time of transition, but the
recovery is postponed until such time as the taxpayer becomes a net dissaver.
References
Alliance USA. ''USA Tax System." Tax Notes, Special supplement 66, no. 11, March 10, 1995.
103
104
Arthur Andersen. Office of Federal Tax Services, "Tax Reform 1995 Looking at Two Options." Arthur
Andersen & Co., SC, May 1995.
Auerbach, Alan J. "Tax Reform and Adjustment Costs: The Impact on Investment and Market Value."
International Economic Review (1989): 939-62.
Auerbach, Alan J., Jagadeesh Gokhale, and Laurence J. Kothkoff. "Generational Accounting: A Meaningful
Way to Evaluate Fiscal Policy." Journal of Economic Perspectives 7 (1993).
Auerbach, Alan J., and Laurence J. Kotlikoff. "National Savings, Economic Welfare, and the Structure of
Taxation." In Behavioral Simulation Methods in Tax Policy Analysis, edited by Martin Feldstein. Chicago:
University of Chicago Press, 1983.
. Dynamic Fiscal Policy. New York: Cambridge University Press, 1987.
Bradford, David F. "Issues in the Design of Savings and Investment Incentives." In Depreciation, Inflation
and the Taxation of Income from Capital, edited by Charles R. Hulten, 13-47. Washington, D.C.: Urban
Institute, 1981.
. Untangling the Income Tax. Cambridge, Mass.: Harvard University Press, 1986.
. "On the Incidence of

Consumption Taxes." In The
Consumption Tax: A Better
Alternative, edited by Charls E.
Walker and Mark A. Bloomfield,
243-61, Cambridge, Mass.:
Ballinger, 1987.
. "What Are Consumption Taxes and Who Bears Them?" Tax Notes, April 18, 1988.
. "Market Value versus Financial Accounting Measures of National Saving." In National Saving and
Economic Performance, edited by B. Douglas Bernheim and John B. Shoven, 15-44. Chicago: University of
Chicago Press, 1991.
. "Treatment of Financial Services trader Income and Consumption Taxes." In Economic Effects of
Fundamental Tax Reform. Edited by Henry Aaron and William Gale. Washington, D.C.: Brookings
Institution, 1996.
Bradford, David F., and the U.S. Treasury Tax Policy Staff. Blueprints for Basic Tax Reform, 2d ed., rev.
Arlington, Va.: Tax Analysts, 1984.
Browning, Edgar K. "The

Burden of Taxation." Journal
of Political Economy 86
(August 1978): 649-71.
Bureau of National Affairs. "USA Tax Act of 1995." Special supplement, report no. 80, April 26, 1995.
Â
104
105
Page 120
Domar, Evsey D., and Richard Musgrave. "Proportional Income Taxation and Risk Bearing." Quarterly
Journal of Economics 58 (1944): 382-482.
Gordon, Roger H. "Taxation of

Corporate Capital Income: Tax
Revenues versus Tax
Distortions." Quarterly Journal
of Economics 100 (February
1985): 1-27.
Hall, Robert E., and Alvin Rabushka Low Tax, Simple Tax, Flat Tax. New York: McGraw-Hill, 1983.
. The Flat Tax. 2d ed. Stanford, Calif.: Hoover Institution Press, 1995.
Howitt, Peter, and Hans-Werner Sinn. "Gradual Reforms of Capital Income Taxation." American Economic
Review 79 (1989): 106-24.
Institute for Fiscal Studies. The

Structure and Reform of Direct
Taxation: The Report of a
Committee Chaired by
Professor J. E. Meade. London
George Allen & Unwin, 1978.
Jorgenson, Dale W. "The Economic Impact of Fundamental Tax Reform." Paper presented at the Hoover
Institution Conference on Frontiers of Tax Reform, Washington, D.C., May 11, 1995.
Kaplow, Louis. "An Economic Analysis of Legal Transitions." Harvard Law Review (January 1986).
. "Taxation and Risk Taking: A General Equilibrium Perspective." National Tax Journal 47, no. 4 (December
1994): 789-98.
"Recovery of Pre-Transition Basis under an Individual Consumption Tax: The USA Tax System." Tax Notes,
August 28, 1995.
Keuschnigg, Christian. "The Transition to a Cash Flow Income Tax." Swiss Journal of Economics and
Statistics 127, no. 2 (1991): 113-40.
Kotlikoff, Laurence J. Generational Accounting Knowing Who Pays, and When, for What We Spend. New
York: Free Press, 1992.
R. G. Ibbotsen Associates. Stocks, Bonds, Bills and Inflation: 1995 Yearbook. Chicago, 1995.
Sarkar, Shounak, and George R. Zodrow. "Transitional Issues in Moving to a Direct Consumption Tax."
National Tax Journal, 46, no. 3 (September 1993): 359-76.
Scarborough, Robert H. "Risk, Diversification and the Design of Loss Limitations under a Realization-based
Income Tax." Tax Law Review 48, no. 3 (1993): 677-717.
Summers, Lawrence H. "Capital Taxation and Accumulation in a Life Cycle Growth Model." American
Economic Review 71 (September 1981): 533-44.
Tait, Alan A., ed. Value-added Tax: Administrative and Policy Issues. Washington, D.C.: International
Monetary Fund, October 1991.
U.S. Congress, Joint Committee
105
106
on Taxation. Discussion of
Issues Relating to "Flat" Tax
Rate Proposals (JCS-7-95).
Washington, D.C.: U.S.
Government Printing Office,
1995.
U.S. Treasury Department. Blueprints for Basic Tax Reform. Washington, D.C.: U.S. Government Printing
Office, January 1977.
. Report on Integration of Individual and Corporate Tax Systems: Taxing Business Income Once.
Washington, D.C Â· U.S. Government Printing Office, January 1992.
Â
Page 121
II
INCOME AND CONSUMPTION TAXATION: THEORETICAL TOOLS AND
ISSUES
Â
Page 123
5
Tax Neutrality and the Investment Tax Credit
This chapter concerns the question of how the rules for calculating the investment tax credit and the
associated rules for calculating depreciation allowances for tax purposes should be structured to assure the
"appropriate" relationship between the subsidy granted to long-lived assets and that to short-lived assets.
Under the U.S. federal income tax a credit is allowed against tax liability equal to a certain fraction of the
cost of qualifying investments. The fraction depends on the durability of the asset, as measured by its "useful
life." For assets with useful lives of less than three years no credit is allowed; assets with lives of between
three and five years qualify for a credit of 3 1/3 percent; those with lives of between five and seven years, 6
2/3 percent; and those with lives of seven or more years, 10 percent. 1 This credit against tax is ignored in the
calculation of tax allowances for depreciation, which are based on the historical cost of the asset.2
Both of these features of the tax rules have a bearing on the choice by an investor between long- and
short-lived assets. The increasing rate of tax subsidy under the investment credit favors long-lived assets by
comparison with a flat-rate credit, while the neglect of the credit in calculating depreciation allowances
favors short-lived assets (for which the depreciation allowance is a more important element in the cash flow).
There is considerable confusion about which of these two biases in the rules is the right one.3 In reviewing
the literature on this issue Sunley focused on the question of whether the investment credit should vary with
the durability of the asset purchased.4 He concluded that neutrality requires a subsidy rate increasing with the
106
107
useful life of the asset in a way qualitatively similar to that prescribed in present U.S. law.
Â
Page 124
This chapter develops Sunley's discussion through the use of simple formal models of the yield from
investment. A principal conclusion is a qualified confirmation of Sunley's view that neutrality requires the
rate of credit to increase with asset life. Furthermore, if tax rules are otherwise fundamentally correct, the cost
basis for depreciation allowances should be net of investment tax credit. 5 Even with this condition, however,
it is not possible in general to present a neutral rule for calculating the investment credit independently of the
rate of interest and of the detailed pattern of returns from the asset.
The essential principle underlying the conclusions of this chapter is that efficiency requires equality of the
before-tax and before-subsidy rates of return on investments in assets of various durabilities. To obtain an
efficient allocation, subsidy and tax rules must be appropriately related to the durability of the assets.
Although it may be possible to design such a relationship in the case of the investment credit applied to assets
within some limited class of durability characteristics (for example, within the class of assets that lose a
constant fraction of their value every year), no simple rule will give appropriate results for all classes. In
particular, the cases examined illustrate the way simple rules relating investment credit to the single
dimension, "asset life," go wrong as the actual pattern of returns varies from exponential decay to point
output, to constant flow output.6
The analysis takes up in turn three special models of investment opportunities: point input, exponentially
declining output; point input, point output; and point input, constant flow output. The first is perhaps the most
familiar to economists. Exponential decay is much the most convenient depreciation assumption to make in
models of economic growth. It has also been common in the tax lib erature.7 The very analytical convenience
that makes exponential depreciation attractive, however, may lead to neglect of the question of how
conclusions are affected by it. Hence a section of this chapter is devoted to the alternative investment models
mentioned.
For the most part I assume that taxpayers use true "economic" depreciation in calculating income subject to
tax. This assumption is important because it assures that if the investment subsidy can be made "neutral," the
income tax system, whether proportional or progressive, will influence investment only via its effect on
savings and not via the composition of the capital stock.8 It is necessary in
Â
Page 125
this context to be clear about the

definition of economic
depreciation, an issue that calls
for a digression on "recapture"
and the treatment of secondhand
assets.
Taxes and Investment Credit in an Exponential World
The model of investment made familiar by Jorgenson is most naturally interpreted as a world in which the
107
108
quantity of capital consists of a number of identical physical units (for example, shovels or machines). 9 Each
machine produces a fixed output per period. As a result of physical deterioration, a fixed proportion of the
stock of machines disappears each period.
In this world the quantity of capital of any particular type is readily understood. Whereas in a more general
case it may not be obvious what quantity of capital I own when I have a one-year-old machine, it is clear how
many units I own if they consist of a fraction of the previous year's stock of identical machines.
The "rental cost of capital," c, also has a ready interpretation in this model: it is the price paid for the use of
one machine for one period. In a competitive system this price will clear the market for machine services,
equating the quantity supplied by capitalists to the quantity demanded by firms for use in production.
Particular interest attaches to the relationship in equilibrium between the rental cost of capital, the market rate
of interest, i, and the price, q, of a newly produced machine. Under may assumptions, in the absence of taxes
the capitalist receives a cash flow from a machine, c Â· exp[-δs], where δ is the fractional rate at which
machines disappear and s is the age of the machine. The capitalist able to borrow and lend at the market rate
of interest, i (I neglect uncertainty throughout), will be indifferent about purchasing another machine when
or
.
If an income tax is imposed at proportional rate, u, the capitalist will evaluate net-of-tax cash flows at his
after-tax rate of return from
Â
Page 126
lending, (1 - u)i. 10 The cash flow from purchasing a machine will also be affected by (1) any investment tax
credit assumed equal to k per unit invested, and (2) depreciation allowances assumed granted at rate D(s) Â·
q, where s is asset age.11 Equilibrium with wealth maximization now implies12
or
where
is the present value to the taxpayer of the tax-allowed depreciation deductions on one unit of tax basis if ra is
the taxpayer's after-tax return. For easier comparison with the equilibrium condition without taxes this can
alternatively be written as
108
109
When the depreciation schedule, D(s), is applied to the effective purchase price, (1 - k)q, rather than the
nominal historical price, q, the equilibrium relationship between c and q (expression 1) is replaced by
.
True Depreciation: A Conceptual Problem
What is the rate of economic or true depreciation experienced by investors under either of these systems?
There is a certain ambiguity about this question arising out of the subsidy, k, and the possible tax advantage
or disadvantage due to the difference between the depreciation allowed in calculating the tax and that actually
occurring.
As far as the replacement cost of the asset is concerned, the matter is clear. Under the exponential decay
assumption the pattern of physical depreciation is described by
Â
Page 127
Because in this case the actual decay takes the form of a reduction in the number of identical physical
machines, the decline in replacement cost value (the price paid by the buyer before receiving any credit) must
follow this pattern as well. The replacement cost of a machine of age s and of type δ must be given by q
exp(-δs), and the instantaneous loss at that age is given by D* (s) Â· q.
This would describe the path of the demand price of the asset if a purchaser of the used asset could obtain the
same tax and subsidy treatment as if he had purchased the same number of identical but newly produced
machines. The value to the original purchaser of the asset will generally be less than this amount because tax
and credit advantages will have already been realized. Special rules are required to prevent, for example, the
buyer of a new machine from obtaining the investment credit and then immediately reselling his asset to
another purchaser, who in turn receives the same credit. The "recapture" rules of the U.S. tax system,
applicable to the investment credit and certain forms of accelerated depreciation, as well as special rules for
the treatment of used assets, are designed to moderate such tax-motivated transactions. 14 To avoid
complexity, I assume that these rules are perfect so that the tax system has no influence on the desired
time-since-acquisition structure of a capitalist's stock of machines.15
Economic Depreciation to the Holder
What is the path of asset value to the holder of one of the machines who plans to keep it perpetually? I refer
to the rate of decrease of asset value along this path as "economic depreciation." It will in general depend on
the tax rules, including the depreciation allowances. Hence to use economic depreciation for tax purposes
requires finding a fixed point: economic depreciation is a function of tax depreciation; a rule is sought that
will make tax depreciation equal economic depreciation.
Let v(s) denote the value of a machine as a function of the time, s, since original construction and purchase,
and let H(s) represent the schedule of depreciation allowancesas distinguished from the schedule of
depreciation rates, D(s), against basisfor tax purposes. Then the cash flow from the asset at t is
109
110

Page 128
so that
The expression for depreciation is found by differentiating expression 4 to obtain
If the value of depreciation in this sense is used for tax purposes (call this schedule H*(s)), then using -v'(s) =
H*(s):
.
Differentiating,
.
This equation has a solution:
Note that this user depreciation schedule is not related to the purchase price of the asset. Such a relationship
will be a derived consequence of market equilibrium. The sum of all depreciation allowances according to
this formula is given by
The right-hand side is the

purchase price, q if q(i + δ) = c,
the equilibrium relationship in
the absence of taxes. And in this
case the economic depreciation
to the holder is equal to
replacement cost
Â
110
111
Page 129
depreciation, given by expression 2. If this relationship does not hold, the depreciation allowances will sum
to a total different from the purchase price.
The Problem of the Investment Credit
This reasoning and the calculations that went before it are based implicitly on the mathematical assumption
that v(s) is a differentiable function. This assumption seems safe enough for all the features of the tax system
except the investment tax credit. Thus expression 3 correctly describes the cash flow except at the moment of
purchase, when there is an additional lump sum accumulation (an infinitely high cash flow for an instant)
equal to the rebated investment tax credit. Similarly expression 4 is correct for s greater than zero, but v(0)
does not correctly describe the value of a new asset to the holder. That value depends on the investment
subsidy and its tax treatment.
One may think of the purchaser of an asset at price q as receiving with the asset a coupon good for kq in cash.
Neglecting the question of whether there is enough tax liability to collect kq (because under present U.S. law
the credit is only usable to pay taxes), there are two ways of viewing this coupon. First, it may be ignored for
tax purposes, in which case presumably the decline in asset value that takes place when the owner tears off
the coupon and cashes it will also be disallowed as a deduction, even though it reflects a real fall in asset
value to the holder. 16 Second, the coupon value may be taken into taxable receipts, in which case the
associated decline in asset value will also be recognized for tax purposes.
Both approaches have the same effect on the net-of-tax cash flow of the purchaser of the asset. Therefore let
V denote the value to the holder when the true depreciation to the holder is used for tax purposes, and when
either of the treatments above is applied to the tax credit. Then
,
and, using
expression 6,
equilibrium
implies
Â
Page 130
Hence from expression 5, in equilibrium
.
Several conclusions may be drawn from this exploration of economic depreciation to the holder, none
perhaps wholly unexpected. First, in equilibrium the ''basis" for tax depreciation purposes, v(0), is the net of
credit price to the buyer, (1 - k)q, either because the subsidy is not taken into taxable receipts and not allowed
as an immediate write-off or because the subsidy is taken into taxable receipts and immediately deducted.
Either approach accurately reflects the path of asset value to the owner and leads to a sum of depreciation
allowances equal to the appropriate cost of the asset. Second, given this use of true depreciation, the tax rate
111
112
has no effect on the equilibrium relationship between the rental cost of capital, c, and the price of a newly
produced machine, q, except as it may influence the rate of interest, i. Third, the investment credit does enter
the equilibrium condition in such a way as to influence the durability of assets emerging, even though the
"correct" basis for depreciation is used. This conclusion brings me back to the principal concern of this paper.
Efficiency and the Durability of Assets
What is the requirement of efficiency in the allocation of investment at a given instant among machines of
different durabilities? For simplicity I concentrate on steady states where the relevant marginal productivities
can be regarded as constant over time. The question can then be rephrased: how should investment be
allocated so as to obtain the highest sustainable flow of net output?
To answer this question, consider the perpetuity obtainable by sacrificing a unit of consumption to purchase
an asset of type δ and subsequently reinvesting sufficient amounts of the proceeds to keep net output
constant. The way to do this is to reinvest the replacement cost depreciation from the gross yield. By doing
this it is possible to maintain a steady flow of (cδ/qδ -δ) where cδ is here interpreted as the consumption-good
flow of output from the new machine, and qδ is taken to be the consumption-good cost of a machine of type δ.
17 (Note the shift from price to real productivity and cost concepts.)
Thus at least if attention is confined to constant net output "steady state" paths, efficiency in the allocation of
each instant's investment requires
Â
Page 131
if both types δ and δ' are employed.
The Investment Tax Credit and Efficiency with True Holder's Depreciation
In order to assess the efficiency implications of various rules for the investment credit, it is necessary to
consider the relationship between the marginal productivity of machines, cδ, and the marginal consumption
good cost, qδ, of the immediately preceding analysis and the rental price of machines, c, and the market price,
q, of a machine in competitive equilibrium. To keep the analysis within manageable bounds I make the usual
growth-model assumption that the production process generates output that is either immediately consumable
or convertible to machines of various types (identified by δ) on fixed terms (depending possibly on the stock
or rate of production of the machine type in question). This means that both c and q can be regarded as
measuring correctly the consumption opportunity cost of, respectively, the use and the construction of a
machine. Since I am now concerned with machines of different durabilities, furthermore, I refine the price
notation, and let cδ and qδ represent, respectively, the rental price of a unit of type δ machine services, and the
price of one new type δ machine.
Under these conditions efficiency in investment can be identified with the equating of the before-tax rate of
return (internal rate of return) from investment. This internal rate of return on type δ is given by
.
Condition 7 shows that when k = 0 this requirement is satisfied in equilibrium, regardless of the proportional
tax rate, u. 18
When k is not zero, condition 7 says that cδ/qδ - δ = (1 - k)i - kδ. While i is the same for all asset types, δ
obviously is not.
112
113
Consider now the question of

the relationship between asset
durability and investment credit
needed (when all assets display
exponential capacity decay) to
obtain efficiency in the
presence of the credit. Because
efficiency requires cδ/qδ - δ to
be the same for all δ, formula 7
can be used to express the
requirement for k:
Â
Page 132
where r is the internal rate of return common to all investments in an efficient program. Let k(δ) be the
relationship between tax credit and durability required for efficiency. Then, solving formula 9 for k(δ)
explicitly,
is obtained.
As expected, the subsidy must increase with durability (decline with δ) in order to obtain neutrality, with a
zero subsidy for a zero life asset (δ = ℜ∞). The "right" subsidy for each durability cannot, however, be
specified in advance unless one can correctly anticipate the equilibrium interest rate.
Point Input, Point Output, and Related Investment Models
To explore the effect of varying the exponential decay assumption I consider in this section the cases of (1)
point input, point output and (2) point input, constant flow output production.
Point Input, Point Output
The assumption here, as it was implicitly in the previous section, is a homogeneous output that can either be
consumed or invested. In this case, however, one cannot naturally associate units such as "number of
machines" with investment. Consequently it will be simpler to consider a unit investment of type d as costing
one unit of consumption and yielding yd units of consumption d time units later. This is to be understood as
the entire return, liquidating the investment.
Efficiency and Equilibrium
The efficiency condition in this model, corresponding to condition 8 in the exponential decay model, may be
derived by noting that the time pattern of an investment with returns after, say, two periods can be reproduced
by investment in a one-period project together with reinvestment of all proceeds for a further period.
Generalizing, this requires for all d and d' representing investments actually undertaken in positive amounts
Â
113
114
Page 133
The internal rate of return, rd, associated with an investment in form d is given by
or
.
It will be immediately verified from formula 10 that efficiency implies equalizing the internal rate of return
on investment opportunities pursued. In the absence of taxes, wealth maximizing capitalists will equate the
internal rate of return at all investment margins to the rate of interest;
for all d, thus satisfying necessary condition 10 for efficiency.
The Effect of an Investment Credit
It will simplify the analysis of the effect of an investment tax credit if one takes for granted the proposition
that if correctly calculated economic depreciation is used for tax purposes, tax rates do not upset the
efficiency of market equilibrium in the sense of condition 10. The effect of a credit at percentage rate k is
then to change the market equilibrium in condition 12 to
where is the after-credit rate of return from investment in asset type d, given by
If expression 14 is used, equilibrium condition 12 can be rewritten:
As in the previous model, the rate of credit, k, must be varied with the durability of assets (interpreted here as
the time, d, between point investment and point return) to assure satisfaction of efficiency condition 10.
Unlike the previous case, there is in this model a method for doing this independently of the equilibrium rate
of interest: k need simply be varied to maintain the constancy of ln(1 - k)/d.
Â
Page 134
An explicit expression for the schedule, k(d), of credits necessary to assure a given common before-subsidy
rate of return, r, can be obtained by solving expression 15 to obtain
.
Point Input, Constant Flow Output
In the last special investment model I assume that in return for a point input of one dollar of forgone
consumption a constant flow of output at rate xT can be generated from the instant of investment until a time
114
115
T periods later, at which point the investment is exhausted.
To analyze the efficiency conditions for a program of investments of different durabilities it will be
convenient to assume that the point input, point output opportunities are also available. Since a point input,
flow output pattern can be reproduced by an appropriate mixture of point input, point output projects,
efficiency requires that the cost of the two methods of obtaining the same effect be the same. In particular, by
undertaking xT/yd units of type d point input, point output investment for all d between 0 and T, a steady
output flow of xT per period is produced over the interval. This investment program costs
and efficiency requires this to equal 1, Let r, defined as in expression 11, denote the common internal rate of
return on all point input, point output projects in an efficient program. Then, using expression 16, efficiency
in the point input, constant flow output case requires
for all "durabilities," T, in use. This is precisely the requirement that the internal rate of return, rT, defined in
expression 18 also equal r for all T employed.
Â
Page 135
In the absence of investment subsidies, wealth maximization will lead to the equating of all investment
returns to the market rate of interest, by the usual arguments. Let denote the after-subsidy internal rate of
return, defined by
.
Equilibrium now requires
for all durabilities, T, employed.
It will not in general be true that

the before-subsidy internal rate of
return associated with this
equilibrium will be the same for
all T, as required for efficiency.
Suppressing the associated
algebra, one can write the
variation in k with T required to
assure satisfaction of expression
17 as
115
116
,
where r is the common social rate of return on real investment and i is the market rate of interest, which by
expression 19 also equals the rate of return to the investor.
As in the two previous models, efficiency requires the subsidy rate to increase with the life of the investment.
Further, as in the exponential decay model, but not as in the point input, point output case, it is necessary to
know the equilibrium situationthat is, to know iin order to obtain the rule for the efficiency-preserving
relationship between credit and asset durability.
Summing Up
The foregoing analysis gives a formal demonstration of two general principles: (1) that a credit varying with
the durability of the investment is required for neutrality of the subsidy with respect to this aspect of
investment choice; and (2) that a different structure of such credits will in general be necessary for different
patterns of returns. Table 5.1 suggests the importance of these principles as a practical matter. This table
illustrates the structure of credits required to
Â
Page 136
Table 5.1
Illustrative schedule of durability-neutral investment tax credita Rates in percent
Exponential depreciationb Point input, Point input,

point outputc constant flow outputd
Durability Neutral Durability Neutral investment Durability Neutral investment

parameter investment credit parameter credit ratef parameter credit rategk(T)
δ ratee d k(d) T
k(δ)
1 1.8 1 20 1 1.0
1/2 3.3 2 3.9 2 1.9
1/3 4.6 3 5.8 3 2.8
1/4 5.7 4 7.7 4 3.7
1/5 6.6 5 9.5 5 4.5
1/7 8.2 7 13.1 7 6.1
1/10 10.0 10 18.1 10 8.2
1/15 12.0 15 25.9 15 11.1
1/20 13.3 20 33.0 20 13.3
0 20.0 ℜ∝ 100.0 ℜ∝ 20.0
a. Under various assumptions about form and duration of return patterns, entries show investment credit required to
generate a rate of return (i) of 10 percent to the investor when the common underlying rate of return (r) is 8 percent.
-
b. This can be interpreted as assuming the asset disappears at rate δe δs, where s is the time since construction.
c. A trait investment at time 0 yields yd at time d.

d. A unit investment at time 0 yields a steady flow xι of output until time T.
116
117
e. k(δ) = 1 - (r + δ)/(i + δ).
f. k(d) = 1 - exp(r - i)d.

g. k(T) = 1 - (r/i)[1 - exp(-iT)]/[1 - exp(-rT)].
induce a common before-credit rate of 8 percent on investments of various types and durabilities while
providing the investors with a common 10 percent after-credit (before income tax) rate of return in every
case.
Unfortunately it is difficult to relate these illustrative figures to actual tax practice or to draw policy guidance
from them beyond what emerges from the qualitative argument of the theory. Indeed, the theoretical
argument is less than satisfactory. Apart from their being cast in steady state framework, the theoretical
conclusions have the shortcoming of being based on the assumption that the market cost of investments
correctly measures the consumption goods forgone. 19 But presumably the relative market prices of various
investment and consumer goods are influenced by the investment tax credit.
Â
Page 137
This issue is too complicated to do more than suggest the nature of the problem here. Suppose, for example,
that a machine of some type can be produced either directly by conversion of consumption goods or
indirectly by the machine itself. The higher the investment credit, the lower will be the equilibrium rental
price machines. As a result, machines will increasingly be produced by other machines rather than by direct
conversion of consumption goods. The outcome is a tax-induced inefficiency in the production of machines.
If the machine-produced machine differs from the directly-converted-from-consumption-goods machine with
respect to durability, it may be possible to offset this inefficiency through appropriate useful life
discrimination in the credit rules.
There is another shortcoming in the theoretical argument that appears in many efficiency analyses: the
absence of guidance on how much difference mistakes make. Remedying this defect seems out of the
question, however, as it would require a knowledge of the elasticities of substitution of each machine type for
others in production, as well as of the production conditions of the machines themselves. (Remember that the
term "machine" is used here to represent all forms of investment.)
If these technical problems are

overlooked, practical difficulties
in using the figures in table 5.1
remain. For one thing, it is not
clear how one should associate
the various durability parameters
of the table to the concept of
"useful life" in U.S. tax law.
Does an asset exponentially
decaying at 10 percent a year
(average life, ten years; half life,
seven years) have a longer or
shorter useful life than one
producing all its output exactly
ten years after construction or
one producing a constant flow
output for, say, fifteen years?
117
118
Second, allowance must be made

for the fact that the figures in the
table are based on requiring a
deduction of investment credit
from the basis for depreciation
and using true economic
depreciation for tax purposes.
Basing his calculations on an unpublished paper by T. Nicholaus Tideman, Sunley derived the investment
credit schedule required for neutrality (in the sense analyzed here) assuming (1) original purchase price used
as the basis for depreciation; (2) exponential decay of assets; (3) sum-of-years method of depreciation for tax
purposes, interpreting 2/δ as the useful life of an asset decaying at rate δ; and (4) a marginal tax rate of 0.48
(needed because tax depreciation is different from economic depreciation). 20 Sunley's neutral credit
Â
Page 138
Table 5.2
Illustrative schedule of neutral investment credit under "plausible" depreciation assumption
Neutral
Durability investment Present U.S.
parametera Useful life credit rate law creditb
δ (years) (percent) (percent)
2/3 3 2.7 3.3
2/5 5 4.0 6.7
2/7 7 5.1 10.0
1/5 10 6.5 10.0
1/6 12 7.3 10.0
2/15 15 8.4 10.0
1/10 20 9.9 10.0
Source: Based on Emil Sunley, Jr., "Tax Neutrality between Capital Services Provided by Long-Lived and
Short-Lived Assets," in U.S. Department of the Treasury, OTA Papers, vol. 1: Compilation of OTA Papers
(GPO, 1978), p. 15. Investor obtains 10 percent return (before tax) on all asset classes; before-credit return is 8
percent on all assets. For details of assumptions see text.
a. Exponential decay rate.
b. Neglects extra credit for employee stock ownership plan.
schedule is shown in table 5.2. It is interesting and somewhat surprising how similar the neutral credit rates
are under the more realistic assumptions made by Sunley and those for exponential decay in table 5.1. The fit
of both of these with present law also appears better than might have been expected, and it looks even better
if one measures useful life by the average or expected life of the asset, 1/δ, instead of 2/δ.
On the other hand, under the exponential depreciation assumption there is clearly a bias in present law against
the long-lived assets, and there is no way of assessing the importance of other patterns of investment returns.
These are but two relatively unimportant reasons for a substantial revision in the way capital income is
measured and taxed. But this is the subject of another paper. 21
Notes
118
119
The author would like to acknowledge helpful discussions with Seymour Fiekowsky, Harvey Galper, and
Roger Gordon on the subject of this paper. Any opinions expressed are those of the author and not of
Princeton University or the National Bureau of Economic Research.
Â
Page 139
1. This description neglects the extra credit that is allowed, contingent on employer contributions to a
qualifying employee stock ownership plan.
2. The Long Amendment (named for the chairman of the Senate Finance Committee, Senator Russell B
Long), a part of the original investment credit enacted in 1962 but subsequently repealed with retroactive
effect, required subtraction of the investment credit from the purchase price of the asset in the calculation of
depreciation allowances.
3. It is sometimes argued that the increasing subsidy for longer-lived assets is justified primarily as a rough
offset to the failure to use as the basis for depreciation the true effective cost to the investor, namely, the
after-tax credit cost.
4. Emil Sunley, Jr., "Tax Neutrality between Capital Services Provided by Long-Lived and Short-Lived
Assets," in U.S. Department of the Treasury, OTA Papers, vol. 1: Compilation of OTA Papers (Government
Printing Office, 1978).
5. That is, the Long Amendment is necessary. It is not suggested that the depreciation rules available under
U.S. law are correct; they certainly are inadequate in a time of inflation.
6. Such shortcomings of simple approximating rules need not imply the procedures should be changed. The
relative performance of different rules of thumb is the issue. It does appear that for purposes of subsidizing
investment a partial write-off for tax purposes would be both simpler and superior to present practices from
the point of view of neutrality.
7. See, for example, Robert E. Hall and Dale W. Jorgenson, "Tax Policy and Investment Behavior," American
Economic Review, vol. 57 (June 1967), pp. 391-414.
8. See Paul A. Samuelson, "Tax Deductibility of Economic Depreciation to Insure Invariant Valuations,"
Journal of Political Economy, vol. 72 (December 1964), pp. 604-06.
9. See Hall and Jorgenson, "Tax Policy and Investment Behavior."
10. The analysis that follows is most easily understood as applying to direct investment by individuals. It will
apply to corporations as well if it is assumed that the same marginal rate of tax applies to shareholder equity
returns and the bond interest of those shareholders. For a fully satisfactory treatment, however, one would
have to resolve analytical problems (for example, the existence of many different shareholder tax rates)
beyond the scope of this chapter.
11. This roughly describes present law.
12. These formulas apply for an investment held forever. Hence it is not necessary to be concerned about the
proceeds of sale of the asset. This will only be wealth maximizing appropriate tax rules, even in the steady
state with constant c, q, u, and i, as is discussed further below.
13. As the derivation indicates, because tax rates and hence after-tax rates of return vary among taxpayers,
the supply price of machine services may vary from taxpayer to taxpayer. This variation is incompatible with
equilibrium in the market for machine services. Such inconsistency may be resolved by sorting out asset
119
120
types by the tax brackets of their owners, as occurs in tax shelter situations. In what follows, to avoid having
to deal with this, assume u is the same for all taxpayers. For an analysis of how
Â
Page 140
opportunities for profit are eliminated when taxpayers face different rates, see David F. Bradford, ''The
Tax System, Saving, and Capital Formation," in George von Furstenberg, ed., Capital Investment and
Saving, vol. 2: The Government and Capital Formation (Ballinger, 1980).
14. See Gerard M. Brannon and Emil M. Sunley, Jr., "The 'Recapture' of Excess Tax Depreciation on the Sale
of Real Estate," National Tax Journal, vol. 29 (December 1976), pp. 413-21, for an interesting and surprising
discussion of recapture rules in the U.S. income tax.
15. Recall that the assumption is that all machines having the same durability coefficient, δ, are identical,
regardless of the time since construction.
16. This is the treatment under the Long Amendment (see note 2).
17. To demonstrate this proposition, let I(t) be the amount reinvested in asset type δ at time t, where the
whole process is regarded as starting at time 0. I(t) will be the difference between the gross yield from the
initial investment plus subsequent reinvestment and cδ/qδ - δ. It will thus satisfy
.
One may quickly verify by substitution that the solution to this integral equation is given by
That is, the program of constantly reinvesting the depreciation flow maintains gross output constant and
thus maintains net output at cδ/qδ - δ. To obtain any higher net output flow would require at least for some
finite time accepting a lower net output flow.
18. This is the point established in Samuelson, "Tax Deductibility of Economic Depreciation."
19. All of this assumes that the second-best rules call for production efficiency. Even this rule may not hold
under some circumstances; see J. E. Stiglitz and P. Dasgupta, "Differential Taxation, Public Goods, and
Economic Efficiency," Review of Economic Studies, vol. 38 (April 1971), pp. 151-74.
20. See Sunley, "Tax Neutrality between Capital Services Provided by Long-Lived and Short-Lived Assets";
and T. Nicolaus Tideman, "Refinements in the Formula for the Price of Capital Services and Their
Application to Tax Neutrality" (1975).
21. The results of this paper and

those of Arnold Harberger's paper
[in Henry J. Aaron and Michael J.
Boskin, eds., The Economics of
Taxation (Washington, D.C.: The
Brookings Institution, 1980] are
complementary. I show that if
depreciation is allowed on the net
price of a capital goodthe purchase
price less the investment tax
creditthen a neutral investment
120
121
credit depends on the pattern of

true economic depreciation and
rises with durability. Harberger
shows that if depreciation is
allowed on 100 (1 - k/t) percent of
the purchase price of a capital good
(where k is the investment credit
rate and t is the income tax rate),
then a constant investment tax
credit is neutral. To illustrate, if the
tax rate is 40 percent and the credit
is 10 percent, then the credit is
neutral with respect to durability if
the firm is permitted tax
depreciation equal to true
economic depreciation on 75
percent of the value of the capital
good; if the firm can depreciate
any large fraction of the value, and
in particular if it can depreciate 90
percent as under the Long
Amendment, then a constant credit
favors short-lived capital goods
and a neutral credit must rise with
durability.
Â
Page 141
6
The Economics of Tax Policy Toward Savings
Any effort to analyze the effect of U.S. federal tax rules on the accumulation of wealth and on the stock of
capital used in production confronts the fact that there is as yet no consistent policy toward savings. An
individual who deposited $100 in a 5.5 percent savings account in November 1977 in order to increase his
Christmas shopping resources a year later would have experienced a rate of return before taxes of minus 3.2
percent when the effect of inflation is taken into account. This would have been further reduced by federal
income tax on his interest "return," and if he were lucky enough to be in the top (70 percent) marginal rate
bracket he would have earned minus 6.7 percent after this tax. If, on the other hand, he had deposited the
same $100 in a "HR-10 plan" of retirement savings as a self-employed person, or into an "IRA" (Individual
Retirement Account) as an employee not covered by an employer-provided plan that qualified for tax-favored
treatment, or if his employer had put an extra $100 into such a qualified plan, and if between deposit and
withdrawal he had reached age 59.5, and if his marginal tax rate had not changed over the period, he would
have received the full, real return of minus 3.2 percent. If his marginal rate had dropped sufficiently between
the two years, for example because he retired, the tax system might actually have generated a positive return
on his savings. 1 If our individual had, instead, decided to purchase equipment for his small manufacturing
business, he might have found the tax system ready to help with a 10 percent investment tax credit, although
it would not have recognized the inadequacy of his future depreciation deductions based on historical cost if
the price level had continued to rise.
121
122
These are relatively simple cases of inconsistency of the tax treatment of savings in the United States. One
could easily go on and as
Â
Page 142
easily construct more striking, if less representative, pathologies. The variety of tax provisions and the
complexity of their interactions make generalization about the effect of taxes on savings difficult. Yet it is
possible to describe the existing system as a combination of relatively simple basic approaches, and to model
the way contradictions among these may be expected to be resolved through adjustments in markets and in
individual behavior. Such an analysis is the objective of this chapter, which will also offer my own view of
desirable directions for change in tax policy toward savings.
The current federal income tax treatment of savings is characterized by considerable variation in the
treatment of any act of saving according to the particular asset acquired, the form of its ownership and
method of financing, the circumstances of its purchase, and the rate of general price change. The resulting
system carries with it costs of a well-known sort to individual taxpayers in optimizing their affairs and to the
revenue collection agency in defending the fisc. But quite apart from these costs of administration, and
certainly far larger in magnitude, are the economic losses due to inefficient resource allocation and the
political losses due to what are perceived as inequities in the working of the tax rules.
The existing tax code is the result of a long process of incremental adjustment, and the uncoordinated nature
of the solutions adopted to a sequence of particular problems bears a significant share of the responsibility for
the system's shortcoming. One does not expect tidy outcomes from the federal legislative process, yet the
present muddle reflects, I believe, more than the inevitable resultant of conflicting interests. It is the
consequence as well of the lack of any clear conception of how the tax system works, and of a persuasive
organizing principle or strategy based on such a conception around which a coherent structure of rules can be
formed.
As far as the question addressed in this chapter is concerned, to the extent there is a dominant model of how
the system works, it is of an essentially partial character. Roughly speaking, this conception holds that taxes
on capital incomeincome taxes, estate and gift taxes, and property taxescause a bias against saving for future
consumption. Lightening the tax on selected forms of wealth, via favorable treatment of capital gains,
accelerated depreciation, and the like, is called upon to correct the allocative inefficiency created by this bias.
Unfortunately, this model is rarely, if ever, connected systematically to the underlying equity problem of fair
distribution of tax
Â
Page 143
burdens. Concessions to savings are generally viewed as unfortunate deviations from the closest thing to an
"organizing principle" operative in the period of development of the modern U.S. tax system. That principle
is taxation according to "Haig-Simons" income. 2 As expressed by Henry Simons [59, 1938, p. 50], this
principle holds that taxes should be based on "income" understood as "the algebraic sum of (1) the market
value of rights exercised in consumption, and (2) the change in the value of the store of property rights
between the beginning and end of the period in question."
122
123
The important Tax Reform Act of 1969 can be regarded as predicated on greater fidelity to Haig-Simons
income, as can many of the features of the Tax Reform Act of 1976. The policy pressures working against
this strategy, based on concern about the adequacy of national investment and about individual savings for
retirement, were especially evident in the legislative process leading up to the Revenue Act of 1978, which
lowered the corporate tax rates, made permanent the investment credit, and substantially lowered the taxes on
realized capital gains.
If "income" is indeed understood to be the sum of consumption expenditure and savings over the accounting
period, such a tendency toward eliminating the taxation of the yield from savings can be interpreted as a
movement toward taxing on the basis of consumption. This has been noted by several observers. William
Andrews [1, 1974], for example, in a celebrated paper in the legal literature, explored the taxation of
consumption both as a model to explain many features of existing law which are hard to reconcile with a
Haig-Simons income tax ideal and as a framework for reform and rationalization of the tax law. In the U.S.
Treasury's [67, 1977] Blueprints for Basic Tax Reform (hereafter cited simply as Blueprints) the similarity of
treatment of many transactions under existing law to that under a consumption ideal was used to motivate
consideration of a consumption base as one of two reform models. Merton Miller and Myron Scholes [53,
1978, p. 49], in the context of explaining how present tax rules permit the avoidance of tax on dividends,
concluded, "So many and so varied, in fact, are the devices for taxfree accumulation, so large are the amounts
invested in such devices and so rapidly are they growing that we have been driven to a conclusion that
transcends in important respects the dividend issues that were our initial motivating concern. To put it
bluntly, the 'income' tax is dead." They argue that this is neither coincidental nor
Â
Page 144
undesirable. In a comprehensive review of the United Kingdom tax system, a committee chaired by Professor
James E. Meade [46, Institute for Fiscal Studies, 1978] (hereafter cited simply as the Meade Report) found
inconsistencies in the treatment of savings in Britain that are similar to those described above for the United
States and concluded with a recommendation for adoption of an expenditure tax base.
Taxation according to Haig-Simons income and taxation of consumption thus offer alternative ideals around
which tax policy may be organized, and because of their very different treatment of savings, the choice
between the two will be referred to throughout this study. I do not propose here to review at any length the
arguments that a consumption base is preferable from the point of view of equity and administrative
simplicity. These were drawn together in [14, Bradford, 1980], and the basic pros and cons presented in that
paper will be taken for granted. The emphasis here will be on extending the efficiency analysis and drawing
out the connections between the choice of tax structure and the accumulation of wealth.
I shall use the term "tax structure" or "structural tax policy" to refer to the set of rules determining the relative
tax rates levied on various transactions, to be distinguished from the question of the overall level of taxation
and the deficit or surplus posture of the budget. Section I of this chapter presents a brief overview of the
federal tax structure as it bears on the savings decision. In Section II, I take up briefly the efficiency aspects
of the choice of structural tax policy toward savings, and, in Section III, the question of how structural tax
policy relates to capital accumulation. In Section IV, attention is shifted from the general issue of the taxation
of savings to the institutional details through which policy is embodied in the individual income tax. The
subject turns out to be sufficiently complex to merit treatment in greater depth in Section V. Section VI looks
at the economics of the corporation income tax, and Section VII presents the policy conclusions I draw from
the story as I see it.
I
An Overview of Federal Taxation of Savings
123
124
Before going into details of the federal income tax, some aggregate statistics may give a feel for the overall
effect of the federal tax system on the reward to saving. In fiscal year 1978 the federal government spent
$541b. 3 This outlay was financed in part by taxes, of
Â
Page 145
$181b assessed on individual incomes, $123b on payrolls, $60b on corporate profits, $25b on various specific
goods and services (excise taxes and customs duties), and $5b on gifts and estates. Miscellaneous receipts
accounted for $7b, with the remainder financed by a deficit of $48b.
These revenue sources can be roughly classed into taxes on exchanges of present for future consumption
("taxes on capital") and taxes unrelated to this present-future trade-off ("taxes on labor"). The corporation
income and estate and gift taxes (with some qualification) should be placed in the former category, while the
payroll and specific taxes may be placed in the latter. Miscellaneous receipts might best be netted out of total
outlays. This leaves individual income taxes and the deficit, which should be regarded as levied on both labor
and capital returns, and for our purposes must be allocated between them. Since we are only after a rough
approximation to average tax rates, let us simply divide the individual tax in the same proportion as the
aggregate division between returns to capital and returns to labor. Finally, assume the deficit amounts
implicitly to a proportional increase in taxes from all sources. 4
Table 6.1 gives a version of aggregate income flows during fiscal year 1978, using national income
accounting data from the fourth quarter of calendar year 1977 and the first three quarters of 1978. These
aggregates are based on definitions somewhat different from
Table 6.1
Returns to capital and labor, 1977 IV-1978 III ($ billion)
Returns to capital 305
Corporate profits with inventory valuation and capital consumption adjustments 152
25 percent of proprietors' income with inventory valuation and capital consumption adjustments 27
Rental income of persons with capital consumption adjustment 23
Net interest 103
Returns to labor 1343
Compensation of employees 1261
75 percent of proprietors' income with inventory valuation and capital consumption adjustments 82
National income 1649
Source: Economic Report of the President, January 1979, Table B-19. (Because of rounding, detail may not add to totals.)
Â
124
125
Page 146
those we might choose for tax purposes (for example, capital gains and losses are ignored), but, again, for
purposes of a rough approximation this need not trouble us. The classification into capital and labor returns
requires an allocation of proprietors' income, which I have based on the conventional stylized fact of a one
quarter-three quarter capital-labor split.
Putting all the pieces together gives us total federal taxes on the returns to savings of $111b for an average
rate of 36 percent, and total federal taxes on labor and consumption of $332b, for an average rate of 25
percent on the labor earnings base. The average rates are such as one might expect to emerge from a
progressive income tax regime in an economy where capital income is skewed toward the upper end of the
income distribution. 5 One might think these figures resulted from a reasonably coherent objective of taxing
according to income. Yet a closer look reveals a bewildering array of tax rules affecting the future yield from
present consumption foregone.
Consider some examples of the features of U.S. law tending to treat the yield from savings favorably relative
to a "pure" income tax:
1. Perhaps best known is the special treatment of capital gains. In the first place such gains are taxed only on
realization by sale or exchange; long-term deferral of tax can greatly reduce it. Furthermore, when assets pass
to heirs by bequest the basis for calculating capital gains to the heirs is set at the value at the time of bequest;
any gain during the giver's lifetime thus goes free of income tax.6 Equally important, 60 percent of capital
gain on assets held a year or more is excluded from the individual income tax base (although an "alternative
minimum tax" puts a floor under the effective rate).
2. Contributions by employers to qualified pension plans are excluded from the taxable income of employees;
all of the earnings of the pension fund are exempt from tax. Similar treatment is accorded contributions to
and earnings on retirement plans for the self-employed and employees not covered by a qualified employer
plan (subject to fairly modest limits). Pension payments received are fully taxable to individuals but generally
at lower rates than apply at the time of the contributions.
3. Life insurance policy holder reserves are lightly taxed in times of stable prices; accruing cash surrender
value is not taxed as income to the holder, nor are the proceeds to the beneficiary taxed when received.
Â
Page 147
4. The entire yield in kind of household durables is excluded from tax. A special case of this is
owner-occupied housing; the associated mortgage interest payments and real estate taxes (though not
maintenance expenses and depreciation) are nonetheless deductible from the individual income base. Liberal
"roll over" provisions allow the taxpayer to avoid realizing gain on sale of a house when he purchases a
replacement, and a $100,000 tax-free capital gain is allowed once in the lifetime of a taxpayer over age
fifty-five.
5. Investment in human capital

that is financed by foregone
earnings in effect gets qualified
pension treatment: increased
future earnings financed
through present earnings
foregone are not taxed until
they materialize. 7 This implies
125
126
that the interest on the deferral

is not taxed. However, unlike
earnings deferred to retirement
years via a pension plan, the
deferred receipts attributable to
education are likely to
encounter higher marginal tax
rates than would have applied
to the earnings foregone.
6. Interest on qualifying state and local bonds is excluded from income subject to tax.
7. It is widely, though not uniformly, believed that the rate at which assets may be written off for tax
purposes is excessive in the absence of inflation. This applies especially to real estate, and it is thought to
occur also on assets accounted for under the Asset Depreciation Range (ADR) procedures.
8. A credit against tax is allowed for up to 10 percent of the cost of business equipment for domestic use. The
full credit is granted for assets with a useful life of seven years or more; reduced credits are given for less
durable assets. Additional investment tax credit allowances are available for corporations making
contributions to an employee stock ownership plan (ESOP).
9. The first $100,000 of taxable corporation income is taxed at rates ranging from 17 percent to 40 percent,
below the flat 46 percent rate applicable to all income in excess of $100,000, and for many small companies
well below the marginal rate applicable to shareholder income.
10. The first $100 of dividends ($200 for a joint return) is excluded from individual tax.
A variety of special provisions lighten the tax load on investment in particular lines of business:
Â
Page 148
1. Although presumably directed at acquiring an asset with some durability, research and development
expenditures may be expensed, that is, deducted in the year incurred, rather than amortized over a period of
years.
2. "Intangible expenses" (that is, outlays not embodied in structures or equipment) of drilling for oil and gas
or geothermal energy may be written off (though they are subject to a special taxthe minimum taxif not
balanced by otherwise taxable receipts from oil and gas or geothermal energy).
3. Although the major oil and gas producers are excluded from the rule, independent oil and gas producers
and most other extractive industries may use "percentage deletion" methods, which tend to overstate,
sometimes dramatically, the allowance for decline in asset value appropriate for proper income measurement.
4. Tax credits are provided for business (and household) purchase of energy-saving equipment (in addition to
the regular investment tax credit).
5. Investment in pollution control facilities may be amortized over five years.
6. Gains from sale of timber, livestock, and some other agricultural products, as well as royalties from iron
ore deposits, are treated as long-term capital gains.
7. Overly generous bad-debt reserve deductions substantially reduce taxes on financial institutions.
126
127
While all of these elements of the tax system tend to reduce the rate of tax on the yield from savings, two
major elements of the tax system work in the other direction. The first is the two-tier system of taxation of
income arising in corporations. While, as we shall discuss, recent thinking on the effects of the corporation
income tax has substantially qualified earlier views, the simple evidence of collection of $60b in revenues
from this source during fiscal year 1978 indicates its significance as a tax on savings. Using our earlier
figures, this tax alone would constitute a 20 percent levy on all returns to capital.
The second element is the tax-increasing effect of inflation working through the procedures by which returns
from capital are measured for income tax purposes. The most obvious defects in these procedures are in the
measurement of capital gains and depreciation. In an
Â
Page 149
analysis of a sample of more than 30,000 individual income tax returns showing capital gains realized on
corporate stock in 1973, Martin Feldstein and Joel Slemrod [30, 1978] made the startling discovery that
whereas individuals were taxed on $4.6b of such gains, adjustment for the increase in price level would have
produced instead a loss of nearly $1b. They found, furthermore, that the difference between nominal gains
and real gains (commonly losses) varied systematically by income class, with the highest income category
experiencing the least divergence and the lowest income category the most. In another study, Feldstein and
Lawrence Summers [33, 1979] looked at the overstatement of income attributable to the use of historical cost
as the basis for depreciation allowances and to current methods of inventory accounting. According to their
estimate, in 1977 the tax burden on corporate sector capital income was larger by $32b than it would have
been with properly indexed depreciation allowances and inventory accounting, an increase in effective
(average) tax rate from 43 to 66 percent on this income flow.
II
Structural Tax Policy: What Should the Tax Rate be on the Return to Saving?
In the introductory section we noted that the political debate about the relationship between tax policy and
capital formation makes implicit use of rather simple economic models. In view of the complexity of the
subject and the generally rudimentary knowledge of economics among the principal actors determining tax
rules this is not surprising. However, a small increase in the sophistication of the economic modeling enables
us to see the ways in which arguments frequently encountered are wrong or misleading. It can also help us
appreciate the difficulty of, on the one hand, finding wholly persuasive grounds for particular policies in a
priori reasoning or, on the other hand, convincingly resolving policy issues by econometric evidence. In this
and the following sections we use rather simple but complete models to examine two broad issues of tax
policy and saving, (1) the normative one of how savings "ought" to be taxed and (2) the positive one of how
taxation of savings will affect aggregate capital accumulation.
Our main conclusions are somewhat negative. On the first issue we see that the answer according to
conventional criteria of welfare economics depends upon various empirical parameters. The prospects
Â
127
128
Page 150
are not bright for estimation of these parameters, let alone the far more extensive list of coefficients required
of a ''realistic" model. On the other hand, it is argued that the configuration of parameter values leading to a
consumption base is not implausible, indeed, it has been frequently encountered in empirical analyses of
savings behavior.
The conclusion on the second issue is of a somewhat different character. It is argued that the relationship
between tax structure and wealth accumulation depends crucially on the constraint assumed to apply to
government borrowing (and lending). Where this is unconstrained, virtually any objective for aggregate
accumulation can be met on the basis of any tax structure. The two issues can then be separated. Attention in
discussions of tax structure can and should be directed toward other concerns, provided the requirements for
overall fiscal policy are understood and taken into account.
The reader may feel a disconcerting lack of connection between the discussion here and ringing phrases about
capital shortages and job creation normally associated with these questions. Wealth often seems to be
regarded as good in itself and it is taken as a matter of self-evident concern that the savings rate in the United
States is lower than that in other countries, or that the expected accumulation over some future period will
fall short or extrapolated "requirements." 8 In some cases these concerns are explicitly related to the need for
capital to enable the employment of additional labor, thus relieving unemployment.
Economic analysis leads us to regard with skepticism policy claims based on attention to the aggregates per
se rather than on consideration of the underlying purpose and cost of accumulation. And most economists
consider that the flexibility of productive technology in the United States is such as to permit full
employment of labor, perhaps excepting short-run circumstances of capacity constraints.9 In this view the
issue of capital accumulation is not one of employment versus unemployment but one of more or less
remunerative employment. Once this is recognized the concern of policy can sensibly be directed to the
question of how attractive are the returns on investment at the margin.
Feldstein [26, 1977; 27, 1977], for example, has considered this issue and concluded that the rate of return on
investment in the United States (which he and Summers [31, 1977] place at roughly 12 percent) considerably
exceeds what would be required to compen-
Â
Page 151
sate for postponing consumption at the present margin. This is because taxes on private investment yield
introduce a "wedge" between extra output generated by the investment and the reward received by the saver
who provides the resources. The extra investment does not take place because the government takes part of
the yield.
It is tempting to conclude from this that tax policy should be changed to remove this distortion. Yet by the
same line of reasoning any positively (or negatively) taxed activity will be at an inefficient level in
equilibrium. If it is the case (as it seems to be) that nonzero tax rates are required to finance government
activity, some inefficiency is virtually inevitable. The question is how much of that inefficiency "should" be
reflected in a tax on the returns to saving.
Views on this question are bound to vary and to depend, among other things, upon whether one is concerned
with the long-run operation of the rules or with the distribution of tax burdens among identifiable groups in
the short run. Even if one focuses, as we do here, primarily on the long-run properties of the system,
conclusions will be sensitive to the underlying model of the economy. Thus neo-classical growth models
generally use savings propensities that are independent of factor prices, though in some models they differ
128
129
according to the form of income [2, Kenneth Arrow and Mordecai Kurz, 1970; 23, Feldstein, 1974; 24, 1974;
39, Ronald Grieson, 1975]. Such models typically rest implicitly on special assumptions, either about
individual savings behavior under different rates of return on savings or earnings from labor, or about the
degree of capital market imperfection. For our purposes, which are to understand the issues involved and to
judge whether persuasive policy conclusions are likely to emerge from more detailed analysis, it is preferable
to have a model in which the fundamental building blocks are individual or household and producer choices.
In the life cycle model of capital accumulation the individual choice problem is the standard one of
microeconomic theory: choose the most preferred bundle of commodities within a budget constraint. In this
case the "commodities" are usually abstracted to quantities of a homogeneous consumption good,
distinguished by date of availability to the individual, either for own use or for transfer to others by gift or
bequest. The bundles available to the chooser are determined by earnings from labor and nonlabor receipts
(in heritances, gifts, government transfers, and so on) as these may be
Â
Page 152
transformed via borrowing and lending in the capital market. In the simplest "perfect" capital market world
without uncertainty the budget constraint requires that the present value of the stream of outlays on
consumption and gifts and bequests (discounting at the market interest rate) equal the present value of the
stream of earnings, inheritances, transfers and so on.
These ideas have been drawn upon widely, certainly since the time of Irving Fisher [34, 1930], who,
incidentally, favored taxation on the basis of consumption. They were applied to the analysis of aggregate
savings behavior in the famous studies of Milton Friedman [35, 1957] in which individuals are viewed as
taking a lifetime perspective in choosing a rate of "permanent" consumption out of "permanent" income.
Earlier Franco Modigliani and Richard Brumberg [54, 1954] had provided a key simplifying assumption, that
an exogenous increase in wealth will lead to a proportional change in consumption in all periods. Initially the
life cycle model was directed toward understanding short-run reactions of aggregate consumption
expenditures to changes in income, as these are statistically defined. 10 Subsequent elaborations (for example,
by the introduction of liquidity constraints) have been concerned with medium-term effects of changes in
fiscal policy [65, James Tobin and Walter Dolde, 1971] or the determinants of wealth and income distribution
[10, Alan Blinder, 1974]. For our purposes, though, a very simple version will suffice.11
In this model an individual is regarded as living for two periods ("working years" and "retirement"). In the
first period he works and consumes, and in the second period he consumes as a retiree the proceeds of his
first-period savings. With a tax assessed at rate tr on the yield from savings and at rate tw on wage income, the
budget constraint of our typical individual is
where Ci is consumption in life-period i; L, labor supplied in the first period; r and w, interest and wage rates,
respectively.
This budget constraint reflects a flat-rate income tax if tr = tw, a pure wage tax if tr = 0. A consumption tax at
flat rate tc would imply the budget constraint.
129
130

Page 153
Dividing both sides of (1) by (1

- tw), it is immediately seen (and
this generalizes in more
complex models; cf. Anthony
Atkinson and Joseph Stiglitz [5,
1980]) that for given w and r a
pure flat-rate wage tax system is
equivalent from the individual
point of view to a pure
consumption tax at flat rate tc =
tw/(1 - tw). Individual
consumption possibilities are the
same under both schemes. It
may be readily verified that the
tax liabilities are also the same
in present value terms, although
the path of cash flow to the
government generally will
differ. We return to this point
below.
Using this model we can see clearly why it is not sufficient to observe that there is a divergence between the
yield on investment and the rate of return received by savers to conclude that the taxation of income from
capital is undesirable. With no restrictions on the use of resources, efficiency requires that the production
trade-offs between consumption in the two periods and between leisure and each of the two types of
consumption equal the corresponding marginal rates of substitution in individual preferences. These
efficiency conditions will result from competitive forces in the absence of taxes. But when taxes must be
nonzero to raise revenue some or all of these desirable equivalences must be violated. In particular, "second
best optimality"doing the best possible when some of the conditions of efficiency must necessarily be
violatedmay imply some taxation of the returns to saving. In other words, in view of the necessity to have
some distortions in order to raise any revenue, a spread between the return on investment and the reward to
savers may be called for.
Note that this is not a point connected

with distributional considerations; the
issue is one of efficiency. This may
be illustrated by considering the case
of a single individual (or a population
of identical individuals) in a system in
which the productivity of labor and
capital are constant and equal to the
(gross of tax) wage and interest rate,
and the government revenue
constraint is fixed in terms of present
value. To emphasize the use of
standard demand analytical tools, let
us assume the individual has available
a total of one unit of time in the first
period, of which he retains a fraction
H for the non-market use
130
131
conventionally called "leisure" or

"home production." In analyzing the
second best problem we follow the
derivation of Atkinson and Agnar
Sandmo [3, 1977], adopting as well
their computation-easing device of
representing the tax on interest as an
equivalent tax, t2, on the
second-period consumption. The
individual is then viewed as
maximizing a utility function U(C1,
C2, H) subject
Â
Page 154
to the budget constraint (1), rewritten as
where
This problem has as parameters t2 and tw. Let C1(tw, t2), C2(tw, t2), and H(tw, t2) be the quantities of first and
second period consumption and home production chosen under this constraint and let V(tw, t2) be the indirect
utility function, that is, the utility level attained at the solution to the problem for each pair of tax rates. It is
readily verified that the loss in utility per unit increase in the tax on labor earnings is proportional to the
amount earned, while the loss per unit increase in the tax on second-period consumption is proportional to the
present value of that consumption. That is:
Here β is the "marginal utility of income," that is, the increment in utility that could be obtained if the budget
constraint (3) were relaxed by the addition of one unit (properly of wealth, not income). (Here and in what
follows, subscript i on a function designates the partial derivative with respect to its ith argument.)
The government's efficiency problem is to choose tw and t2 to make V as large as possible and still collect
revenue of a given discounted value, PVR(tw, t2), per taxpayer:
Ri represents tax receipts in life period i collected from our typical individual, and G is the fixed revenue
constraint. Wage tax receipts occur in the first period,
second period receipts are
131
132
Â
Page 155
A solution to this problem involves equality between the trade-off of tw and tr which leaves V constant and the
trade-off which leaves PVR constant. That is to say, a necessary condition of second-best optimality is
Using equations (4) to (7) we can reduce this condition to an expression in terms of demand functions and
their slopes. A rather direct substitution leads to the following version of (8):
where YL(tw, t2) represents before-tax labor earnings. Rearranging terms in (9), we obtain the following
condition on and , the second-best optimal values of t and t :
w 2
All of the terms in (10) are easily given an operational interpretation and it is immediately clear that in
general both tw, and t2 will be nonzero in a second-best optimum. Thus, because of the government revenue
requirement (5), both the present versus future consumption and the leisure versus first-period consumption
choices will be distorted even when both taxes may be varied freely. 13
For the second-best optimum to involve zero taxation of the returns to saving it is necessary for the
coefficient of on the left-hand side of (10) to be zero. Hence, a second point immediately apparent is that
the key sensitivity affecting this issue is the responsiveness of labor supply, not savings, to the tax rates.
Loosely speaking, (10) implies the second-best t2 is 0 when the supply of labor is roughly equally sensitive to
both the labor and capital taxes.
A great deal of attention has been paid to the interest elasticity of savings and the interpretation of the record
has been a matter of considerable controversy.14 Yet (10) suggests that this property of the economic system
taken by itself tells us little about whether the rate of taxation of capital should be high or low. The quantity
C2/(1 + r) will just equal savings in the absence of taxes. The sensitivity of savings to taxes thus appears in
the right-hand side of (10). A highly positive response of savings to the rate of return, which would
Â
132
133
Page 156
correspond to a large negative value of , does not per se determine the size or even the sign of t in the
second-best optimum. 2
To illustrate, consider the case of Cobb-Douglas preferences (11), an assumption of particular interest
because of its frequent application in the life cycle savings literature: 15
These preferences have the well-known implication that a constant proportion of the consumer's budget (in
this case, his wealth, including the market value of time reserved for home use) will be spent on each
commodity. Referring to the individual budget constraint (3), we can write down the demand functions (12),
The last expression in (12), the demand for home production time (one minus the labor supply), indicates that
income and substitution effects so balance as to cause the labor supply to be wholly independent of the prices
in the system. Hence it follows immediately that, for these preferences, . From (10) the
second-best optimal tax rate on capital income is zero, with revenue raised through wage taxation alone.
Notice that this result holds in spite of the zero elasticity of savings with respect to rate of interest, as seen in
the demand function for first-period consumption in (12).
In [28, 1978] Feldstein has emphasized this possibility. As he points out, the result for the Cobb-Douglas
preferences is an instance of the general proposition due to Atkinson and Stiglitz [4, 1976] that a tax on labor
alone will suffice for optimality (in the sense used here) whenever the individual's utility function is separable
between leisure and other commodities. This condition holds when utility can be written as the sum of a
function of leisure and a function of the remaining commodities, as is true for Cobb-Douglas preferences.
The views are often expressed that (1) the distortion in investment incentives created by a "tax wedge"
between true yield and the return to investors is ipso facto ground for concluding the taxation of income from
capital is too high, but that (2) in any case this "wedge"
Â
Page 157
doesn't matter much if (as several students of the historical record have concluded) the rate of saving is
insensitive to the rate of return. Even our simple model suffices to show that neither of these propositions
holds in general.
III
Structural Tax Policy and Capital Formation
Related to the common view that the taxation of returns to investment implies an inefficiently small quantity
of investment is the notion that a shift in taxes from capital income to labor earnings will necessarily cause an
increase in aggregate accumulation of wealth. That this notion is similarly incorrect is well illustrated by the
special case of Cobb-Douglas preferences.
133
134
In a world populated by our model life cycle savers, the aggregate privately held wealth will consist of the
difference between labor earnings and first-period consumption of the young generation. Letting W(tw, t2)
denote this wealth accumulation by an individual, we have
A government budget constraint of some form will determine a relationship between tw and t2, and it is clear
from (13) that increasing tw, to compensate for a decrease in t2 may, depending upon individual preferences,
lead to a reduction in W. A labor income tax rate of 100 percent would obviously leave nothing for either
consumption or private capital accumulation. But as the Cobb-Douglas example indicates, one need not look
only to such pathological cases to find the possibility that reducing taxes on the returns to savings and raising
taxes on labor earnings will lower capital formation. From (12) we see that with Cobb-Douglas preferences
(13) becomes
Private wealth accumulation in this case, which would not be regarded as particularly implausible by those
who employ life cycle models of savings, is wholly independent of t2, the tax rate on second-period
consumption, and directly and negatively related to tw, the tax rate on wage income. The Cobb-Douglas
preferences give rise here to a response to taxes that might be described as "Keynesian," in that only income
effects are evident. Extra labor tax revenues are drawn in fixed proportions from first-period consumption
and
Â
Page 158
savings, while all of the receipts of interest income taxes come out of consumption, in interesting contrast to
the Kaldorian view in which capitalist income is wholly devoted to accumulation. Thus under quite
reasonable behavioral assumptions a shift from capital to labor taxes can result in a reduction in private
wealth.
The Importance of Taking into Account Government Saving
In the discussion of accumulation above attention was focused on private saving and wealth. The earlier
analysis of efficiency, however, implicitly allowed for government saving and dissaving by invoking a
budget constraint cast in present-value terms. To model government saving introduce a pattern of real
government exhaustive expenditure of G1 per young person and G2 per old person. Constraint (5) now
becomes
Constraint (15) is readily seen to be equivalent to (5) with G1 + (G2/(1 + r)) = G, so that this reformulation has
no effect on the efficiency analysis.
This form of the budget constraint draws attention to the fact that a cash budget surplus with respect to an
individual in early life can be invested (potentially in a reduction of outstanding government debt) to help
cover outlays in later life. The term R1 - G1 in (15) represents government wealth accumulation with respect
to an individual. The total wealth, WT, accumulated either directly by a representative young person or
indirectly on his behalf by the government is then given by
134
135
and the effect of tax policy on total wealth, private plus government, per young person, is determined by the
properties of the labor supply function, (1 - H(tw, t2)), and the first-period consumption demand function,
C1(tw, t2). In the case of Cobb-Douglas preferences (11), (16) becomes
Equation (17) tells us that as we shift from capital to wage taxation as a method of finance, overall wealth
accumulation increases, when
Â
Page 159
account is taken of government saving, even while private wealth is reduced.
This case illustrates sharply the necessity of considering the entire budget framework in assessing the effect
of tax policy on capital accumulation. Thus a policy which might appear from the earlier analysis to be one
which would encourage capital formation may have the opposite effect if choice is made from among those
combinations of taxes sufficient to assure a flow of revenues of specified discounted value (per person).
There is a useful reminder in this of the limitations of cash flow accounting for government activities. Where
a change in the rules involves a change in the timing of tax payments (or government outlays) within the life
cycle, preserving a given tax burden per individual may require running a cash flow budget surplus or deficit.
In this example, a shift from t2 to tw implies a government surplus with respect to young people if a fixed tax
burden is to be maintained.
We remarked above that there is an exact equivalence between a flat rate wage tax and a flat rate
consumption tax in that for an appropriate choice of rates both the individual budget constraint and the
present value of tax revenues are the same. It may be instructive to verify the generalization that there is a
similar equivalence between a change in tw and appropriate changes in t2 and t1, the latter being the tax rate on
first period consumption. With the three taxes the individual budget constraint is
If we multiply both sides of (18) by we obtain the budget constraint which would result from
labor tax rate and consumption tax rates, and , where
Since the budget constraints under the two sets of taxes can be obtained from one another by multiplying
through by a positive constant, the individual's alternatives are precisely the same, as will
Â
135
136
Page 160
be his choice of labor effort and timing of consumption. To verify that the revenue constraint is equally
satisfied by both sets of tax rates, let PVR denote the present value of revenues from all three taxes over the
life of a young person,
Now note that the individual budget constraint (18) requires that the sum of the present value of the
consumption stream (including the value of the ''consumption" of leisure) and the present value of taxes paid
equal a constant, the value of the time endowment:
Since the hypothetical tax change does not change the budget constraint (and hence does not change the
choice made), it does not change PVR.
Now consider the path of wealth accumulation under various tax regimes that are equivalent in the sense just
described. Our individual variables H, C1, and C2 should now be imagined as functions of the three tax rates,
tw, t1, and t2, but the form of expression (13) for private wealth accumulation is unaffected. Since H and C1
will not change under the tax alternatives we are considering, we see from (13) that private wealth
accumulation will change, decreasing with increases in tw. But from (16) we see that the growth of
government wealth accumulation exactly offsets the private change. Aggregate wealth accumulation is
independent of the choice among combinations of tax rates with the equivalence just described. This includes
as a special case the class of combinations of flat labor and flat consumption taxes (t1 = t2 = tc).
The Effect of a Cash Flow Revenue Constraint
Note, however, that to exploit this equivalence the government must be sensitive to the requirement for
budgetary surplus (in the case of a wage tax) or perhaps deficit (in the case of a consumption tax). It is
sometimes suggested that a cash flow budget balance requirement is the appropriate constraint on tax policy.
While this analysis indicates that such a principle makes no economic sense, it may be politically appealing
and we may briefly consider its implications.
Â
Page 161
In the model we have been using thus far, essentially a one-person world, a cash flow constraint fixes t2, and
the alternatives available vary along the spectrum from labor tax to capital tax (and beyond), assuming the
possibility of negative rates (subsidies) as we vary tw and t1 to obtain the revenue required in the first life
period. A more interesting set of issues is raised if we introduce a population age structure.
For this purpose we may

conveniently adopt Samuelson's
[57, 1958] original assumption
of steady population growth, so
that each new generation is (1 +
n) times as numerous as the
preceding one. With this
assumption, tax revenues per
136
137
young person in a system

steadily reproducing itself will
be given by
It will be seen that this constraint on the choice of tax rates differs from (20) in, in effect, discounting second
life period taxes at rate n instead of rate r. In view of this difference, unless n = r, the behavior of the system
will not be independent of the choice among tax regimes equivalent in their impact on the individual budget
constraint. 16
Consider, for example, what happens as we shift from a wage tax to a consumption tax. As we know, if tax
burdens are held constant in present value terms this change in tax structure will not change consumption
patterns or aggregate wealth accumulation, although it will affect the ownership of the wealth, since it will
alter the government budget surplus. Under the cash flow constraint the government is not allowed to run a
surplus (or deficit). The wealth accumulation effects are all through private savings, as described by (13),
suitably modified to incorporate the tax on first period consumption. Specifically, in this case the household's
choice of consumption sequence and labor supply can be written as functions of the two tax rates, tw and tc(=
t1 = t2), so (13) becomes
As we raise tc we generally can lower tw by an amount determined by revenue constraint (22). Depending
upon the trade-off permitted
Â
Page 162
by the budget (which will depend on n) and the responsiveness of the demand functions to the tax rates,
wealth accumulation may be either increased or decreased.
Developing a point made by Hall [46, 1968], Summers [63, 1978] stresses these effects in a recent paper in
which he analyzes tax policy in a life cycle model. Summers's model is more realistic than ours in allowing
for many life-periods, with earnings from labor in later as well as earlier periods, although it is less realistic
in incorporating no choice of labor supply. While Summers's principal focus is on the welfare gains
obtainable in his model from eliminating the tax on capital income, he devotes considerable attention to
differences in the two ways of doing this, in our terms via tc or via tw. These differences are of two sorts. One
concerns the incidence of a change in tax regimes. A sudden shift from capital to consumption taxation
penalizes (relatively) those who have accumulatedprimarily those at or near retirementwhile a shift to wage
taxation penalizes those in the early phases of accumulation. Summers makes the important point that it may
be possible to use a combination of two taxes to generate an improvement for everyone through a
combination of the two taxes, if the shift generates sufficient efficiency gains.
The second difference between tc and tw as replacements for tr is in the effect on capital formation. The tax
alternatives considered by Summers are confined to those maintaining cash flow government budget balance.
Under this constraint, the equilibrium wealth stock is found to be inversely related to the use of the wage tax.
Although it involves more periods, Summers's model assumes Cobb-Douglas preferences, and we can make
use of our earlier analysis of wealth accumulation with these preferences to understand his conclusion about
the difference between tw and tc.
137
138
Cobb-Douglas preferences lead to a division of the discounted value of after-wage-tax receipts into outlays
on consumption gross of tax in the various periods according to the exponents on the consumption terms in
the utility function, α1 and α2 in (11). A proportional change in net wage receipts leads to a proportional
contraction of outlays for consumption. Since private wealth accumulation is built up from the differences
between net wage receipts and consumption outlays, a decrease in (1- tw,) causes a proportional decrease in
aggregate wealth (assuming it is positive to begin with). A change in tc, on the other hand, changes the
consumption received for a given gross-of-tax outlay, but, because of the Cobb-Douglas
Â
Page 163
preferences, it has no effect on the path of savings. Wealth accumulation depends only on tw in the
Cobb-Douglas case, for which (14) is the explicit form of (23).
As this analysis of a simplified version of Summers's model suggests, his particular results, while useful, are
strongly dependent both upon the details of the preference structure and on the government cash flow
constraint. The first dependence means one must draw empirical conclusions with considerable caution, since
the preferences assumed, while typical for life cycle models, are rather special. Perhaps more important is
recognizing the crucial effect of the cash flow restriction. In models such as Summers's, structural tax policy
(for example, the choice between a consumption or income tax) has a bearing on the determination of
aggregate wealth only because of the implicity accepted limit on government saving. Without such a limit
(which throws away an important policy instrument) fiscal policy is able to influence aggregate wealth
according to whatever may be the social objectives while structural tax policy is chosen on other criteria.
Certainly as far as the simple life cycle story is concerned, those critics of the consumption base ideal who
say that its claimed capital accumulation effects can be achieved through government saving without
structural change are correct.
By operating the fiscal system so as to collect more in taxes (in a present value sense) than is paid out in
current exhaustive expenditure per taxpayer the government can accumulate additional wealth, while by
operating with a regular deficit in this sense the government can decumulate wealth. The effect of this on the
aggregate wealth of the economy will depend upon the reaction of private savings. If private capital
formation is reduced by less than one dollar for each dollar's worth of government capital formation, budget
surpluses will lead to net additions to aggregate wealth, deficits to net reductions.
Structural Tax Policy and Intergenerational Links
An important issue bearing on the reaction of private savers to government accumulation is the extent and
nature of preferences about intergenerational transfersgifts and bequests. The model savers we have been
considering thus far apply all proceeds of work and saving remaining after taxes to own consumption. We
can obtain a sense for the reaction of these savers to government wealth accumulation
Â
138
139
Page 164
by reference to expression (13) for total wealth (per young person), and expression (14), corresponding to the
Cobb-Douglas special case. If the relationship of tw, to t2 is established, for example, by efficiency
considerations such as were expressed in (10), marginal adjustments in the overall level of taxation will be
made by proportional changes in these rates. The effect of such changes on aggregate wealth will be
determined by the combination of labor supply effects (expressed in H(tw, t2)) and first-period consumption
effects (expressed in C1(tw, t2)). For the special Cobb-Douglas case a unit increase in both tax rates leads to an
increase in aggregate wealth accumulation of α1 tw w; intuitively, all of the extra tax collected on the
price-insensitively supplied labor adds to the aggregate wealth.
Because our model individuals are concerned only with own consumption, government accumulation affects
them only via tax rates within their lifetimes. When the generations are linked by gifts and bequests,
however, the entire future chain of consequences of government accumulation bears on present individuals'
decisions, and we can no longer regard their consumption choices as dependent only on the tax rates.
Analytically, the linking of generations has roughly the effect of postulating an "immortal" individual
expressing preferences over consumption and labor supply in every period. In that setting an increase in
wealth accumulation by the government substitutes in some measure for private wealth accumulation
motivated by a desire to benefit future generations. Similarly, an extra unit of government debt is treated, in
effect, as a liability in the balance sheets of existing taxpayers, because of the implied obligation of future
taxpayers to redeem that debt.
The degree to which government accumulation substitutes for private accumulation is an issue with a long
history in the economics literature. Its bearing on the question of whether the burden of government
expenditure can be shifted toward or away from future generations is by now widely recognized. Directly
related is whether the effect of government spending on aggregate demand differs according to whether it is
tax or debt financed. Here we stress whether the government accumulation is regarded by individual
accumulators in the aggregate as different from private wealth. Under strong conditions of intergenerational
links by voluntary transfer the answer to all three questions is no.
Tobin and Willem Buiter [64, 1980] have presented an excellent overview of the literature on this subject and
in [17, 1979] they re-
Â
Page 165
view the empirical evidence bearing on the issue. 17 For us the principle question is whether the choice of
structural tax policy needs to be concerned with wealth accumulation effects. Tobin and Buiter conclude that
the conditions for the "debt neutrality" hypothesis, of a fully offsetting implicit liability in taxpayer balance
sheets for an increase in government debt, are implausibly stringent, and that the evidence does not support it.
As they point out, however, the hypothesis of debt neutrality cannot be rejected on the basis of their
econometric analysis. My conclusion is that, while this must be regarded as an important open issue, fiscal
policy recommendations should be based on an assumption of significant nonneutrality. This implies that the
overall fiscal balance, budget deficit or surplus, is a crucial instrument in determining aggregate real capital
accumulation.
Whether it is the crucial instrument is a question that can only be addressed rigorously in a carefully specified
optimizing model. Yet we can see that the answer is yes in the case in which the individual ignores
government wealth accumulation. The issue is then wholly one of intergenerational distribution, and this is
captured by net government accumulation. The question of optimal tax structure can then be cast just as we
have done in Section II, except that the requirement (15) that the present value of taxes collected with respect
139
140
to an individual equal the present value of government expenditures on his behalf is replaced by a net
accumulation or decumulation requirement. The wealth accumulation objective enters the problem in what
may be thought of as a lump-sum way, with the result as we have suggested that structural tax policythe
relationship among the various tax rates in the systemis divorced from capital formation issues.
Qualifications and Conclusions
Before we turn to a discussion of some specific tax structure issues, three remarks are in order concerning the
interpretation of measured wealth (and saving). First, one should not overlook the fact that the stock of
measured wealth held by households, directly or indirectly via financial intermediaries, includes the sum of
government debt and real capital, and is thus larger than the economy's stock of real claims on future output
when government debt is positive (and would be smaller than it if government were a net creditor, for
Â
Page 166
example, by holding a portfolio of corporate bonds). Intuitively, the aggregate wealth of the citizenry as one
would normally measure it is different from the true collective wealth. As we have seen, whether it is the
measured wealth or the underlying "true" wealth that represents the aggregate of what individuals perceive as
wealth depends upon one's conclusion on the debt-neutrality issue.
A second distinction important in policy analysis is between measured wealth, which includes claims on
foreign individuals and institutions, including foreign governments, and the stock of capital with which U.S.
labor cooperates in domestic production, which may be owned by residents of other countries. 18 The
importance of this emerges most sharply in a model of a small open economy in which returns to foreign
owners of domestic capital are not taxed. In this case the size of the capital stock combined with labor is
wholly determined by the world interest rate, and taxation of domestic wealth-holders is of absolutely no
relevance for the determination of the domestic capital stock, much as it may have to do with the domestic
wealth stock. If domestic savers decide to accumulate less, the world capital market will supply sufficiently
more to keep the rate of return on domestic investment equal to the world interest rate, with suitable
allowance for risk.
If domestic capital stock is the subject of policy concern, it is necessary to attend specifically to this feature
of an open economy. This idea has not been wholly absent from U.S. policy determination. The investment
tax credit, for example, applies only to the acquisition of eligible property for domestic use. This rule
illustrates as well the way in which interaction of different parts of the tax system may produce surprising
results. Because U.S. income tax is creditable against the home income tax for most foreign companies, the
investment credit (which offsets a tax they in effect don't pay) may be of little value to them. In this case the
investment credit may have no effect on the size of the domestic capital stock, but merely an effect on its
ownership, moving it out of the hands of foreigners and into the hands of U.S. residents. In order to keep the
discussion within bounds, international capital flows will not be given further explicit attention in this
chapter, although the example suggests the issue may deserve future study.
The third distinction deserving mention is that between measured wealth and wealth including the value of
claims on future payments not embodied in existing pieces of paper. An example is the element
Â
140
141
Page 167
of unacknowledged government debt in the anticipated payments of retirement benefits under Social
Security. 19 For our purposes, the point to stress is that the net wealth accumulation effect of tax structure
choices is not in general correctly captured by current cash flow deficit or surplus. A rule change which
involved reduction of taxes early in the life cycle, and increases equal in present value a bit later in the life
cycle, would cause a cash budget deficit, even though it would not affect the wealth position of existing
taxpayers. (The fact that the increased taxes will occur within the lifetime of existing taxpayers is what
distinguishes this from the standard debt-neutrality situation.)
In this section and the preceding one I have tried to lay out the basic economics of tax policy toward savings
in a way which corrects oversimplifications often encountered. The analysis remains, I hope, sufficiently
simple to allow debate and study of the key points on which policy turns. With this as background, let us turn
to a look at some of the details of tax institutions through which the treatment of savings is determined.
IV
Savings in the Individual Income Tax: Problems of the Half-Way House
Not much space in the U.S. internal revenue code is devoted to tw and tr. Instead the portion of the statute
concerned with the income tax specifies tax schedules to be applied to a tax base (called "income")
attributable to individuals and to such legal entities as corporations and trusts, and provides an elaborate
system of rules defining that base. In this section we are concerned with how this actual system relates to the
concepts considered above. It will simplify matters to think of this section as applying to the taxation of
individuals only. Although much of the analysis will carry over to the subsequent examination of the
corporation income tax, the latter introduces new problems which are more easily treated separately.
As we have seen, the simple economic models are well suited for straightening out our thinking and even for
drawing some conclusions about the desirable directions of policy, but they are not readily related to actual
tax institutions. In the model used in Section II, for example, the choice between income and consumption as
a tax base is simply between setting tw = tr and setting tr (or t2) = 0. Furthermore, there is no need to choose
between these two options. We
Â
Page 168
can set the flat rates tw and t2 at any level. Since all the transactions involved (sales of leisure, purchases of
future consumption) are assumed easily observable on markets of a conventional sort, assessing tax liabilities
is a simple matter. In actuality, however, the implementation of a system to achieve with any precision the
effect of any given rates of tax, tw and t2 is very difficult. And the problems multiply when the objective is not
to assess the same flat rate taxes on all individuals, but rather to tailor tax burdens to individual
circumstances.
In the actual system, instead of payments of wages and interest being the only and readily monitored forms of
compensation for the services of labor and savings we have a complex array of rewards, including such
elements as pension rights, in-kind benefits of various types, accruing changes in value of assets ranging from
common stock to the inventories of proprietorships, and so on. These do not separate themselves into rewards
to labor and rewards to saving. This in itself is not a problem if one has concluded that the objective of tax
policy should be taxation of periodic Haig-Simons income (the sum of consumption and the change in net
worth that has accrued over the accounting period, and hence, neglecting transfers, and aggregate of factor
rewards). It is a problem if one wishes to tax the returns to saving more or less heavily than labor rewards.
We cannot accurately separate returns to labor and wealth and even measuring the aggregate of these two
141
142
presents severe difficulties.
The fundamental difficulty is measuring changes in wealth on an accrual basis, that is, changes that need not
be directly revealed by current market transactions. This point has been developed at length by, among
others, Andrews [1, 1974], Blueprints, and Bradford [14, 1980], and we need not repeat these expositions.
Important for us here is that under present rules some elements of accruing increases in wealth go untaxed for
long periods (as with accruing capital gains) while in other cases deductions designed to approximate
accruing wealth reductions (the principal example being depreciation allowances) overstate the true amounts.
These two possibilities correspond to the two approaches available for exempting the returns to saving from
tax, that is, for implementing a consumption base. To do this the basic accounting technique takes one of two
forms: the "standard treatment" allows a deduction from an otherwise familiar income base for saving (and
inclusion of the corresponding dissaving); the "alternative treatment" simply ignores both the act of
Â
Page 169
saving and the return flow it generates. (In terms of our simple model, the first method corresponds to a
common rate of tax, tc, on consumption in the two periods, while the second corresponds to tw.)
Added to the elements of consumption tax accounting that are virtually inevitable in a practical income tax
are many devices more or less explicitly designed to encourage savings in one form or another. These
measures generally amount to allowing consumption tax accounting for selected transactions. Thus, for
example, the standard consumption treatment is accorded pension saving undertaken by an employer on
behalf of an employee, and a partial form of standard treatment is given to investment eligible for accelerated
depreciation or the investment tax credit. These rules have the effect of providing a deduction for investment,
to be followed by subsequent inclusion of return flows in the tax base. The alternative consumption treatment
is also found, as in the case of state and local bond interest or, in partial form, long-term capital gains. Full
alternative treatment of long-term gains would involve complete exclusion from the base. Current rules for
individuals provide deferral of tax until the gain is "recognized," normally the time of sale or exchange of the
asset, with 60 percent of the gain excluded from tax at that time. Thus we frequently find side by side in U.S.
law income tax and consumption tax treatments of economically very similar transactions. The resulting half
way house has been called a "hybrid" system by Andrews [1, 1974], and as he points out the availability of
both treatments accounts for a whole array of nettlesome problems. These are revealed most sharply in the
case of leveraged tax shelters.
The Paradigm of the Leveraged Tax Shelter
The essence of a leveraged

shelter (insofar as it does not
involve some sort of outright
sham) is to finance an
investment eligible for
consumption tax treatment by
borrowing under income tax
rules. A good example of such a
procedure would be borrowing
to purchase tax exempt bonds.
The deductible interest payment
is shared by the government
while the return is not. If the
yields were the same there
would be a net profit to the
142
143
taxpayer, in principle with no

investment whatsoever (the
borrowing being secured by the
tax exempt bonds). The interest
deduction "shelters" otherwise
taxable income, while the yield
it finances goes free of tax.
Â
Page 170
Both the economic and political reaction to this situation in fact are characteristic. The economic response
manifested in the market place is a reduced interest rate on the tax exempt bonds, so that they are attractive
only to taxpayers with marginal tax rates in excess of the tax exempt-taxable interest differential. (This has
the effectalso characteristic of sheltersof reducing the progressivity of the income tax.) The political reaction
is on the one hand to limit the supply of tax exempt bonds (although local governments are ingenious at
finding ways to issue eligible securities; for an instance of complexity of rules see the regulations governing
the ''arbitrage bonds" used by state and local governments in refunding their tax-exempt debt) and on the
other to write rules circumscribing the deductibility of interest payments. In this case, the rules disallow
interest deductions on debt incurred to purchase or hold tax-exempt bonds, a criterion that presents obvious
problems of interpretation. Implementation of this policy requires "tracing" rules, associating particular
liabilities with the tax exempt bonds on the asset side of the taxpayer's balance sheet.
The transactions giving rise to tax shelters typically involve not fully tax-exempt return on investment but
rather a combination of partial write-off of the cost of the investment and favorably taxed return, generally
via the long-term capital gains device. Thus the antishelter rules built into the tax code by the Tax Reform
Act of 1976 (subsequently somewhat simplified by the Revenue Act of 1978) were directed at particular
activities (cattle raising, film producing, and so on) in which depreciation allowances are faster than
economic calculus would imply (the evidence, indeed, is their very attraction as shelters) and in which the
return is typically converted to capital gains. By accounting for the two parts of a leveraged transaction in
very different waysthe borrowing according to income rules and the asset purchase according, roughly, to
consumption tax rulesthe result is not merely that income goes untaxed. Rather, a form of arbitrage profit
through the tax system is provided that may even make attractive investments with negative social return.
As in the example of tax exempt bonds financed by borrowing with tax deductible interest, what is involved
here is not simply lightly taxing the returns to saving. In our extreme case there is no saving by the taxpayer
at all, simply a choice of appropriate composition of a portfolio with net market value zero. The taxpayer
seeks
Â
Page 171
to equate the returns at all margins of portfolio choice (with appropriate allowance for risk). Neglecting risk,
this implies a rate of return on saving equal to the general market interest rate times one minus the saver's
marginal tax rate after taking into account the effect of sheltering. The shelter transaction may generate
infra-marginal gain for the taxpayer, but it may lead to no incentive for saving at the margin. The tax
arbitrage profit merely supports a lower social return on the "favored" assets and the transactions costs (often
significant) involved.
143
144
As in the illustrative case of tax-exempt bonds, the arbitrage process would go on indefinitely were it not for
the limited supply of the asset eligible for the consumption tax treatment, the nonlinearity of the income tax
rate structure (so that the rate of tax saving per dollar of deduction falls while the tax per dollar on the return
flow rises as more of the tax arbitrage is undertaken) and, the political response, special rules written into the
law to limit the manipulation of portfolios for this purpose. This is done either by restricting the income tax
treatment of the borrowing side (for example, putting conditions on the interest deduction) or the
consumption tax treatment of the asset purchase side (for example, restricting the netting out of capital gains
against depreciation deductions).
Even without attempting to take into account the special rules that have been enacted to restrict the extent of
income sheltering, it is difficult to predict the net effect of the hybrid rules on saving and the allocation of
capital. In the next section we look somewhat formally at a series of characteristic features, with surprising
implications in some cases.
V
Economic
Analysis of the
Hybrid System
By studying simple models of the hybrid rules, controlling for various elements in turn, we can learn about
the way they work through a market economy. To start with, consider a system in which there are two real
investment sectors, ℜϒ and C. Assume initially income from investment is calculated according to good
Haig-Simons principles in both sectors, and assume an income tax is levied at the flat rate of 40 percent
(including refunds for negative income) on all taxpayers. With an interest rate of 10 percent, the after-tax
yield on savings to all taxpayers is 6 percent. Because interest is deductible, the yield on real investment in
both sectors will be 10 percent.
Â
Page 172
Now imagine a decision is made to change the rules for calculating the income subject to tax in sector C.
(This might be because the rules are easier to administer, or to encourage investment in sector C, or for some
other reason.) In place of Haig-Simons income rules, the consumption base practice of immediate write-off of
investment outlay is adopted. To compensate for the revenue cost, the tax rate is raised to 50 percent for all
taxpayers. The original allocation of resources will no longer be compatible with equilibrium. While the
interest rate and rate of return are equal in both sectors, the mixed income and consumption tax rules present
an opportunity for tax arbitrage.
This could be exploited in any number of ways. The taxpayer might, for example, borrow $50 and add to this
the tax reduction (or refund) of $50 to finance a $100 investment in sector C at no out-of-pocket cost. From
the annual gross return of $10 the interest charge of $5.00 would be payable, as would tax on the net of
interest return of $5.00. There would remain $2.50 annually, representing a pure surplus, inasmuch as the
taxpayer has sacrificed no consumption or other asset acquisition to obtain it.
In response to the changed rules we would expect resources to flow into sector C out of sector ℜϒ, and the
interest rate to change, until no further tax arbitrage is possible. The outcome might, for example, involve an
interest rate and rate of return of 12 percent in sector ℜϒ and a rate of return of 6 percent in sector C. Our
investor who borrows $50 at 12 percent and adds his tax rebate of $50 to purchase a $100 machine in sector
C obtains a gross income of $6.00, just sufficient to cover his interest payments on the loan, leaving him with
no increased tax liability and no net cash flow.
144
145
The analysis even at this stage establishes several useful points. First, the "tax break" to investors in sector C
does not imply a higher rate of return to such investors in equilibrium. Indeed, it is the essence of the
equilibrium condition that the rate of return received by investors must be the same in all sectorsotherwise
there would be a shift in investment flows. Secondly, a tax break to investors in one sector of the economy
may not at all increase the rate of return received by them or others after adjustment has taken place, taking
into account the necessity of raising tax rates to obtain the required revenue. In our example, both before and
after the rule change investors obtain a net return of 6 percent, representing a 40 percent tax on a 10 percent
interest rate in the initial situation and a 50 percent
Â
Page 173
tax on a 12 percent interest rate in the new situation. These illustrative figures happen just to balance. It is
possible to construct fully spelled out models in which the net return to savers is either increased or
decreased by a tax break of the type described. 20
Third, if there is a policy reason to wish for a shift of resources into sector C, there is nothing objectionable
about the method of giving a "tax preference" in our flat-rate tax world. It is to be emphasized that even
though our hypothetical investor receives an "artificial" deduction of $100 for his $100 investment financed
entirely from borrowed money and the tax refund, there is no real reasonother than a cosmetic oneto be
concerned. There is no reason to treat the borrower any differently from the person who puts his "own money
at risk." The horrors of tax arbitrage are a phenomenon of disequilibrium, and the possibility for tax arbitrage
will only strengthen the speed of response toward the desired subsidy of sector C investment. Even if the
sector C subsidy is not the explicit policy objective, but merely a by-product of some administrative necessity
(as is the case usually argued for allowing certain farm enterprises to use cash accounting methods) the tax
arbitrage gain vanishes in equilibrium so that the policy concern is not equity across taxpayers, but efficiency
of resource allocation.
Achieving a Consumption Tax with Hybrid Rules
Consider next the consequences of taking the further step of allowing consumption-type accounting for real
investment in sector ℜϒ as well as C. For illustrative purposes, assume this requires an increase in tax rate to
60 percent in order to keep the economy on the desired path of aggregate accumulation. With neutrality
between sectors thus restored, we would expect resources to return from sector C to sector ℜϒ until the yields
from real investment are equalized. Let us suppose that the effect of the new rules is to expand aggregate
saving so that the new equilibrium yield is 9 percent. If the interest rate is also 9 percent, as it would be with
universal Haig-Simons accounting, the leveraged sheltering device will permit tax arbitrage. By borrowing
$40, for example, and financing the remainder of the purchase of a $100 machine from the tax saving due to
the $100 write-off, the taxpayer can generate a gross return of $9.00, out of which he must pay $3.60 interest
and $3.24 income tax (=($9.00 - $3.60) Ã 0.6). Such possibilities will be eliminated only when the interest
rate has
Â
145
146
Page 174
adjusted to 22.5 percent. With this condition taxpayers will obtain a yield of 9 percent whether they
accumulate financial assets accounted for on Haig-Simons income principles or real assets, accounted for on
consumption tax principles. These rules thus effect a consumption base, without acknowledging it.
To summarize the analysis of the flat-rate system, we have seen that the predictable effect of introducing
consumption tax principles with respect to real investment in a particular sector of the economy is
inefficiency in the allocation of capital. The effect on the rate of return on savings depends upon the
associated adjustment in the tax rates and the rates of return obtaining in the various sectors in the new
equilibrium. It is difficult to predict the signs or sizes of these effects, or to relate the equity properties of the
resulting system to the usual debates about the relative merits of income and consumption as alternative tax
bases. When the consumption principles are extended to all real investment in a flat-rate income tax system a
consumption base is effectively achieved, with its associated efficiency and equity properties, even though
taxpayers holding financial assets are observed to receive interest payments and pay "income tax" on them.
The rate of return received by savers in the new equilibrium, relative to that obtaining under pure income tax
rules, will depend on all those elements analyzed in Section II, but it is assured that it will equal the yield on
real investment.
Problems Due to Graduated Rates
The story is different with a graduated rate structure. Consider, for example, the case in which marginal
income tax rates vary from 0 to 70 percent, as is true in the United States. With true Haig-Simons income
measurement, the initial equilibrium will show tax-payers obtaining a yield on savings ranging from 3 to 10
percent, with indifference about portfolio composition among financial assets and real assets in both sectors.
Now introduce the consumption type accounting for real assets in sector C. (For the moment neglect the
required increase in tax rates to make up the lost revenue; the point here concerns the problems arising from
progressive rates.) It is clear that these new rules have absolutely no immediate effect on the prospects of a
taxpayer with a marginal rate of 0. For him the deduction of the outlay on a machine provides no tax benefit.
For the 70 percent taxpayer, on the other
Â
Page 175
hand, the payoff from the leveraged shelter is high. Borrowing $30 at 10 percent enables him to buy a $100
machine in sector C with no out-of-pocket expenses. The gross return of $10 covers $3.00 of interest and
$4.90 (= ($10.00 - $3.00) Ã 0.7) of tax, for a surplus of $2.10.
Presumably the same sort of reallocations as took place in the flat rate example will now occur. Suppose the
rate of return on real investment in C goes to 6 percent, that on real investment in ℜϒ to 12 percent, and the
interest rate to 12 percent. This leaves an opportunity for tax arbitrage by the 70 percent taxpayer. Borrowing
of $30 will still buy him a $70 tax reduction to help pay for a $100 machine in sector C. The return of $6.00
will cover interest of $3.60 and taxes of $1.68 for a surplus of 72 cents. To eliminate this possibility, the
process of shifting resources from ℜϒ to C must go further. 21 This arbitrage possibility is eliminated when
the rate of return on real investment in sector C is, say, 4 percent, and both the rate of interest and yield on
real investment in ℜϒ are 13.33 percent.
The effect on the change in rules with no increase in tax rates would be in this instance to raise the net yield
on saving for all taxpayers, although the result could be reversed by the required increase in tax rates. Note
the outcome, characteristic of such tax shelter situations, that ownership of the tax favored asset, capital in
sector C, is concentrated in the hands of the high-bracket taxpayers, who are thus observed to be the only
146
147
ones able "to afford" to take advantage of the tax loophole. Everyone else will be better off with financial
assets or capital in sector ℜϒ.
Consumption Tax Equivalent?
In contrast to the case of a universal flat rate of income tax, the application of consumption tax accounting to
all real investment does not produce the equivalent of a genuine consumption base in a graduated rate system.
In the model's equilibrium all saving of taxpayers below the top marginal rate bracket will be in the form of
financial assets (loans) while the top-bracket taxpayers have the equity interest in the entire capital stock.
This relationship between the yield on real capital and the interest rate will again be determined by the
no-arbitrage condition, which is simply that the high-bracket taxpayer be indifferent between fully taxed
interest and the effectively tax exempt real investment yield. Thus, for the highest-bracket
Â
Page 176
taxpayers the real investment yield and actual net-of-tax return on savings are the same. For all other
taxpayers the rate of return on savings exceeds the real investment return.
The implication of this simple model, that with consumption tax rules applied to real investment only all real
capital will be owned by the highest-bracket taxpayers, would presumably be qualified by attention to such
matters as portfolio diversification. Yet this feature of the model surely describes a genuine pressure existing
in the U.S. tax system, and one that would grow were steps taken greatly to extend the consumption tax
treatment of real investment (accelerated depreciation, investment tax credit, and so on) while leaving intact
the Haig-Simons income tax treatment of ordinary borrowing and lending. Furthermore, the presumption that
low-bracket bond-holders (including pension funds) now receive a rate of return (adjusted for risk) exceeding
the yield on most real investment would be strong were it not for the possibility that inflation has resulted in
sufficiently understated depreciation to offset the general bias in the other direction.
Two elements are missing from the story thus far. One is the political reaction to a situation in which
high-bracket taxpayers are able to shelter so much of their income by way of leveraging. This would be likely
to lead to rulessuch as the minimum tax on preference incometo restrict the extent to which individual
taxpayers can take advantage of the favorably taxed investment opportunities. The second missing element is
the endogenous change in marginal rates in the process of getting to the new equilibrium. As the high-bracket
taxpayers engage in tax arbitrage their success in sheltering income moves them into lower brackets, and this
will affect the equilibrium configuration.
Endogenous Marginal Rates
When we acknowledge that marginal tax rates depend upon the way taxpayers arrange their affairs we
introduce a further complication. As the high-bracket taxpayer shelters income by current deductions he
moves into lower brackets, but at the price of increased future taxes and, more to the point, higher future
marginal rates. We can analyze this by constructing a model in which all of the adjustment to eliminate tax
arbitrage opportunities takes the form of changes in marginal rates over the taxpayer's lifetime.
Â
147
148
Page 177
To illustrate, consider a taxpayer initially confronting a marginal tax rate of 50 percent in both the current and
the next period in a system of pure income taxation, with both the rate of interest and yield on real investment
of 10 percent. If we introduce consumption tax treatment of real investment there will be an opportunity for
tax arbitrage by borrowing to finance real investment. To this point we have considered the way a change in
the relationship between the interest rate and the real yield on investment can bring about equilibrium.
However, the reduction in marginal rate in the first period attributable to the deduction of the real investment
from the tax base, and the increase in second-period marginal rate due to the increased taxable income in the
second period can also remove the possibility for tax arbitrage. For example, if the deduction reduces the
marginal rate to 45 percent in the first period, and the increased income in the second period raises the
marginal rate to 58 percent, no further arbitrage is possible when the interest rate and the yield on investment
remain at 10 percent. 22
We can obtain some insight from a simple formal representation of a hybrid system with endogenous
marginal rates. In the two-period model introduced in Section II, let B stand for the amount borrowed (on an
income-accounting system for tax purposes), and S for the amount of "shelter" investment. Let Xi stand for
the "taxable" income in period i, and T(Xi) for the tax liability as a function of taxable income. Our model
individual must now choose C1, C2, H, B, and S to maximize U(C1, C2, H) subject to
Particular interest is attached to the trade-off between first- and second-period consumption at the point of
solution to this problem. One way to characterize this is by a "consumption possibility" frontier defined as the
locus of maximum C1 for each C2, given H:
Â
Page 178
The Lagrange multiplier of the constraint in this problem is the discount factor, -dC1/dC2, along the
consumption possibility frontier. It is readily verified that
If the shelter investment is undertaken in positive amounts the effect is to raise T'(X2) and thus, we see from
the first equation in (26), to lower the marginal return on savings. The version of the discount factor shown in
the second equation of (26) draws attention to the fact that marginal rates would have to be the same in both
periods to achieve the proper consumption tax result.
If we were to consider a multiple-period lifetime model, the problem of optimizing the illustrative taxpayer's
affairs would become more complicated, particularly if issues such as realization of capital gains, tax-exempt
148
149
interest, and so on were introduced. However, the possibility seems clear that endogenous determination of
the patterns of marginal tax rates over the life cycle may have a significant bearing on the effect of tax rules
on the reward to saving. In the numerical example, introducing the consumption tax treatment of real
investment reduces the after-tax rate of return on savings (the increase in second-period marginal rate from
50 percent to 58 percent drops the after-tax return from 5 percent to 4.2 percent), even without consideration
of the upward shift in the tax schedule one might expect to make up for lost revenues. The net effect of such a
shift in tax policy is made the more difficult to predict by consideration of the changes at margins other than
consumption timing, such as the timing and extent of labor force participation, brought about by the
endogenous effect on marginal tax rates.
Allowing Consumption Treatment of Financial Assets
The likelihood that marginal tax rates will play an important role in eliminating tax arbitrage opportunities is
increased by the presence in the tax system of an element of consumption treatment of financial assets. If this
were unrestricted, the effect would be to remove the constraint on the supply of the assets eligible for
consumption treatment that permitted equilibrium to be achieved through a differential between the rate of
interest and the rate of return on tax-favored assets. If the taxpayer has an unrestricted option to treat any
partic-
Â
Page 179
ular loan or borrowing on consumption tax principles, he can create a tax-favored asset in the form of a loan
to himself. The asset purchase (the loan made to himself) is given consumption treatmentthe outlay is
deducted from the tax base; all return flows are includedwhile the offsetting asset sale (the borrowing from
himself) is given income treatment. If the rates of tax on the deduction of the asset purchase and the
subsequent returns flow are the same, this produces a pure arbitrage profit, and it is only through tax rate
differentials that this possibility can be eliminated without introducing such refinements as portfolio
considerations, transactions costs, and so on.
The principal route by which financial assets obtain standard consumption tax treatment is via pension
savings. These are generally restricted in that additions to an individual's accumulation are permitted only up
to a limit related to labor earnings. There are, furthermore, legal inhibitions to pledging pension benefits as
loan collateral. Such constraints tend to limit the potential tax arbitrage and resulting adjustment of marginal
rates over the life cycle, although it is most unlikely they eliminate the effect.
Summing Up
The simultaneous existence in the individual income tax of rules treating certain forms of saving according to
a consumption ideal and other forms (notably borrowing) according to a Haig-Simons income ideal set up the
potential for profitable rearrangements of individual wealth portfolios with little or no inherent merit. These
profits are limited in part by the finiteness of the supply of assets accounted for under the consumption
model, and in part by nonlinearities and special rules which relate to the accounts of the individual (the floor
under itemized deductions, the investment interest deduction limit, limited netting of interest and long-term
gains, "at risk" rules and so on). Where the supply limit is the operative force, inefficiency in the allocation of
wealth results from bidding for the tax advantages; the social rate of return is lower in equilibrium in the
tax-favored activities than in those taxed by adequate income measurement rules. And because the tax
advantage is most valuable to those with the highest tax rates, assets accorded consumption-tax treatment
tend to migrate to the portfolios of high-bracket taxpayers. 23 After taking into account the required
compensation in tax
149
150

Page 180
rates to maintain revenues the net result, on incentives to save, on progressivity of the system, or on virtually
any other objective of broad appeal, is wholly unpredictable. Predictable is inefficiency in the allocation of
investment resources, distortion in individual portfolios, transactions costs directed solely to achieving tax
advantage, and the appearance, at least, of horizontal inequity.
In the next section we turn to the place of the corporation income tax in this story.
VI
The Corporation Income Tax
In an ideal Haig-Simons income tax there would be no need for a separate tax on corporations. Corporate
equities in individual portfolios would be evaluated at the beginning and end of the accounting period, and to
the difference would be added any cash distributions received on these securities. Nor is this a wholly
fanciful scheme. In modified forms it was proposed seriously by the Carter Commission [56, Royal
Commission on Taxation, 1965] for enactment for Canada and at least for serious discussion in the U.S.
Treasury's Blueprints. The method is most practical in cases of actively traded securities, hardly practical for
small corporations not traded on a major stock market. For this and no doubt other reasons, there exists a
separate tax on the income of corporations. Though the subject of a separate subchapter of the Internal
Revenue Code is devoted to this tax, income measurement for most corporations is broadly similar to that for
individuals. 24 The most interesting tax issues concern the treatment of distributions out of the corporation
and the consequences of various sorts of restructuring of a corporation's affairs.
The views of most economists about the corporation income tax have been greatly influenced by the work of
Harberger [43, 1962; 44, 1966], who modeled the tax as a levy on the services of capital as a factor of
production in the corporate sector. If the fraction of corporate finance taking the form of debt (with
deductible interest) is fixed, the rate of tax on corporate equity capital consists of the sum of the flat statutory
corporate rate (currently 46 percent; we neglect the fact that the first $100,000 of corporation income is
subject to lower rates ranging from 17 to 40 percent) and the individual taxes, consisting of ordinary income
tax on dividends and any capital gains taxes incurred or accrued due to the corporation's activities. While the
individual taxes vary according to the circumstances of the tax-
Â
Page 181
payer, Harberger concluded the net effect of the system is to tax corporate capital more heavily than capital
employed elsewhere. Using reasonable values for a simple two-sector (corporate and noncorporate) model of
the U.S. economy, he concluded that after all real-locations had taken place the corporation income tax was
equivalent to an extra tax on returns to capital in the aggregate. More elaborate general equilibrium modeling
by John Shoven and John Walley [58, 1972] using the same basic approach supports Harberger's conclusions.
Recent contributions to the economic theory of taxation and corporation finance have challenged the
adequacy of these models. A number of authors [13. Bradford, 1977; 47, King, 1974; 48, 1975; 49, 1977; 52,
Miller, 1977; 61, Stiglitz, 1973; 62, 1976] have questioned the assumption of fixed proportions of debt and
equity finance, pointing out that, like the investment decision, the choice of financing method is subject to the
control of corporate managers and should be expected to depend upon tax rules. In particular, these writers
have emphasized various ways in which the corporation and individual tax rates interact to influence
financial structure and, via financial structure, investment decisions. The effort to model optimal corporate
150
151
financial policy under tax rules now applicable in the United States has revealed numerous instances of
inconsistency in the treatment of economically similar transactions, inconsistencies of much the sort
previously discussed in connection with leveraged tax shelters.
Stiglitz [61, 1973] and King [47, 1974] argue that the deductibility of interest payments means the tax on
corporation profits will not affect the firm's investment criterion. A project with a rate of return just equal to
the rate of interest will, if financed by borrowing, generate no corporation tax. A project with higher return
will generate a positive surplus regardless of the rate of tax assessed on that surplus. The phrase ''that surplus"
rather than "taxable income" is used advisedly here, but for the moment assume the two are the same. How
then account for the existence and even predominance of equity financing of U.S. corporations?
The answer suggested is the tax advantage of equity finance at the individual level, in the form of deferral of
tax and low capital gains rates. Wealth accumulated within the corporation is subject only to the corporation's
income tax until it is realized by the shareholder, when it is subject to a "toll charge" in the form of individual
income
Â
Page 182
taxation. For individuals with high enough marginal tax rates this may be more attractive than accumulation
outside the corporation. In the simplest models the question is simply whether the corporation or individual
marginal rate is the higher one.
With uniform and constant tax rates all around, this argument would lead one to expect dominance of one
form of finance over the other. Conventional wisdom presumes a bias against equity, leading to the
expectation that in the absence of uncertainty the corporation tax will be wholly avoided by the use of debt
finance. The next degree of sophistication in the modeling seems to imply either all debt or all equity finance,
depending upon which tax rate is higher. In his presidential address to the American Finance Association,
Merton Miller [52, 1977] suggested a progressive tax schedule at the shareholder level may be instrumental
in bringing these two forces into balance. The explanation, which follows a line developed earlier by Martin
Bailey [7, 1974], is based on the fact that there is a distribution of taxpayers of varying marginal income tax
rates. Those with marginal rates above the corporate rate will wish to hold equity, and those with marginal
rates below the corporate rate will wish to hold debt. 25 Thus the distribution of wealth between high-bracket
and low-bracket individuals (in view of the endogeneity of individual marginal rates, we should say
"presumptively" high- and low-bracket individuals) determines the proportion of debt and equity in the
aggregate. Feldstein and Slemrod [30, 1978] have shown that such a view of the determinants of the
aggregate portfolio is consistent with plausible parameters for the tax rate distribution and risk attitudes in the
United States.
This line of argument as

presented in simple form here
clearly leaves out of account
important elements of the
explanation of actual portfolio
choices at the individual level
and management response at the
corporate level. Such
ingredients as risk attitudes,
transactions costs, and
differential information would
be part of a model directed at
such a goal. Yet the simple
151
152
model does suggest that the

Harberger view of the matter as
one of differential taxation of a
factor of production in a
particular sector should not be
accepted without question. The
new view gives a central role to
the corporate tax rate, since it
puts what amounts nearly to a
ceiling on the taxation of returns
to wealth.26
Although the general approach described above appears promising as an aid to understanding the effect of tax
policy on corporate
Â
Page 183
financial structure, it does not address the puzzling persistence of dividends as a method of distribution of
corporate funds. This issue here is not the choice between retention and distribution, but between dividends
and other forms of distribution. In view of the favorable tax treatment of capital gains, it is difficult to explain
the extent of dividend payments in the United States. 27 A corporation with cash available for distribution has
the option instead of purchasing its own shares in the stock market. In such a case (neglecting details of
transactions cost, minimum lot size, and so on) stockholders have the opportunity to generate the same cash
flow from their portfolios as would obtain from dividend payment, but the rate of tax would be at most that
applicable to capital gains. Indeed, in most cases, shareholders would have a positive tax basis in shares sold,
so that even capital gains tax would continue to be deferred on the bulk of the distribution.
The nature of the equivalence between dividends and share repurchase is worth spelling out in some detail.
Consider the case of a corporation wishing to distribute a specified sum, say D dollars. If this amount is paid
out in dividends a shareholder holding fraction s1 of the common stock will experience a portfolio change: he
will find himself with (1 - m1)s1D dollars extra cash (where m1 is his marginal income tax rate) and fractional
ownership s1 in a corporation with D fewer dollars in the corporate treasury than before the dividend. The
portfolio also embodies certain liabilities for future tax payments; of particular interest is the "basis," b1, of
the individual's ownership claim to the firmusually the amount he paid for the shareswhich is unaffected by
the dividend. Generally the shareholder will accompany receipt of the dividend with further portfolio
rearrangements, but first the portfolio experiences the shift: extra cash, change in character of the equity
claim due to withdrawal of funds from the corporation, and unchanged basis in the equity claim.
Consider now the alternative of share repurchase, under the assumption that transactions are costless. If
shareholder i sells a sufficient number of shares to realize a cash flow of exactly s1D his fractional ownership
will be just preserved at s1. The number of shares outstanding will, however, have been reduced; let us say a
fraction, a, has been "annihilated." The shareholder will thus experience a capital gain realization in the
amount of s1D - b1a, and will be liable for tax on this gain in the amount of g1(s1D - b1a), where g1 is the
applicable marginal rate. Then net effect on the portfolio of the two
Â
152
153
Page 184
events, dividend distribution versus proportional share repurchase, is thus to leave each shareholder in the
same position with respect to ownership claim in the corporation, and in the following position with respect
to cash and future tax liability:
Change in Change in
cash position portfolio basis
Dividend (1 - m1)s1D 0
Share repurchase (1 - g1)s1D + g1b1a -b1a
Except in unusual circumstances the second line in this picture dominates the first for individual
shareholders; share repurchase is unambiguously more advantageous than dividend distribution. For
individuals g1 is less than m1; usually g1 = 0.4m1. Share repurchase thus produces a higher net cash flow even
if the basis in the shares is zero. With nonzero basis the cash flow advantage of share repurchase is greaterthe
tax would even add to the cash flow in the event the basis in the shares sold exceeded the current market
value. When the basis is positive the disadvantage of reduction in basis of the portfolio (implying higher
future taxes) must be taken into account. Unless the tax rate on capital gains is expected to be higher in the
future than in the present, however, the "change in basis" effect can at most negate the term g1b1a in the
expression for cash flow, leaving the advantage due to g1 being less than mi.
The tax advantages of share repurchase have not gone wholly unnoticed by tax lawmakers. Thus the internal
revenue code provides, in effect, that a share repurchase structured so as to be effectively equivalent to a
dividend will be taxed as such. Yet relatively minor precautions on the part of the corporation and individual
shareholders seem sufficient to avoid this penalty. The sums of money involved in this choice are not
inconsiderable. In tax year 1976, individuals reported some $25b in dividends on their tax returns. The U.S.
Treasury estimates the average marginal rate of tax on dividends in the neighborhood of 40 percent, implying
some $10b in tax liability. A reduction of this tax bill by three quarters seems a conservative estimate of the
tax savings that would have been realized by the share repurchase alternative to dividend payment. Nor is
repurchase of own shares the only way to accomplish this. Purchase of shares in other companies (taking
advantage of the 85 percent exclusion of intercorporate dividends from the corporation tax base) or
Â
Page 185
merging with another firm in a purchase for cash achieves the same lightly taxed distribution of corporate
funds. These methods of distribution are in fact extensively used. Thus the continuing substantial volume of
dividend payments poses a challenge to tax analysis.
One possible explanation is the higher rate of tax applied to capital gains at the corporate level. While a
corporation may deduct 85 percent of dividends received, leading to a marginal tax rate of at most 7 percent
on dividends paid to a corporation, long-term capital gains realized by corporations are taxed at 28 percent.
Even for corporate shareholders, however, share repurchase will be superior to dividends if the basis in the
shares is sufficient (the term g1b1a in our schematic representation of alternatives) and if the expected
ultimate realization of the basis in the shares is sufficiently distant in time (so that the change, -b1a1 in
portfolio basis is of minor importance). But that corporate shareholders for which dividends are preferable to
153
154
capital gains can explain the extent of observed dividends is a possibility meriting further study. Dividends
qualifying for the 85 percent deduction reported on corporate tax returns in 1974 totaled $11b, by comparison
with the $21b reported on individual returns in that year.
Miller and Scholes [53, 1978] have suggested another explanation for the paradox of dividends in one of the
rules in the individual income tax intended to limit tax arbitrage through leveraged shelters. They single out
the restriction of the deduction for investment interest to the sum of $10,000 and investment income, together
with the availability of tax-free accumulation (in their example, in life insurance) as combining to eliminate
the tax on dividends at the margin. For a taxpayer bound by the constraint on deductibility of investment
interest, an additional dollar of dividends generates the capacity for an additional dollar of deductible interest
on leveraging borrowing, which shelters the dividend from tax. 28 Miller and Scholes do not show that the
taxpayer is, in fact, indifferent to this relaxation of the interest deduction constraintwhat they show is that he
can by offsetting purchase of life insurance restore his original portfolio characteristics with no increase in
current tax liability.
While this illustrates yet again the surprising ways in which the present income and consumption base tax
rules work in combination, it may be questioned whether it is yet a satisfactory explanation for the continuing
role of dividend distributions. Data from income tax returns indicate that a minuscule fraction of taxpayers
actually
Â
Page 186
finds the investment interest deduction limit a binding constraint. Miller and Scholes suggest that the
corporate officers actually making dividend distribution decisions may be the ones at this constraint, but one
hesitates to base a theory on such apparent self-dealing, particularly since the argument is not that these
shareholders gain from dividends but rather that they can neutralize the income tax consequences of
dividends.
Another approach to the study of the effects of the tax rules on the extent and form of distributions from the
corporation is to accept the existence of dividends as evidence of an unobserved cost of other forms of
distribution, offsetting their tax advantage. In [13, 1977], I present a model which may alternatively be
interpreted as a system where nondividend distributions are prohibitively costly or one where all distributions
are taxed as dividends. Attention is focused on the trade-off between dollars inside the corporation (after
taxes at that level) and dollars in the pockets of shareholders (after taxes at that level). One might expect that
this trade-off rate would be a crucial determinant of the rate of distribution from the corporation. However,
because the tax is assessed on a realization transaction, and not on accrual, it functions as a "toll charge" for
the passage of funds, and does not affect the relative attractiveness of holding the funds inside or outside the
corporation. The shareholder is interested in the net rate of return on funds inside the corporation compared
with that on funds outside. The toll charge simply determines his fractional interest in what happens inside
the corporation, and it has no bearing on the decision whether to withdraw that fractional share this period or
next.
A second surprising implication of this view of the taxation of distributions is that an increase in earnings
retained in the corporation may be expected to be reflected less than dollar for dollar in an increase in the
market value of outstanding equity. In the simple model I analyze, which incorporates only a tax on
distributions, an extra dollar inside the corporation raises the value of equity be one minus the tax rate. In
other words, purchasers of stock discount the net worth of the corporation by the rate of tax they would have
to pay to withdraw this wealth. This assures there will be no difference to the shareholder between realization
(conversion to cash) of his investment by sale of the stock and by distribution of corporate net worth. Where
there is only a fiat rate tax on distribution its level is irrelevant to the choice to distribute or retain. It does
have a bearing,
154
155
Â
Page 187
though, on the issue of new equity, since a dollar inside the corporation is valued at less than a dollar by the
stock market. While equity may be expanded via retained earningsthe basis for this expansion is "locked in"
to the corporation, and must bear a toll charge to get outissue of new shares to the public for cash would be
irrational. A further implication of this model is that a reduction in the tax on distributions would have
primarily, if not only, a wealth effect on current owners of equitysince the toll charge discount on the
corporate net worth is reducedand would have no effects on the investment or other policies of the firm,
except insofar as the reduction is sufficient to make new equity issue attractive. Interestingly, a reduction in
toll charge describes rather precisely the effect of most proposals for "dividend relief" to offset the "double
taxation'' of corporation income distributed in this form. 29
Naturally none of these conclusions will characterize precisely the real system in which the toll charge
(individual tax on dividends) varies from shareholder to shareholder, in which a tax (the tax on realized
capital gains) is assessed on the equity value changes attributable to retained earnings, whether or not these
are discounted in the marketplace in anticipation of the toll charge, and in which alternative routes for
withdrawing funds from the corporation are available. To consider just the last point, if alternative routes for
withdrawal are costly, but these costs are independent of the tax rules, we might well expect dividends, and
possibly the aggregate of all forms of distribution, to be responsive to the terms on which funds can be
transferred from corporation to shareholder via dividends. Just such a responsiveness has been found in
empirical work by Feldstein [21, 1970] and by King [49, 1977]. Some evidence supporting the validity of the
toll charge view is provided by the observed market values of closed-end mutual funds, which generally fall
significantly below net asset value. And while the estimation problems are severe, studies of the effect on
stock market values of dividends have found some discounting of wealth inside the corporation [37, Gordon
and Bradford, 1979; 50, Litzenberger and Ramaswamy, 1979].
Summing Up
This section is difficult to summarize because of the unsettled state of economic analysis of the particular
combination of devices by which
Â
Page 188
corporate shareholders are taxed in the United States. Without pretending that the model fully explains
observed behavior, I tend toward the "tax shelter" view of this system, as attributed above to Miller [52,
1978]. A ceteris paribus increase in the corporate rate may be expected to lead to a change toward bond
finance, but to have an effect on real investment only as this in turn leads to an upward shift in individual
marginal rates. Conversely, a ceteris paribus upward shift in individual rates should cause a move toward
equity finance which tends to offset the individual rate change.
In this view the corporate rate plays a very important, even dominant, role in the taxation of returns to saving,
and a large divergence between individual and corporate rates (as would occur, for example, if individuals
were put on a pure consumption tax system without changing the corporation tax rules) would produce severe
strains. However, the taxation of distributions to equity holders, primarily individual income taxation of
dividends, may not have the significant allocation effects often attributed to it. Rather the present taxation of
155
156
distributions has its principle effects on the way in which new equity is created (for example, by retention of
earnings or mergers). Mischievous consequences could be averted by a more consistent flat rate taxation of
distributions, a possibility to which we shall return in the concluding section with its suggestions for policy
change.
VII
Conclusions
At the outset we noted the lack of a consistent tax policy toward savings in the United States today, as
indicated by the great diversity of ways in which the rewards to saving are taxed. The confusing state of the
law reflects in part the inadequacy of the economic analysis of the accumulation of wealth that has made its
way into policy debates. Even trained economists have commonly applied to this issue models that fail to
incorporate significant interactions of the various tax rules bearing on the savings decision. For example, the
government's revenue requirement may be left out, with its implication that a decrease in one tax rate must
ordinarily be offset by an increase in another. More common is neglect of the way in which an individual's
marginal tax rate bracket, in the graduated individual income tax, varies over his lifetime in a way depending
upon the amount, method, and time pattern of his saving.
Â
Page 189
Such omissions can seriously affect policy conclusions. My objective in this chapter has been to provide an
analysis which does not omit any of the important features of the present and feasible alternative tax
treatments of savings, but remains at a reasonably simple level. While the exposition is addressed to readers
with some literacy in economics the principal points made can be summed up in nontechnical terms in this
section, which will also take up policy implications.
In spite of numerous provisions in the law to lighten the burden of taxes on the reward to saving, the rough
calculation carried out in Section I suggests an average federal tax rate of 36 percent on the annual yield from
wealth. This may not seem extraordinarily high when compared with out similarly rough estimate of an
average rate of 25 percent on the returns to labor. It is somewhat misleading, however, to conceive of the tax
on savings in annual terms. Most saving is directed toward rearranging the timing of expenditures within the
life span of the individual (or married couple), primarily the provision in early life for educational and other
expenses of child rearing and for retirement. For these decisions what is of interest is the price today of future
dollars. A flat rate tax on the annual yield from savings translates into a geometrically increasing set of taxes
on the purchase of dollars successively far into the future. For example, if the annual yield before taxes is 10
percent, a 36 percent tax on that yield raises the price today (that is, the required current saving) of a dollar a
year hence from $0.909 to $0.940, a modest 3.4 percent increase. But it raises the price of a dollar in
retirement at age sixty-five, purchased at age thirty-three, from $0.047 to $0.137; that is, it nearly triples the
price.
With a modicum of analytic license most of the rules in what is called the individual "income" tax can, as far
as the treatment of savings is concerned, be described as some combination of "income" and "consumption"
tax rules, as these terms are technically understood. The distinguishing feature of a consumption tax is that
the returns to savings are effectively exempted from tax, either by literal exemption of the yield or by a
combination of a tax deduction for the act of saving followed by full taxation of the dissaving that takes place
when the yield is realized. Because of this a consumption tax system can also be loosely equated with a
system of taxation of earnings from labor only, with the details of the equivalence depending upon how
graduated rates are implemented and how inheritances and other transfer payments are treated.
156
157

Page 190
It is sometimes argued that because of the tax "wedge" between the return on investment, for example in
machinery, and the aftertax reward to the saver providing the wherewithal, too little investment is undertaken.
This overlooks the unfortunate necessity for there to be some such wedges if the government revenue
requirements are to be met. In Section II we considered the question of what combination of taxes on labor
services and rewards to savings will reduce to a minimum the wasteful disincentive effects while meeting the
revenue target. We showed that the answer cannot be determined a priori; rather it depends upon how
individual saving and labor supply decisions respond to the tax rate. This general result is well known to
economists. What still seems less widely understood is that the responsiveness of the rate of saving to the
yield on saving taken by itself does not determine the optimal balance between consumption and income
taxes. In particular, we examined an economy in which a pure consumption tax was seen to be the best choice
even though the savings rate was wholly independent of the rate of return.
In Section III we used the same basic model to demonstrate the similar indeterminacy of the relationship
between tax rules and the accumulation of wealth in the economy. Thus it is possible that a combination of
lightly taxing labor earnings and heavily taxing the rewards to saving could lead to greater wealth
accumulation than the reverse policy.
Perhaps more important are two related points emerging from our analysis of aggregate accumulation. First,
it is necessary in assessing the influence of tax policy on saving to take into account both government and
private saving. In one example a shift of taxes from the return on savings to labor earnings resulted in a
decrease in private savings but an increase in the sum of private and public savings. This suggests the validity
of the notion, surprising at first sight, that accumulation effects are not crucial determinants of structural tax
policy, as in the choice between income and consumption as a tax base. The key is the ability of government
through its overall budget policysurplus or deficitto determine the level of aggregate wealth. This ability in
turn depends upon a less than full substitutability of public for private saving in individual portfolios. That is,
government surpluses achieved by a general increase in tax rates will result in larger aggregate saving if
individuals do not react by reducing their private wealth stocks by an equal amount.
Â
Page 191
In this circumstance, it should be emphasized, government has no choice but to have a policy toward
aggregate accumulation. Budget policy will affect the stock of true claims on future consumption, embodied
either in real capital for in financial claims on foreigners. A balanced cash budget represents a particular
choice of policy, but it cannot be shown in the present framework of analysis to be preferred to other policies.
A rising stock of (real) government debt diminishes the real inheritance of future generations, shifting
consumption to those presently alive; government saving has the opposite effect. Zero saving by the
government is simply one choice about the distribution of wealth between current and future generations.
A second point to emerge from the analysis in Section III is that the effect of budget policy on wealth
accumulation cannot be determined by reference to cash budget surplus or deficit, or even by this residual
plus the change in value of paper claims on government due to inflation. Other policy choices have the effect
of changing individuals' claims on future payments from the government or their anticipated liability for
future taxes. Consequently, in carrying out a change in tax (or other) policy with a particular
wealth-accumulation objective in view, it is necessary to take a more refined approach to analysis of the net
budget position. It may be that such a modification of budgetary analysis would help resolve some of the
157
158
puzzles of recent economic policy. Here the point to emphasize is that if a particular structural tax policy
choice is found to have an undesirable effect on accumulation under a given cash budget constraint, this can
be offset through changing the constraint.
While specific provisions of the income tax can be described as expressing consumption or income tax
principles, our discussion in Section IV showed that the system as a whole cannot properly be viewed as a
point in a spectrum between consumption and income tax bases. In this respect the usual analysis is rather
misleading. What we actually have is not a position between two poles in any very pure sense, but rather
different rules applicable to transactions that are virtually identical economically. In organizing their affairs
individuals will pay attention to after-tax outcomes, and the applicability of both income and consumption tax
rules to similar assets introduces opportunities for profit obtained by portfolio rearrangements of no
significance to the taxpayer other than their effect on his tax liability. I refer to such transactions as "tax
arbitrage." Borrowing with deductible interest to make an investment subject to the
Â
Page 192
investment tax credit is an

example. The net effect of these
taxpayer responses may be very
different from that of the
application of either pure
consumption or pure income tax
rules.
The essence of the competitive market mechanism is its tendency to eliminate profit opportunities, as when
high returns attract firms to a new industry. The analysis of how the competitive system reacts to eliminate
"tax profits" was the subject of Section V. Three basic adjustment mechanisms were emphasized. The first is
administrative or political: special rules to inhibit observed instances of tax arbitrage. An example is the rule
against deducting interest payments incurred on borrowing to acquire or hold tax exempt bonds. This sort of
reaction to tax arbitrage accounts for much of the complexity of the U.S. income tax.
The second basic adjustment mechanism is change in the rates of return on assets depending upon their tax
treatment. Thus, for example, tax exempt bonds have a lower equilibrium yield than taxable bonds. Similarly,
assets eligible for the investment tax credit have a lower equilibrium yield (before tax) than the interest rate
on taxable bonds (after suitable allowance for risk differences). Unless different real assets are treated the
same this translates into an inefficient allocation of investment, with too much of the lightly taxed investment
opportunities undertaken relative to the more heavily taxed. Note that this inefficiency is different from that
due to the "tax wedge" between rate of return on real investment and yield to the saver which we analyzed in
Section II. My own guess is that this second sort of inefficiency due to present rulesan unnecessary wasteis
more significant than the first.
The third adjustment mechanism is via the progressive rate schedule applicable to the tax base of each
individual. Deductions from the base in the present, for example accelerated depreciation allowances, tend to
lower the present marginal rate of an individual taxpayer, while the subsequently enlarged return flow of
postponed tax base raises the marginal rate applicable then. Both the second and third forms of adjustment,
when combined with a government revenue constraint, render indeterminate the effect on wealth
accumulation of tax devices, such as accelerated depreciation, adopted with the specific intention of
encouraging capital formation.
A comprehensible picture of the individual income tax emerges from this analysis. Fitting the corporation
income tax into this picture was the objective of Section VI. The corporation tax can be
158
159
Â
Page 193
viewed as analogous to a tax on interest accumulating within a marketable savings account (with such
accumulations excluded from depositors' individual income tax), where equity owners correspond to
depositors. Withdrawls from these accounts (dividends) are subject to an additional tax, as are gains from sale
of the accounts (realized capital gains). The "interest rate" on these accounts is, of course, risky. Accounts of
this type will be attractive to certain savers, including those for whom a relatively high return compensates
for the risk and those for whom the individual tax on savings in other forms is high relative to the corporation
income tax inside these "accounts."
We can think of corporations as financing their real investment portfolios by a combination of the sale of
bonds and these special "savings accounts." Thus the broad outlines of this complex tax structure can be
readily grasped. This system cannot be represented as a simple add-on to the individual income tax. The
workings of the tax reflect a balancing of its component elements. An increase in the corporate tax rate, for
example, will not simply drive real investment out of the corporate sector. Instead it will lead to a shift away
from equity finance toward bond finance, with a further ripple of adjustment taking place in the tax rates
applicable to individual shareholders. To take another example, a reduction in the tax rate on dividends, such
as would be implied by some actively considered forms of integration of corporate and individual taxes,
would be likely to be reflected primarily in an increase in market value of existing shares.
While the "savings account" model is a useful approximation, the details of the actual system display myriad
inconsistencies of the same sort discussed earlier in connection with tax arbitrage in the individual income
tax. The taxation of distributions to shareholder "depositors" depends very much upon the particular route
chosen to tap the corporate vessel. In the case of closely held corporations, for example, the operation of the
system is best described as a game between tax legislators attempting to assure distributions are taxed as
dividends and tax lawyers structuring the distributions to obtain the preferential capital gains treatment. Just
how any shareholder "depositor" is taxed depends importantly upon such issues as how often he rearranges
his portfolio, matters peripheral to the proper objectives of tax policy. In short, a simple model of the
combination corporation-individual income tax should not blind us to its many
Â
Page 194
capricious features, or to the uncertain state of its overall effect, or to the vast cost of administering it.
Implications for Policy
The power of economic reasoning to determine desirable policy is limited. In may view, however, the
description above of the functioning of the U.S. system of taxing the rewards to saving, when combined with
reasonable guesses about empirical relationships and broadly attractive criteria of equity, supports explicit
rejection of the Haig-Simons income base and adoption of a consumption base as the fundamental principle
upon which tax policy should be based. I have developed this viewpoint at some length elsewhere [14,
Bradford, 1980]. Without going into details, let me here note three general propositions underlying may
preference. First, the value of a person's potential consumption over his lifetime is an appealing standard of
well-off-ness; a consumption base will correlate more closely with this measure than will an income base.
Second, a consumption base will be desirable on efficiency grounds when people can be described as
159
160
separating their work and savings decisions, first deciding how much to work and then how much of earnings
to allocate to provision for the future; while certainly not precisely true, this seems a reasonable description
of behavior, as is suggested by its being typically assumed in economic studies of consumption over the life
cycle. Third, it is as a practical matter almost impossible to design a consistent and workable income tax
system, particularly in an inflationary world, as the present hybrid system illustrates; it is not difficult to
implement the consumption principle.
It is usual to formulate policy recommendations in the form of limited, "politically feasible" changes. A shift
to a consumption base for the present "income" tax may not seem to qualify in this respect. Yet this chapter
has emphasized the likelihood that marginal adjustments to the existing rules, not designed to advance toward
the consumption base objective, will lead us further in the direction of inefficiency and complexity. In
particular, present methods of encouraging the accumulation of wealth have been shown to be of dubious
effectiveness.
Furthermore, the present hybrid is not vastly different from a consumption tax. Blueprints offers a fully
spelled out consumption base system, which is shown to be not as distant from present rules
Â
Page 195
as might have been expected. To implement a consumption base system what is required is consistent
treatment of savings and investment by the two basic methods. Either a deduction for saving or investing is
allowed, in which case all associated dissaving is taxed. Or no deduction for saving or investing is allowed, in
which case none of the return is taxed. Ample precedent exists in the present system for the first method, as
in tax-sheltered retirement savings plans. The second approach is generally found in modified form, as in the
treatment of accruing capital gains. The natural tendency of tax lawmakers as they consider how to stimulate
capital accumulation is to adopt some version of these consumption tax devices: investment credit, sheltered
retirement savings, accelerated depreciation, reduced rates of tax on capital gains, and so on. All that is
required is to coordinate these steps in a plan of transition toward a full consumption base system.
We can do no more here than outline a program that could accomplish this.
1. Remove present limits on savings plans, such as employer-provided retirement programs, Keogh plans,
and IRAs, qualifying for tax shelter treatment. In particular eliminate ceilings on contributions and the
restriction on the withdrawal of funds to retirement years.
2. Phase out all taxation of interest receipts, dividends, and capital gains, except as these are realized through
withdrawals from tax-sheltered savings plans.
3. Accelerate depreciation deductions over a period of years until full first-year write-off of investment is the
rule; similarly allow immediate deduction of inventory purchases, while proceeds from sale of assets written
off or of expensed inventories are to be included in full in the tax base.
4. Introduce the possibility of "negative saving" in tax qualified plans, that is borrowing, the proceeds of
which are immediately taxed. Repayments would in turn be fully deductible. This device is needed, for
example, to allow a new business to take advantage of immediate write-off of investment. 30 Steps 1 and 4
constitute establishment of the "qualified accounts" described in Blueprints.
As has been observed, in some form these basic devices are already in use, and their implementation seems to
be attractive to
Â
160
161
Page 196
lawmakers. What has not been undertaken is any attempt to rationalize the system as a whole. To eliminate
the serious undesirable effects of the hybrid system it is essential that politically less attractive measures be
taken as well:
5. Phase out ad hoc savings and investment incentives, such as the investment tax credit.
6. Phase out the deduction for interest payments other than in the circumstances described in step 4.
The last step, in particular, will seem radical, but as the analysis of Section V has shown, it is required if a
consistent and simple treatment of savings is to be achieved in a graduated rate tax system.
Before we turn to the taxation of

corporations a point should be
emphasized which has received
only passing mention in the text,
although it is emphasized in
Blueprints and [14, Bradford,
1980]. That is that the adoption
of these rules would essentially
eliminate problems of
adjustment of the tax base for
the effects of inflation, a serious
defect in the present system.
Blueprints suggests that the corporation

income tax be eliminated in a move to a
full consumption base. However, our
discussion above of the combination
corporation-individual income tax
system as a "savings account" together
with a tax in the nature of a "toll
charge" on distributions to stockholders
suggests an alternative treatment.
Adoption of an explicit tax on
distributions, other than debt repayment,
would maintain the present
governmental claim on existing
corporate capital and a collective
participation in new equity investment
in the future. To accomplish this add to
the program described above the
following:
7. Phase out the corporation income tax.
8. Phase in a flat-rate tax on corporate distributions with respect to equity holders, whether in the form of
dividends, stock repurchase, or purchase from public holders of stock in other corporations. Introduce a credit
computed at the same flat rate on corporate sales of equity to the public. Such a credit might be held in an
interest-bearing account to offset future distribution tax liability.
These steps would effect a

complete overhaul of the U.S.
tax treatment of savings. Their
161
162
elaboration into a detailed

program could easily occupy
the full space of this chapter.
Taken together they may
Â
Page 197
seem a radical package. Yet they

need not be taken abruptly, nor
are they wholly unfamiliar
devices, either in actual tax
practice or in expert discussion.
And the argument of this chapter
is that little short of such a
package is likely to do much to
restore order to the tax treatment
of savings. Indeed, many
measures taken or advocated in
isolationfor example the
reduction in tax rates on capital
gainsare likely to diminish the
equity of the tax system and to
inhibit the efficient allocation of
investment resources, while
having a wholly indeterminate
effect on the aggregate
accumulation of wealth. There
seems to be no adequate
substitute for thinking the system
all the way through and taking the
consequences.
Notes
Among the many individuals who have assisted in the preparation of this chapter the author would like
particularly to thank William Andrews, Willem Buiter, Larry Dildine, Barbara Hewson, Don Fullerton, Roger
Gordon, and Harvey Rosen. The views expressed here are his own and do not represent positions of
Princeton University or the National Bureau of Economic Research.
1. For example, a person in the 50 percent marginal rate bracket who makes a $100 deposit in a HR-10 plan
receives a deduction of $100 in calculating taxable income. He thus gives up $50 in current consumption. If
he cashes in his account after retirement the next year, when his marginal tax rate is 25 percent, he realizes 0
75 Ã $105 50 = $79.13, or $72.90 in terms of November 1977 dollars. Thanks to the tax system he receives
a return of 46 percent on his savings.
2. For a discussion of this concept see Richard Goode [36, 1977]. For policy recommendations based on this
ideal see, for example, U.S. Treasury [67, 1977], Joseph Pechman [55, 1977], George Break and Pechman
[15, 1975], Royal Commission on Taxation [56, 1966], U.S. Congress [66, 1959].
3. This and the following related figures are taken from U.S. President [68, 1979].
162
163
4. This treatment of the deficit amounts roughly to the assumption that its servicing plus repayment will be
supported by a proportional expansion of future taxes. Just how deficits affect wealth accumulation in the
aggregate will be considered further in the text.
5. Note, though, that the elderly tend to receive a high proportion or income in the form of yield on pensions
and other savings, and they also make up a major fraction of the lowest income category. For a discussion of
some of the difficulties this creates for displaying the distributional characteristics of tax alternatives, see
Blueprints, pp. 154-56, 176-79.
6. Under present legislation the rules would change to give heirs the basis of a decedent dying after
December 31, 1979. However, the effective date of the transition to this ''carryover-basis" rule has already
been postponed once, and it is the subject of continuing legislative attention.
Â
Page 198
7. Michael Boskin [11, 1976] has emphasized the importance of this.
8. For a compact discussion of the "capital shortage" controversy of the mid-1970s see Robert Eisner [20,
1977].
9. Obviously I am here glossing over the question of what "full employment" may mean.
10. In a recent examination of the evidence from aggregate time series data, Robert Hall [41, 1978] concludes
that for this purpose the life cycle model is superior to the prominent alternatives.
11. The model we shall use is based on one introduced by Paul Samuelson [57, 1958], and has been
employed for similar purposes by, among others, Peter Diamond [18, 1965; 19, 1973] and Feldstein [28,
1978].
12. This formulation implies the tax rate is expressed on a "tax exclusive" basis, that is, as a percentage of the
net of tax outlay on consumption. The term, contrasted with a "tax inclusive" expression of the rate, is due to
the Meade Report.
13. Note that the leisure-second

period consumption choice will
be distorted if t2ℜ≠ tw, as will be
true for either tw or t2 equal to zero
(if any revenue is to be raised).
14. For a recent summary and analysis of the evidence see Boskin [12, 1978].
15. The popularity of the Cobb-Douglas assumption, which has strong implications, rests primarily on its
analytical tractability. Its property of unitary wealth elasticities may be considered plausiblewe have noted
the use of this property by Modigliani and Brumberg [54, 1954]but the Cobb-Douglas form shares this with
all other constant elasticity of substitution functions, indeed, with all other homothetic preference structures.
16. This feature of the model with cash-flow constrained government plays a central role in Atkinson and
Sandmo [3, 1977].
17. See also Buiter's [16, 1979] formal modeling of fiscal policy in a steadily growing economy.
18. Under U.S. law, U.S. citizens and U.S. residents are normally subject to U.S. tax (with allowance for
foreign income taxes) with respect to income from all sources. I use the terms "foreigners" and "residents of
other countries" loosely to refer to those not subject to U.S. income taxes.
163
164
19. This particular example has inspired much of the recent discussion of debt-neutrality. For a review see
Buiter and Tobin [17, 1979].
20. A classic analysis of this sort is that by Arnold Harberger [43, 1962] of the corporation income tax.
21. Note here a case in which a higher tax rate may generate an investment incentive. Hall and Dale
Jorgenson [42, 1967] argued on similar grounds that a higher corporate tax rate would have provided an
investment incentive in the United States in the 1960s.
22. More precisely, the required second-period tax rate is 57.9 percent. With these marginal rates the
combination of $55 of borrowed money with a tax rebate of $45 finances a $100 investment, with gross yield
of $110. Taxes due are 0.579 Ã ($110.00 -$5.50). This leaves just $60.50 to repay the lender.
Â
Page 199
23. For further discussion of this effect and empirical evidence of its importance, see Martin Bailey [7, 1974].
24. Important exceptions are the treatment of dividends received (85 percent is excluded from the taxable
income of the receiving corporation) and capital gains (all are subject to a flat 28 percent tax rate).
25. Miller did not point out that there will be profitable tax arbitrage opportunities tending to drive all
marginal tax rates to the corporate tax rate. High marginal rate taxpayers should borrow, with deductible
interest, to buy stock, while low marginal rate taxpayers should sell stock short to lend more.
26. Indeed, with "radical tax arbitrage" as described in the preceding note, the corporate tax rate would tend
to evolve into the rate of tax at the margin on wages and interest!
27. Note, however, the significant exception: capital gains are taxed more heavily than dividends to corporate
shareholders. This will be discussed further in the text.
28. It is significant in this context that investment interest cannot be set off against long-term capital gains.
Note also that Miller and Scholes ignore the effects of the dividend on the subsequent taxation of gain from
the sale of the stock. A cash dividend presumably lowers the price of shares (the ex-dividend effect) and
hence lowers any capital gains tax due on ultimate sale of stock. This effect would increase the attractiveness
of dividend distributions.
29. For an extensive discussion of this and related forms of integration of corporation and individual income
taxes see Charles McLure [51, 1979].
30. Without this device a new business making an investment of $1,000 and having no receipts would have a
negative tax base of $1,000. By borrowing $1,000 on a taxable basis the business could "bank" the deduction,
with interest, to a further period of positive receipts. Blueprints shows how this device might also be used by
households to accomplish tax averaging over a period of years.
References
1. Andrews, William D. "A Consumption-type Cash Flow Personal Income Tax," Harvard Law Review,
April 1974, 87(6), pp. 1113-88.
2. Arrow, Kenneth J., and Kurz, Mordecai. Public Investment, the Rate of Return, and Optimal Fiscal Policy.
Baltimore: The Johns Hopkins Press, 1970.
164
165
3. Atkinson, Anthony B., and Sandmo, Agnar. "The Welfare Implications of Personal Income and
Consumption Taxes" (revised as "The Taxation of Savings and Economic Efficiency," mimeo), Discussion
Paper 10/77, Norwegian School of Economics and Business Administration. Bergen, Norway, 1977.
4. Atkinson, Anthony B., and Stiglitz, Joseph E. "The Design of Tax Structure: Direct versus Indirect
Taxation," Journal of Public Economics, July-Aug. 1976, 6(1-2), pp. 55-75.
5.. Lectures on Public Economics. New York: McGraw-Hill, 1980.
6. Bailey, Martin J. "Capital Gains and Income Taxation," in The Taxation of Income from Capital. Edited by
Arnold C. Harberger and Martin J. Bailey. Washington, D.C.: The Brookings Institution, 1969.
Â
Page 200
7.. "Progressivity and Investment Yields under U.S. Income Taxation," Journal of Political Economy,
Nov./Dec. 1974, 82(6), pp. 1158-75.
8. Barro, Robert J. "Are Government Bonds Net Wealth?" Journal of Political Economy, Nov./Dec. 1974,
82(6), pp. 1095-1117.
9.. "Reply to Feldstein and Buchanan," Journal of Political Economy, April 1976, 84(2), pp. 331-36.
10. Blinder, Alan S. Toward an Economic Theory of Income Distribution. Cambridge, Mass.: MIT Press,
1974.
11 Boskin, Michael J. "Notes on the Tax Treatment of Human Capital," in Conference on Tax Research
1975. U.S. Treasury. Washington, D C.: U.S.G.P.O., 1976.
12.. "Taxation, Saving, and the Rate of Interest," Journal of Political Economy, April 1978, 86(2), Part II, pp.
S3-S28.
13. Bradford, David F. "The Incidence and Allocation Effects of a Tax on Corporate Distributions,"
Discussion Paper No. 7738, Center for Operations Research and Econometrics, UniversitÃ© Catholique de
Louvain, Belgium, August 1977 (rev. forthcoming in Journal of Public Economics).
14.. "The Case for a Personal Consumption Tax," in What Should Be Taxed Income or Expenditure? Edited
by Joseph A. Pechman. Washington, D.C.: The Brookings Institution, 1980.
15. Break, George F., and Pechman, Joseph A. Federal Tax Reform: The Impossible Dream? Washington,
D.C.: The Brookings Institution, 1975.
16. Buiter, Willem H. "Government Finance in an Overlapping Generations Model with Gifts and Bequests,"
in Social Security Versus Private Saving. Edited by George M. Von Furstenberg. Cambridge, Mass.:
Ballinger Publishing Company, 1979.
17., and Tobin, James. "Debt Neutrality: A Brief Review of Doctrine and Evidence," in Social Security
versus Private Saving. Edited by George M. Von Furstenberg. Cambridge, Mass.: Ballinger Publishing
Company, 1979.
18. Diamond, Peter A. "National Debt in a Neoclassical Growth Model," American Economic Review, Dec.
1965, 55(5), Part I, pp. 1126-50.
19.. "Taxation and Public

Production in a Growth
165
166
Setting," in Models of
Economic Growth, Edited by J.
A. Mirrlees and N.H. Stern.
London: Macmillan, 1973.
20. Eisner, Robert. "Capital Shortage: Myth and Reality," American Economic Review, Feb. 1977, 67(1), pp.
110-15.
21. Feldstein, Martin S. "Corporate Taxation and Dividend Behavior," Review of Economic Studies, Jan.
1970, 37(1)(109), pp. 57-72.
22.. "Social Security, Induced Retirement and Aggregate Capital Accumulation," Journal of Political
Economy, Sept./Oct. 1974, 82(5), pp. 905-906.
23.. "Tax Incidence in a Growing Economy with Variable Saving Rates," Review of Economic Studies, Oct.
1974, 41(4), pp. 505-13.
24.. "Tax Incidence with Growth and Variable Factor Supply," Quarterly Journal of Economics, Nov. 1974,
88(4), pp. 551-73.
Â
Page 201
25.. "Perceived Wealth in Bonds and Social Security: A Comment," Journal of Political Economy, April
1976, 84(2), pp. 331-36.
26.. "National Saving in the United States," in Capital for Productivity and Jobs. Edited by Eli Shapiro and
William L. White. Englewood Cliffs, N.J.: Prentice-Hall, 1977.
27.. "Does the United States Save Too Little?" American Economic Review, Feb. 1977, 67(1), pp. 116-21.
28.. "The Welfare Cost of Capital Income Taxation," Journal of Political Economy, April 1978, 86(2), Part
II, pp. S29-S52.
29. Feldstein, Martin S., and Slemrod, Joel. "Personal Taxation, Portfolio Choice and the Effect of the
Corporation Income Tax," National Bureau of Economic Research Working Paper No. 241. Cambridge,
Mass.: NBER, April 1978.
30.. "Inflation and the Excess Taxation of Capital Gains on Corporate Stock," National Tax Journal, June
1978, 31(2), pp. 107-18.
31. Feldstein, Martin S., and Summers, Lawrence. "Is Rate of Profit Falling?" Brookings Papers on
Economic Activity, 1977 (1), pp. 211-28.
32.. "Inflation, Tax Rules, and the Longer Term Interest Rate," Brookings Papers on Economic Activity. 1978
(1), pp. 61-109.
33.. "Inflation and the Taxation of Capital Income in the Corporate Sector," National Tax Journal, Dec. 1979,
32(4), forthcoming.
34. Fisher, Irving. The Theory of Interest. New Haven, Conn.: Yale University Press, 1930.
35. Friedman, Milton. A Theory of the Consumption Function. Princeton, N.J.: Princeton University Press,
1957.
166
167
36. Goode, Richard. "The Economic Definition of Income," in Comprehensive Income Taxation. Edited by
Joseph A. Pechman. Washington, D.C.: The Brookings Institution, 1977.
37. Gordon, Roger H., and Bradford, David F. "Stock Market Valuation of Dividends: Links to the Theory of
the Firm and Empirical Estimates," Department of Economics, Princeton University, Mimeo. Princeton, N.J.,
May 1979.
38. Green, Jerry, and Sheshinski, Eytan. "Approximating the Efficiency Gain of Tax Reforms," Harvard
Institute of Economic Research Discussion Paper No. 516. Cambridge, Mass., Nov. 1976.
39. Grieson, Ronald E. "The Incidence of Profits Taxes in a Neo-Classical Growth Model," Journal of Public
Economics, Feb. 1975, 4(1), pp. 75-85.
40. Hall, Robert E. "The Implications of Consumption and Income Taxation for Economic Growth,"
Proceedings of the National Tax Association, 1968, pp. 125-45.
41.. "Stochastic Implications of the Life Cycle-Permanent Income Hypothesis: Theory and Evidence,"
Journal of Political Economy, Dec. 1978, 86(6), pp. 971-88.
42. Hall, Robert, and Jorgenson, Dale. "Tax Policy and Investment Behavior," American Economic Review,
June 1967, 57(3), pp. 391-414.
Â
Page 202
43 Harberger, Arnold C. "The Incidence of the Corporation Income Tax," Journal of Political Economy, June
1962, 70(3), pp. 215-40.
44.. "Efficiency Effects of Taxes on Income from Capital," in Effects of Corporation Taxation. Edited by M.
Krzyzaniak. Detroit. Wayne State University Press, 1966.
45. Harberger, Arnold C., and Bailey, Martin J, eds. The Taxation of Income from Capital. Washington, D.C.:
The Brookings Institution, 1969.
46. Institute for Fiscal Studies. The Structure and Reform of Direct Taxation, Report of a Committee Chaired
by Professor J. E Meade. London: George Allen and Unwin, 1978.
47. King, Mervyn A. "Taxation and the Cost of Capital," Review of Economic Studies, Jan 1974, 41(1)(125),
pp. 21-35.
48.. "Taxation, Corporate Financial Policy, and the Cost of Capital: A Comment," Journal of Public
Economics, Aug 1975, 4(3), pp. 271-79.
49.. Public Policy and the Corporation. New York: John Wiley & Sons, 1977.
50. Litzenberger, Robert H., and Ramaswamy, Krishna. "The Effect of Personal Taxes and Dividends on
Capital Asset Prices: Theory and Empirical Evidence," Bell Laboratories Economic Discussion Paper No.
136, Murray Hill, N.J.: Bell Laboratories, Jan. 1979.
51. McLure, Charles E. Must Corporate Income Be Taxed Twice? Washington, D.C.: The Brookings
Institution, 1979.
52. Miller, Merton H. "Debt and Taxes," Journal of Finance, May 1977, 32(2), pp. 261-75.
53. Miller, Merton H., and Scholes, Myron S. "Dividends and Taxes," mimeo. Chicago: University of
167
168
Chicago School of Business, November 1978.
54. Modigliani, Franco, and Brumberg, Richard. "Utility Analysis and the Consumption Function: An
Interpretation of Cross Section Data," in Post Keynesian Economics. Edited by K. K. Kurihara. New
Brunswick, N.J.: Rutgers University Press, 1954.
55. Pechman, Joseph A., ed. Comprehensive Income Taxation. Washington, D.C.: The Brookings Institution,
1977.
56. Royal Commission on Taxation. Report of the Royal Commission on Taxation. 6 vols. Ottawa: Queen's
Printer, 1966.
57. Samuelson, Paul A. "An

Exact Consumption Loan Model
of Interest with or without the
Social Contrivance of Money,"
Journal of Political Economy,
Dec. 1958, 66(6), pp. 467-82.
58. Shoven, John, and Whalley, John. "A General Equilibrium Calculation of the Effects of Differential
Taxation of Income from Capital in the U.S.," Journal of Public Economics, Nov. 1972, 1(3-4), pp. 281-321.
59. Simons, Henry C. Personal Income Taxation The Definition of Incomes as a Problem of Fiscal Policy.
Chicago: University of Chicago Press, 1938.
60. Smith, Adam. "An Inquiry into the Nature and Causes of the Wealth of Nations," in An Inquiry into the
Nature and Causes of the Wealth of Nations, by Adam Smith. Edited by Edwin Cannan. New York: The
Modern Library, 1937.
Â
Page 203
61. Stiglitz, Joseph E. "Taxation, Corporate Financial Policy, and the Cost of Capital," Journal of Public
Economics, Feb. 1973, 2(1), pp. 1-34.
62.. "The Corporation Tax," Journal of Public Economics, April-May 1976, 5(3-4), pp. 303-11.
63. Summers, Lawrence H. "Tax Policy in a Life Cycle Model," National Bureau of Economic Research
Working Paper No. 302. Cambridge, Mass.: NBER, Nov. 1978.
64. Tobin, James, and Buiter, Willem H. "Fiscal and Monetary Policies, Capital Formation and Economic
Activity," in The Government and Capital Formation. Edited by George M. Von Furstenberg. Cambridge,
Mass.: Ballinger Publishing Company, 1980.
65. Tobin, James, and Dolde, Walter. "Wealth Liquidity and Consumption," in Consumer Spending and
Monetary Policy The Linkages. Federal Reserve Bank of Boston Conference Series No. 5. Boston: FRBB,
June 1971.
66. U S. Congress. Tax Revision Compendium of Papers on Broadening the Tax Base. House Committee on
Ways and Means, 86th Cong., 1st Sess Washington, D.C.: U.S.G.P.O., 1959.
67. U.S. Treasury. Blueprints for Basic Tax Reform. Washington, D.C.: U.S.G.P.O., Jan. 1977.
68. U.S. President. Economic Report of the President. Washington, D.C.: U.S.G.P.O., Jan. 1979.
168
169
Â
Page 205
7
Issues in the Design of Savings and Investment Incentives
The view is widespread that the rate of accumulation of capital in recent years in the United States has been
too low. This conclusion is based in part upon a comparison of the rate of net investment, especially of
business fixed investment, relative to aggregate output both with its past value in the U.S. and with its value
in the other advanced industrial countries. It is based as well upon symptoms that are presumptively traceable
to a slow-down in capital formation, most especially the apparent cessation in the growth of labor
productivity.
A deterioration in the rate of return to savers resulting from the interaction of inflation with an unindexed tax
system is often cited among the reasons for this shift in performance of the U.S. economy. In particular, the
failure of the tax rules to permit correct accounting for depreciation and capital gains, together with a not
obviously explicable reluctance of businesses to use the LIFO (Last In First Out) inventory procedures
permitted under the tax law, result in a burden of tax on the return from investment that increases with the
rate of inflation. Furthermore, because the tax treatment of nominally denominated assets also ignores
inflation, considerable if poorly understood stress is placed upon the financial structure of the economy in a
period of rapid increase in the price level.
A consequence of this view, that there is a problem and that taxes have something to do with it, has been a
movement to increase the incentive to save and invest by changing the rules. Somewhat oddly, there seems to
be relatively little interest in approaching this by asking what steps would be necessary to correct for
inflation. Nor has there yet appeared an explicit strategy of shifting taxation away from an income and toward
a consumption base, a policy not necessarily related to inflation. Perhaps because of the perceived complexity
of
Â
Page 206
the former and unfamiliarity with the latter, policymakers and the interest groups actively involved with the
issue have been attracted to ad hoc measures.
Most prominent and widely discussed among these have been proposals to allow taxpayers to write off the
cost of acquisition of productive assets, for purposes of calculating income subject to tax, more rapidly than
the economic definition of income would imply. Bills sponsored by House Ways and Means Committee
members Conable and Jones (the "10-5-3 proposal"), by committee chairman Al Ullman, and by the Senate
Finance Committee (the "2-4-7-10 proposal") would in various ways provide for a grouping of assets into
categories, for which relatively rapid write-off for tax purposes would be allowed. Other changes in tax rules
recently enacted or currently discussed with a similar objective of reducing the tax on capital income from
present inflation-influenced levels are lower effective rates on long-term capital gains, exemption of a limited
amount of dividends and savings account interest from individual income taxation, and relaxation of
restrictions on tax deferred saving of the sort now allowed via employer sponsored pension programs,
Individual Retirement Accounts (IRAs) and Keogh Plans.
169
170
These phenomena and policy

choices raise issues of economic
analysis, both theoretical and
empirical, which have attracted
an appropriate amount of
professional attention. Perhaps
the most important are whether
the facts indeed warrant the
conclusion that there should be
more capital formation and
whether, if so, rule changes of
the sort under consideration are
likely to call it forth. This paper
dodges those difficult questions.
It considers rather the problem of
designing tax and related rules to
promote capital formation. 1
To put the matter somewhat more precisely, I take up in this chapter the characteristics of and interactions
among measures to effect savings and investment incentives (henceforth "S-I incentives") in the context of an
income tax system that is inadequately indexed for inflation. Although the issues have been separately
addressed many times, a treatment that is at once unified and reasonably simple is lacking.2 Furthermore,
existing analyses that may incorporate more realistic detail have failed to appreciate sufficiently what legal
commentators call the "pressures" introduced to the tax system by inflation and by the present ad hoc
measures to deal with it. These pressures are by and large created by the opportunity to make money by
undertaking transactions which have offsetting effects on
Â
Page 207
the balance sheet, but different tax consequences, a process I have referred to as "tax arbitrage." It is these
opportunities for arbitrage profit and how the system eliminates them that I shall stress in the following
pages.
Underlying the analysis in this paper is the view that rules which do not work well in a simple model world
will also not work well in the complex real world. This is the justification for confining attention almost
wholly to situations of no uncertainty, no borrowing and lending constraints, and uncomplicated financial
relationships. There is no doubt that the extreme sorts of outcomes that emerge in simple models, such as
conclusions that a taxpayer's wealth will consist all of one asset, or involve large borrowing, will often be
prevented in actuality by information and other uncertainty-related costs. It is clear that a proper treatment of
uncertainty is necessary to a full understanding of capital market equilibrium. However, arbitrage among
relatively risk-free assets represents a significant subset of the transactions that must be dealt with by the tax
system. The qualitative character of outcomes predicted by the certainty models is observed in the real world.
Furthermore, the learning process is obviously still incomplete and cheap high-speed information handling is
increasingly extending sophisticated tax arbitrage to a wider market. Thus problems now manifested in the
"aggressive" behavior of a few taxpayers are quite likely to be seen more generally in the future.
In section I, I sketch out the criteria applicable to choice among savings and investment incentives and offer a
classification of measures differing according to the transactions to which they apply and the way they work.
Section II looks at the major elements of this structure in the context of stable prices, while section III takes
up the difficult problems posed by inflation. There is a brief summing up in section IV.
170
171
I
Background on Savings and Investment Incentives
First, I shall use the terms savings and investment more or less interchangeably, but to the extent there is a
distinction, investment refers to the acquisition of a real asset, while saving refers to the foregoing of
consumption. Second, as to what I mean by savings and investment incentives, remember that the
background for this discussion is an income tax system. An income tax, by definition, embodies a
Â
Page 208
saving or investment disincentive in that it creates a divergence between the rate of return on investment and
the yield to the saver. Inflation may increase this tax wedge. Thus the incentives we are considering are
relative to existing disincentives, whatever their merits.
It is important to be aware of a resulting ambiguity of the term savings or investment incentive as applied to
the measures studied here. As I have stressed elsewhere [1980], in view of the government's budget constraint
an incentive such as the investment tax credit may be bought at the price of higher tax rates than would
otherwise be possible. The net effect may be an increase in the tax wedge applicable, at least for some savers.
Criteria for Choice Among Savings and Investment Incentives
At the risk of banality, I would suggest that the criteria for choice among S-I incentive measures can be
summarized by the familiar trinity of equity, efficiency, and simplicity.
Equity
As usual, equity is the most difficult criterion to deal with. To start with, one of the objectives of currently
considered S-I measures is to offset inequities that have been perpetrated by inflation, and thus concentration
on the static characteristics of rules may miss part of the point. This is a particular aspect of the more general
problem of distinguishing between transition and steady-state effects. It is regrettable that I shall have most to
say about the better-understood steady state properties of S-I incentives.
As we shall see, there are essentially two sorts of available S-I incentives. The firstwhich I call C-tax
measurestend to equalize the rate of return received by savers at all levels of economic well-being. The
secondwhich I call direct grant measurestend to raise all rates of return received by savers, relative to the
social rate of return, but do not alter the differentials among after-tax rates of return on savings characteristic
of a graduated income tax. Even with a fixed structure of tax rates the difference in relative individual
welfare involved is not a priori certain, as it has to do with the lifetime pattern of earnings and consumption,
but it is plausible that the former class of incentives (basically deductions from the income tax base) is
relatively more favorable to high bracket taxpayers than the
Â
171
172
Page 209
latter (basically investment grants or credits). If desired, such differences could be offset by adjustments in
marginal tax rates.
This would leave horizontal equity differences between the two approaches, and these are essentially the
same as those involved in choosing between a consumption base and an income base for taxation. Relative to
a consumption tax an income tax penalizes those who postpone consumption, whether because they simply
prefer to do so, or because their labor earnings, gifts, transfers, etc., occur early in life. 3
Efficiency
As Auerbach [1979b] has emphasized, we are dealing here with a problem in the economics of the second
best. This means that, for example, a measure creating a divergence in the real rates of return on investment
in machines of different durability may not be inferior to a measure, similar in most other respects, that
causes these rates to be equalized. Our analysis of S-I incentives will simply point out the distortions they
engender.
There are three margins of trade-off of particular interest. The first is that between present and future
consumption. An income tax introduces a wedge between the trade-off available to individuals, through
borrowing and lending or through real investment and production, and that available socially via the
production processthe social rate of return on investment. It thus inherently involves an inefficiency, albeit a
potentially second best one, since revenue must be raised somehow. When an income tax is assessed at
different rates on different individuals, there is also a violation of exchange efficiency: different individuals
have different marginal rates of substitution of present for future consumption. As has been mentioned, S-I
incentives of the C-tax type ameliorate both sorts of wedges, while direct grant measures simply shift the
distribution of private rates of return, after taxes, upward relative to the social rate of return.
The second margin is between investment in different forms or different sectors. This is a matter of
production efficiency. If the real social rate of return is not the same in two activities, an opportunity exists to
increase consumption in all periods by shifting resources from the low yield to the high yield activity. Such a
situation commonly arises under an income tax with the weaknesses typical of actual tax accounting systems.
Individuals have an incentive, for example, to push investment in owner-occupied housing to the point
Â
Page 210
that the marginal social (and private) return equals the after-tax return in fully taxed industries. Similar
inefficiencies are predicted by theory when S-I incentives are limited to particular classes of assets, for
example, manufacturing equipment, or to particular industries. A particular case is the difference in social
rate of return to capital of different durabilities that is predicted when an investment credit is not
appropriately varied with the service life of the asset. At a more refined level of analysis, similar comments
would apply to the risk characteristics of real investments.
A third significant margin, between different assets in the household's portfolio, also involves questions of
risk bearing. Efficient allocation of risk will normally imply a certain division of each individual's portfolio
among real asset types and among financial instruments such as bonds and shares. Equilibrium portfolios
with taxes may be expected to diverge from efficiency, and S-I incentives often worsen the distortions
characteristic of the existing income tax, typically in the direction of increasing debt-equity ratios in the
aggregate and concentrating debt ownership relatively in low tax bracket hands (including life insurance and
pension fund portfolios). 4
172
173
The degree to which the S-I incentive displaces the existing income tax presents a further aspect of
efficiency. It is reasonable to suppose that the basic outlines of the existing tax system will be maintained,
with the relatively minor addition of S-I incentive features. It is also the case that the existing system has
solved badly many problems of measuring income from capital, notably in the treatment of owner-occupied
houses, accruing capital gains, and state and municipal bond interest, and in the absence of inflation
adjustments. If the effect of the S-I incentive is to reduce reliance on these aspects of the income tax, it also
diminishes the inefficiencies associates with such defects of income measurement.
Simplicity
It is difficult to say very much in general about the potential tendency of S-I incentives to complicate further
or to simplify compliance with and enforcement of the tax law. There is a certain risk that S-I incentives will
bring with them hard-to-administer rules to prevent ''abuse," as in the present rules disallowing deduction of
interest traceable to the purchase or holding of tax exempt bonds, or will require inherently complex
calculations, as in the case of some aspects of inflation adjustment.
Â
Page 211
One desirable characteristic that might be included under this heading is the degree to which an S-I incentive
automatically adjusts to a changing rate of inflation. Arguments in favor of proposals, for example, for
accelerated depreciation, often turn on their ability to offset the current rate of inflation. The measures
typically will be inappropriate for other rates of inflation, implying the necessity for further rule adjustments
when conditions change. Other S-I incentives may be more or less robust to varying inflation rates.
A Classification of Savings and Investment Incentives
As has been indicated, currently employed or discussed measures to encourage saving and investment can,
with a little license, be placed in two broad categories, the class of consumption tax ("C-tax") rules and the
class of direct grant rules. These rules in turn may be applicable to either real or financial assets, and they
may apply to the purchase or sale of assets (a stock notion) or to the yield from assets (a flow notion). This
generates an eight-way classification.
The usual approach to implementing a consumption tax base ("standard" C-tax treatment) is to permit the
taxpayer to deduct from a conventional income tax base the net purchase of assets during the accounting
period. Various S-I incentive measures have this character, notably including accelerated depreciation of real
assets (the standard C-tax treatment would carry acceleration to the logical extreme of immediate expensing).
Contributions by employers to a qualified pension plan on behalf of employees are subject to the standard
C-tax treatment, since the procedure is equivalent to paying out the contributions in wages and allowing the
employees to deduct the amounts saved in this form. Subsequent pension dissaving upon retirement is then
included in the employees' income tax base. Similar rules apply to saving through Keogh Plans and IRAs,
further examples of S-I incentives of the C-tax type.
A classical consumption tax, levied at a constant rate over time for a given taxpayer, is equivalent in its effect
to exempting the yield from saving or investing from the income tax, and I refer to this as the "alternative"
method of implementing a consumption tax. This approach is also used in S-I incentives, particularly in the
deferral of tax on capital gains accruing in either real or financial assets and in the reduced rate of tax
imposed on such gains upon realization. Reducing the rate of corporation income tax can also be viewed as
Â
173
174
Page 212
belonging in this class of S-I incentives with respect to real assets. The exclusion of dividend and interest
receipts (up to a limit) from individual income tax represents an application of the alternative consumption
tax treatment to financial assets. One might also include here the exemption of state and municipal bond
interest, although this is evidently not a measure designed to encourage saving and investment generally.
In addition to these two consumption tax approaches on the markets for real and financial assets, we can
distinguish incentives having the character of a direct grant, which is not subject to tax or equivalent to a
deduction from the income tax base. The prime example of this in the U.S. is the investment tax credit (ITC),
which provides the investor (who has sufficient tax liability), in effect, a cash grant equal to a fraction of the
cost of a real asset. Unlike the closely related techniques of accelerated depreciation or immediate expensing
of investment outlays, the subsidy provided by the ITC is independent of the investor's marginal tax rate.
There is no program in the U.S. system obviously corresponding to this with respect to saving in the form of
financial assets. In other countries, there exist direct subsidies to saving of a character similar to the U.S.
investment tax credit: the public treasury supplements individual savings by direct grants, independent of the
recipient's marginal tax rate.
For completeness, we may note the possibility in principle of S-I incentives of the yield exemption type
analogous to the investment tax credit. Such measures would involve providing the owner of real or financial
assets with an extra return not subject to income taxation. We would then have four approaches to the subsidy
of each of the two asset types, real and financial, a total of eight hypothetical subsidy techniques. Table 7.1
displays the eight-way classification. Cells 6, 7 and 8 appear to be essentially empty in the U.S. today.
In the next section, we look at the way these different incentive measures work, and interact, under conditions
of stable prices (or well-indexed income measurement rules, regardless of inflation).
II
Savings and Investment Incentives in the Absence of Inflation
Consumption Tax Incentives
To understand the way the different incentives work, how they differ, and how they interact, we are best
served by considering their
Â
Page 213
Table 7.1
A classification of savings and investment incentives in an income tax system
Asset type
S-I incentive type Real Financial
Consumption tax Standard C-Tax rules (1) (2)
rules Accelerated Qualified pension saving
depreciation
Expensing research
and development
Alternative C-Tax rules (3) (4)
Capital gain special Capital gain special rules
rules
174
175
Direct grant Grant for purchase of assets (5) (6)

Investment tax credit Savings premium programs
(not U.S.)
Supplement to asset yield (7) (8)
(after taxes) No known examples No known examples
application in a context with a minimum of complication. Therefore, we focus on the case of an investment
with a risk-free return. To start with, assume away also the graduated structure of tax rates.
It will be sufficient, furthermore, for most of our purposes to study the impact of the various rules on one
particular sort of real investment opportunity, the exponentially decaying machine. This is the model made
familiar by Jorgenson and colleagues. 5 A new machine costing one dollar produces at a rate of output valued
at c, after allowing for payments to cooperating factors, and this is thus the rental rate a producer would be
prepared to pay for the use of the machine. If the machine is of durability δ its rate of output declines at the
constant relative rate δ; the smaller δ, the more durable the machine. The output rate of an s-year old machine
is thus ce-δs. Since an s-year old machine is just equivalent to e-δs new machines, economic depreciation takes
place at rate δe-δs. It is generally assumed that machines of different durabilities are in use at a given time,
and their output rates will differ. When it is necessary to be explicit about this I write c(δ) for the output rate
of a new machine of durability δ.
Simple wealth maximizing considerations will determine who will wish to own machines under various
conditions. If the market rate of interest is given by i, a capitalist subject to a marginal tax rate m, will base
his borrowing and lending on the after-tax interest rate, (1 - m)i. Such a taxpayer will be willing to offer for a
new machine
Â
Page 214
any amount up to the discounted (at rate (1 - m)i) sum of rental payments on the machine, net of taxes, plus
depreciation allowances. This demand price for the real asset is a result of pure arbitrage considerations and
has nothing to do with the capitalist's time preference or propensity to save. Since the supply price of a
machine is 1, the elimination of arbitrage profit requires
Explicit integration leads from (1) to the familiar condition of equilibrium
We may describe as the social rate of return, r(δ), on an asset of durability δ the internal rate of return on a
unit of consumption foregone. That is,
Solving explicitly gives us
Conditions (2) and (4) tell us that in equilibrium with an income tax the social rate of return on investment in
machines of different types is the same. The allocation is thus characterized by production efficiency. 6 While
175
176
the common social rate off return in this equilibrium equals the interest rate, i, the savers receive a lower rate
of return, (1 - m)i = (1 - m)r, whether they save in the form of financial or real assets. Note that as the
marginal tax rate m does not appear in equilibrium condition (2), this analysis would continue to hold with
different tax rates applicable to different capitalists.
Accelerated Depreciation
Consider now the way the accelerated depreciation for tax purposes influences the equilibrium outcome.
There is no single interpretation to be given to this notion, but a natural approach in this context is to assume
that in reckoning income tax the capitalist is allowed to treat the machine with actual durability parameter δ
as though it were of durability δ*. For example, since any likely measure of the average effective or expected
lifetime of an asset of type δ will be inversely
Â
Page 215
proportional to δ, allowing investors to assume δ*= 2δ could be taken to represent a halving of service lives
for tax purposes.
If the taxation of interest is as before, the no-arbitrage equilibrium condition with accelerated depreciation is
given by
which reduces to relationship (6) among rental rate, interest rate, and depreciation and tax parameters:
We may verify by substitution that if tax and economic depreciation are the same (δ = δ*), condition (6)
reduces to (2). Increasing δ* relative to δ reduces the equilibrium rental rate, given i, and results in the return
to savers, (1 - m)i, being higher in relation to the social rate of return, c - δ.
If accelerated depreciation is carried to the extreme of instantaneous write-off, δ* = ℜ∝, equilibrium

condition (6) reduces to
This is the characteristic equilibrium condition for a flat rate income tax system in which real investment is
given standard consumption tax treatment. Since c - δ is equated over all durabilities, this equilibrium is
characterized by production efficiency, and since c - δ = (1 - m)i, the return to the saver, (1 - m)i, is just
equated to the social yield on real investment. Note that in this equilibrium the market interest rate exceeds
the social rate of return by the factor 1/(1 - m), so that the tax on interest just takes away the excess over the
social return.
Returning to condition (6), we may inquire about the relationship between δ and δ* needed to assure that in
equilibrium the social return, c(δ) - δ, is equated at some value for all asset types, a condition for production
efficiency. If we let δ* (δ) stand for the depreciation rate allowed under the tax laws when the true
depreciation rate is δ, a little algebra shows that production efficiency requires a particular
relationship,
176
177
Â
Page 216
It is assumed here that the rate of return to the saver, (1- m)i, is below the social rate of return on investment,
.
Condition (8) may be more

readily interpreted if we state
the objective of the accelerated
depreciation scheme to be
obtaining a specified
proportional difference, β,
between the interest rate and the
common social rate of return on
investment, that is, to effect in
equilibrium the relationship
. With this
substitution, (8) becomes
(assuming β less than m)
Increasing the degree of acceleration toward effectively eliminating the tax wedge between private and social
return toward effectively eliminating the tax wedge between private and social return involves setting β
closer to m. From (9), we see this does involve raising δ* toward ℜ∞, but (9) also tells us that to avoid
inefficiency in the allocation of investment it is necessary to add a term, related to the interest rate, that itself
tends to ℜ∞.
To get some sort of feel for the inefficiency which might be engendered by failing to calibrate the tax
depreciation appropriately to the interest rate and the applicable marginal tax rate, consider the particular case
described in table 7.2.
The first column of table 7.2 shows the true depreciation rate of assets. The second shows the tax
depreciation rate required to obtain the effect of a 50 percent relief from a 50 percent marginal tax rate, if the
before-tax interest rate is 12 percent. Notable is the fact that the ratio of tax depreciation to true depreciation
rate, near 2 for assets
Table 7.2
Illustrative acceleration schedule and effects of using simple scaling
Actual Social return (c - δ)

b
depreciation rate Neutral accelerated when δ* = 2.6δ
a
(percentage) depreciation, δ* (δ) (percentage) (percentage)
2 10 10.3
5 16 9.5
10 26 9.0
50 106 8.5
177
178
Assumptions: i = .12, m = .5, β = .25
a. Calculated according to text expression (9).

b. Derived from text expression (6).
Â
Page 217
with an expected lifetime of two years, rises to 5 for assets with expected 50-year lives. Depreciation of
long-lived assets must be "more accelerated" than that of short-lived assets. Failing to calibrate the degree of
acceleration in this way, by employing instead a simple proportional increase in depreciation rates,
disadvantages more durable relative to less durable assets.
The third column of table 7.2 illustrates this by showing the social rate of return in equilibrium on assets of
different durabilities when instead of the usual acceleration scheme a simple proportional shortening of lives
is employed. In this example the proportionality factor is set to achieve the same incentive effect as the
neutral pattern for assets with an expected life of 10 years. The result is a spread of roughly two percentage
points, or 25 percent, between the equilibrium social return on the 2-year asset and that on the 50-year asset.
This difference is the pure social gain that could be obtained at the margin by shifting investment from the
least to the most durable assets in the table.
In assessing whether a given differential in rates of return or tax on rates of return is "large," it is well to keep
in mind the proverbial power of compound interest. With 50 years of reinvestment at 10 percent, $1
accumulates to nearly $150; at 8 percent it accumulates to a little over $50. It thus may make sense to be
concerned about differences in equilibrium rates of return that appear to be small.
Before leaving this exercise in calculating tax depreciation rates we should note that the assumption we have
employed of a fixed interest rate involves a sort of self-contradiction. In view of the arbitrage potential, a
fixed interest rate implies a fixed incentive to save, unless the tax rate declines. If the new equilibrium is to
involve a larger amount of consumption foregone, it will presumably require a higher interest rate.
Presumably also, the tax rate will need to be higher than otherwise to cover the revenue losses due to
accelerated depreciation. If the tax depreciation rule is not well designed, it is possible that it would do no
more than generate a higher interest rate, higher tax rate, and some deadweight loss. 7 Much the same can be
said of any S-I incentive.
While exponential depreciation does not encompass all investment opportunities, we can learn several lessons
from this analysis. We see that a mechanical compressing of the life of an asset for tax purposes will not in
general generate an appropriate balancing of incentives across short- and long-lived assets, that the
adjustment required to
Â
178
179
Page 218
maintain efficient resource use may be moderately complicated to derive, and that it will in general depend
upon the interest rate and the applicable marginal tax rate. However, the problem of designing a schedule of
accelerated depreciation allowances is made to appear simpler in this case than it is when other possible
patterns of the decline of asset value are taken into account. Each alternative requires, in principle, its own
version of (9).
Arguably, this says nothing more than that in this sphere, as in others, there is a problem of choosing a
sufficiently good approximation. A further oversimplification is not so easily dealt with. The marginal tax
rate in the equilibrium condition, (6), and in the expression for a neutral acceleration schedule, (9), is not a
constant. Taxpayers with different marginal tax rates will have different demand prices for the same real
asset. Establishing equilibrium requires assuming constraints on arbitrage (for example, borrowing limits), or
a mechanism that produces its own constraints. We shall return to this issue in the context of an alternative
approach to the application of consumption tax rules to real investment.
Partial Expensing of Real Investment
We know that taxation according to income properly measured, using economic depreciation, does not upset
the aspect of production efficiency which we are studying, nor does taxation according to consumption
principles, which involves immediate expensing of investment outlays. One suspects then that a "mixture" of
the two approaches, appropriately designed, should share this virtue. As Harberger [1980] has recently
argued, such is indeed the case. 8
Specifically, consider the effect of allowing in this system the immediate expensing for tax purposes of a
fraction x of the investment in the machine. This means that for given value of i, the demand price of the
capitalist is increased by xm less the loss in value of depreciation allowances. If the basis for depreciation is
reduced by just the amount expensed, elimination of arbitrage profits implies
Again, explicit integration plus some algebra reduces (10) to
Â
Page 219
Thus, for the case of exponential depreciation (and this generalizes easily to all patterns of depreciation)
equilibrium with a flat rate income tax allowing partial expensing of investment outlays, together with
economic depreciation applied for tax purposes to the unexpended basis, is characterized by production
efficiency in the sense that the social rate of return to marginal investment in capital of all durabilities (in use)
is the same, and given by (1 - xm)i.In this equilibrium the interest rate exceeds the social rate of return:
while the rate of return received by the saver in either form is
179
180
The relationship holding when immediate expensing of all real investment is allowed is found by setting x =
1, whereby (1 - m)i = r. The return to the saver equals the social return. This repeats the result above that
immediate expensing of investment accomplishes the elimination of the tax "wedge" on the return to saving
and thus effects consumption taxation, even though interest income is subject to tax (and interest outlays are
allowed as a deduction). 9
Note, though, that, as in the case of accelerated depreciation, the single marginal tax rate now enters the
equilibrium condition. If the relationship between the interest rate, i, and the equilibrium rental rate, c(δ), is
given by (11), the demand price for a unit machine of a taxpayer with marginal rate m', possibly different
from m, will be given by
This demand price (obtained by evaluating the right-hand side of (10) for m = m', given (11)), derived from
pure arbitrage considerations, will be greater than the supply price, 1, if m' > m and less if m' < m. This
accords with intuition, most clearly for the case x = 1. For we know that the application of consumption tax
principles to the real investment amounts to exempting the yield from tax. Any taxpayer can then be assured
the rate of return r = c - δ. This will
Â
Page 220
just equal the after-tax interest rate for the taxpayer with marginal rate m. For the taxpayer with rate m' > m,
the after-tax return on lending or the cost of borrowing is less. Hence the situation presents an opportunity for
arbitrage profit, with the high bracket taxpayer borrowing to finance the acquisition of machines, each time
earning a pure profit at the expense of the tax system.
Although one must be cautious about a mechanical interpretation of this model, it is instructive to push it to
an equilibrium. Such is permitted by the impossibility of holding negative quantities of the real asset (short
sales of real assets seeming too far fetched). Tax arbitrage profits are eliminated when the marginal rate
applicable to equilibrium condition (11) is mmax, the highest rate among taxpayers. At that point, all real assets
are owned by top bracket taxpayers, while the portfolios of lower bracket households are entirely in loans.
The yield on saving by the top bracket taxpayer is related to the social return according to (13) (with m =
mmax) while taxpayers with lower rates obtain higher after-tax yields in the usual way. Those with a
sufficiently low marginal rate (below αmmax) receive a return on saving in excess of r. When α = 1, all
taxpayers below the minimum tax bracket will have an after-tax interest rate in excess of the social rate of
return on investment.
Adjusting the Treatment of Interest
The phenomenon of lightly taxed assets migrating to the portfolios of high bracket taxpayers is not a new
discovery. What seems yet to be recognized is the possibility of offsetting the effect through varying the rate
of inclusion of interest receipts in (and deduction of interest outlays from) the income tax base. That is to say,
a partial standard consumption tax treatment of real investment will be compatible with an appropriately
partial alternative consumption tax treatment of borrowing and lending. The argument is general, not
dependent on the exponential depreciation assumption, and is simply a matter of discovering the inclusion
rate needed to foreclose arbitrage profit.
Because of the tax rebate, the taxpayer with marginal rate m can finance a fraction αm of the outlay on real
assets by the tax reduction due to the immediate expensing of the fraction α. In a pure arbitrage transaction
the remainder is financed by borrowing at interest rate i. If a fraction γ of interest payments is deductible
180
181
from the tax base together with a fraction (1 - α) of δ, the decline in value of the asset during the period, the
net of tax proceeds from rental will be
Â
Page 221
(1 - m)c + m(γ(1 - αm)i + (1- α)δ). This must cover the sum of interest payments and the actual loss in asset
value to the holder. The former is simply (1 - αm)i. The later differs from δ because of the Treasury's claim
on the value of the asset. A decline of δ in the market value of the asset implies a decline of (1 - αm)δ in the
value of the private owner's share. Thus the no arbitrage profit condition is given by
which reduces to
The coefficient of i in (16) will be identically 1 if
When the proportion of interest allowed as a deduction is given by γ, defined in (17), taxpayers in all brackets
are indifferent between lending and purchasing real assets. The equilibrium interest rate just equals c - δ, the
social rate of return on real investment, and the rate of return received by the saver is (1- γm)i. When α = 1,
the purchasers of real assets are allowed immediate expensing. The corresponding value of γ is zero: interest
is not subject to tax. Both rules are precisely those of a consumption tax system with tax rates constant over
time for each taxpayer, although graduated across taxpayers. When α = 0, real assets are allowed only
economic depreciation as a deduction in calculating the tax base, the principle of true income taxation. In this
case, γ = 1; interest is taxed in full. Table 7.3 shows the values of γ for various combinations of write-off rate
and marginal tax bracket.
The reader may verify that the demand price for a dollar's worth of exponentially depreciating real assets,
given by the right-hand side of (10), is exactly one dollar, independent of the taxpayer's marginal rate, when
the discount rate (1 - m)i is replaced by (1 - γm)i and c is replaced by the equilibrium value i + δ.
Cells 2 and 3
Thus far we have focused on the interactions between standard C-tax treatment of real investment (cell 1 of
table 7.1) and alternative C-tax
Â
181
182
Page 222
Table 7.3
Rate of inclusion of interest income and deduction of interest expense corresponding to various rates of
expensing real investment (percentage)
Fraction of Marginal tax rate
real asset
expensed 10 20 40 50 70
0% 100% 100% 100% 100% 100%

25 77 79 83 86 91
50 53 56 63 67 77
75 27 29 36 40 53
100 0 0 0 0 0
Note. Entries are calculated according to text expression (17) for γ, with δ = fraction of real asset allowed as
immediate deduction and m = investor's marginal tax rate.
treatment of interest (cell 4 of table 7.1). The analysis of standard C-tax treatment of financial assets (cell 2 of
table 7.1), typified by Keogh Plan saving is straightforward. As far as its interaction with the appropriate
treatment of interest is concerned, the argument is basically the same as applied to real investment. The
important difference is that there is now no possibility to equilibrate away tax arbitrage profits through a
differential in the returns on the two forms of saving.
To illustrate, if saving in a
pension plan could be used to
secure a loan subject to
conventional income tax rules,
there would be a tax arbitrage
profit obtainable through
borrowing (to make the example
particularly graphic, let it be
from the pension fund itself) and
depositing the funds, together
with the tax refund due upon
deducting the deposit, in the
pension fund. This involves no
change in consumption and no
real change in portfolio. But the
interest on the borrowing is
deductible, while the interest on
the offsetting ''lending" is not
taxed. Since the underlying asset
is exactly the same, there is no
possibility for this profit to be
eliminated through yield
differentials: the earnings of the
fund are the interest paid on the
borrowing.
Controlling this arbitrage profit requires either direct limits on the arbitrage process or an offsetting change in
the treatment of interest. (Another possibility is found in the endogenous adjustment of marginal rates over
the life cycle; see Bradford [1980, pp. 47-49]). Current rules follow the first approach, setting ceilings on
182
183
annual additions to tax-favored pension savings, restricting the pledging of
Â
Page 223
pension wealth as loan collateral, prohibiting or penalizing withdrawal of funds before retirement, and so
forth. While these rules no doubt inhibit arbitrage profit, they also tend to eliminate the incentive effect of the
programs. An individual who has reached the ceiling confronts the usual (1 - m)i yield on incremental
savings, as does one sufficiently deterred by the restrictions on the pension asset to stop short of the statutory
ceiling. Significantly lifting these restrictions on the standard C-tax treatment of financial assets calls for
associated changes in the taxation of interest, along the lines discussed in connection with the standard C-tax
treatment of real investment.
Discussion of cell (3) of table 7.1 requires a closer look at the rules for taxing sales of depreciable assets
(importantly real estate) and certain tax favored activities, such as ship building and timber production. While
there is no doubt an important story to be told here, it extends beyond my knowledge of the rules. However,
we can readily see that there is a potential for "double-dipping," with assets subject to the incentive effects of
both standard and alternative C-tax treatment. This would have the consequences of production inefficiency
and portfolio distortion as a function of marginal tax brackets of the sort analyzed in connection with
standard C-tax treatment of real investment with no change in the taxation of interest.
Direct Grant Incentives
We can move more quickly through the discussion of direct grant incentives since the basic analytical
approach is now familiar. Furthermore, there is as a practical matter only one program in question, that of a
direct subsidy to real investment expenditures. This happens to be administered through the tax system, in the
form of the ITC. Two features distinguish it from a directly appropriated grant. First, the credit may only be
applied to settlement of positive tax liability. This may mean the subsidy is not available to many firms at a
given time. Certainly it introduces incentives for somewhat artificial financing arrangements and creates
administrative complexity, an aspect of the ITC about which I have nothing new to say. Second, the subsidy
is not reflected in the basis for depreciation. That is, while the cost of an asset to the taxpayer is the price net
of tax credit, depreciation allowances are calculated as though the full price had been paid. This is nothing
more than a mismeasurement of income,
Â
Page 224
and the appropriate analysis is the same as that above in connection with accelerated depreciation. 10
Zero arbitrage-profit equilibrium with an ITC correctly calibrated for durability, together with allowances for
depreciation calculated on the basis of the purchase price of the asset net of tax credit, will be characterized
by production efficiency in the sense we have been using, and neutrality with respect to the portfolios of
wealth holders. The latter property, indeed, depends only on the use of economic depreciation (of the net of
credit asset value) for tax purposes. We may see these conclusions easily in the exponential depreciation case,
where the elimination of arbitrage profits requires
183
184
where k is the fraction of the unit purchase price of the asset available as a credit against tax. Upon explicit
integration and simplification, this condition reduces to
We see that, as m does not appear in the equilibrium condition, arbitrage profit opportunity is eliminated for
individuals in all tax brackets by (19). However, to obtain equality of the social rate of return, c - δ, over all
asset types, δ, in use in equilibrium, a particular relationship, k(δ), between the credit rate and the durability
of the asset is required. It follows from (19) that to obtain a given value, , of the social rate of return requires
k(δ) to satisfy
This relationship is illustrated in table 7.4. In the second column are shown the credit rates necessary to
obtain a social rate of return of 9 percent where the associated interest rate is 12 percent, assumptions
paralleling those of table 7.2. (The savers thus receive a rate of return of (1 - m)12 percent, depending upon
individual marginal tax rates.) It will be seen that a considerable variation is required to implement a subsidy
that does not distort the choice of asset lives. To give an idea of the effect of failing to take into account the
need to calibrate the credit for durability, the third column of table 7.4 shows the social rate of return on
assets of different lives when a flat rate credit of 13.6 percent is applied uniformly. This is
Â
Page 225
Table 7.4
Illustrative neutral tax credit schedule and equilibrium social rate of return using a flat rate
Actual depreciation Neutral ITC rate, Social return (c - δ) when
ak(δ) (percentage)
rate, δ (percentage) k(δ) = .136b (percentage)
2 21.4 10.1
5 17.6 9.7
10 13.6 9.0
50 4.8 3.5
a. Calculated according to text expression (20) for , i = .12.
b. Calculated according to text expression (19) for i = .12, k = .136
the credit rate which induces a social yield of 9 percent for an asset with a 10-year expected life (still
assuming an interest rate of 12 percent). Because of the bias toward short-lived assets this entails, investment
in less durable assets is driven to the point of very low social yield (it would go to zero for an asset with zero
life) while the social rate of return on long-lived assets exceeds 9 percent (tending to 10.4 percent for a
nondepreciating asset).
Certain general conclusions

emerge from this discussion of
direct grant incentives for the
purchase of real assets. First, if
economic depreciation of the net
of credit cost of the asset is
184
185
employed in calculating taxable

income, we do not encounter
differential effects on the
demand price for assets as a
function of the applicable
marginal tax rateeffects which
called for a change in the
taxation of interest under the
consumption tax approach to S-I
incentives. A corollary is that
increasing the rate of credit does
not affect the relative taxation of
the returns to saving as m varies
across taxpayers. In the case of
the consumption tax approach,
as the level of incentive is
increased, the rate of return on
consumption foregone obtained
by all savers tends toward
equality with the social yield on
investment. In the direct grant
approach, as the subsidy level is
increased the whole structure of
individual rewards to saving
goes up relative to the social
return on investment, but there
is no tendency toward equating
the private yields of taxpayers in
different marginal rate brackets.
Designing a credit structure to avoid production inefficiency may be difficult. 11 While in the particular case
of exponential depreciation the formula is not particularly complex, it requires knowledge of both the target
social rate of return and the interest rate that will call forth the private saving necessary to generate precisely
that rate of
Â
Page 226
return. Furthermore, a different formula applies to each pattern of depreciation. 12 In view of the difficulty tax
authorities have in determining the facts about depreciation, a requirement that the credit rules be written
with a detailed knowledge is a severe one.
Finally, we may note that the consumption tax approach to S-I incentives tends to substitute rules with few
measurement problems for the income tax rules which are subject to many problems. The greater the C-tax
type of incentive, the less important are the shortcomings of the income tax residual. The direct grant
approach does not share this property.
III
Savings and Investment Incentives with Inflation
Background on Inflation and Taxation
185
186
The previous section considered the properties of various S-I incentive measures under conditions of stable
prices. Inflation brings with it new problems of income measurement.13 An ideal indexing system would
solve those problems, and if we had such a system, the preceding analysis would be all that were needed. We
do not, however, and the notion that an extra tax on the reward to saving is a consequence of inflation has
motivated much of the recent movement to enact S-I incentives. In this section we consider, still in the simple
model, the consequences of steady state inflation and the effectiveness of various measures to offset it.
To start with, let us review the way in which an unindexed tax system affects the equilibrium in the market
for real assets, first in a flat rate tax system. With a steady rate of inflation π, the nominal flow of rentals
obtained from a machine of age s is ce(π-δ)s, while under historic cost depreciation the allowance for tax
purposes is δe-δs. The no-arbitrage-profit condition becomes
Provided π is less than δ + (1 - m)i, this condition can be reduced to
When π = 0 this reduces in turn to condition (2).
Â
Page 227
We can see from (22) that with historic cost depreciation and no indexing of interest inflation influences both
the relative social yields from assets of different durabilities and the real return received after taxes by savers.
The latter is given by (1 - m)i - π, and thus whether it is increased or decreased by inflation depends upon
whether the equilibrium interest rate increases by more or less than 1/(1 - m) per point increase in inflation. 14
Letting stand for the real interest rate, such an adjustment would imply
An adjustment of this magnitude is

apparently counter-factual in the
U.S. recently. But we might ask
whether it is likely even in our
simple model. Substituting (23)
into (22), a little algebra allows us
to conclude that it would imply
inflation would lead to no change
in the equilibrium social rate of
return on nondepreciating
machines (δ = 0), but an increase
in the equilibrium social rate of
return for all less durable
machines, rising monotonically
with δ to for
instantaneously depreciating assets
(δ = ℜ∞). If the law of diminishing
returns applies independently to
186
187
investments of different
durabilities, this condition would
imply a reduction in the capital
stock in the aggregate. If the
capital stock desired by private
wealth holders is a declining
function of the private rate of
return, this is incompatible with the
assumption, expressed in (23), of a
constant value of (1 - m)i - π. In
this sense there is a presumption
that, absent correction of income
measurement, inflation will lead to
a decline in the real private rate of
return, (1- m)i - π, a decline in the
overall capital stock, but an
increase in the stock of the most
durable forms of capital (δ near
0).15
All of the analysis thus far is

based on the assumption of a
single marginal income tax rate,
m. We can see from (22) that
with unadjusted depreciation,
the demand price for investment
goods will vary with the tax
bracket of the investor. The
rental rate in (22), which is the
minimum required to cover
taxes, depreciation, and interest,
declines with m. If the rental rate
and interest rate are such as to
permit a taxpayer with marginal
tax rate m to break even (on
purchase of a machine of given
durability), taxpayers in higher
brackets will have an
opportunity for pure arbitrage
profit through borrowing to
finance real investment while
low bracket taxpayers
Â
Page 228
will want to sell machines and lend. This is a situation similar to that just discussed. In that case equilibrium
in the model required all real investment to be in the hands of top bracket taxpayers. While to elaborate on
the details would require more space than merited, there does not seem to be any analogous way of achieving
equilibrium in this case. Just how actual markets clear is not obvious, but it is safe to say that without
correction of income measurement for tax purposes inflation generates distortions in the composition of both
187
188
the real capital stock and individual portfolios.
Inflation Adjustment
Full inflation adjustment of the accounts would involve, first, converting all dollar amounts in the calculation
of the tax to consistent units, that is, dollars of equal purchasing power. A natural choice of units is current
dollars. Inflation adjustment thus calls for increasing the basis of assets sold in the calculation of capital gains
and in the determination of depreciation allowances, as well as for similar changes in inventory accounting.
These adjustments are conceptually straightforward.
Less obvious are the changes called for in the treatment of interest. In principle, the value of the lender's asset
after the payment of interest is exactly what it was at the beginning of the period. Payments considered
"interest" in the usual income tax rules are intended to have this character, and, by and large, they do when
prices are stable. Thus, if a depositor withdraws the interest paid by a savings bank during a year, the nominal
balance is constant. Regarding this interest as income is thus correct when there is no inflation, but when
there is inflation it overstates the income of the depositor and understates that of the bank by the loss in
purchasing power of the nominal balance over the year. Correcting the accounts calls for associating with
interest payments a sum corresponding to the bank balancepresumably one would call it the "principal"and
allowing the creditor a deduction for the loss in purchasing power of the principal during the year, while
assessing the debtor with additional income in equal amount. 16
Following this procedure, the net of tax interest rate, that is, the discount rate applicable to nominal cash
flows, becomes (1 - m)i + mπ. In the case of steady inflation, this is equivalent to adjusting interest payments
to reproduce the effect of taxing on the basis of the real interest rate, i - π. It may be readily verified that
adjusting both
Â
Page 229
the real and the financial sides

of the accounts in this way
restores the calculus of
equilibrium to full equivalence
to the no inflation case with the
same real interest rate.
Because the adjustment of interest called for involves equal and opposite changes in the tax bases of debtors
and creditors, precisely the same economic effect can be accomplished by a change in the interest rate in a
flat rate tax world. To see this, note that with only depreciation allowances adjusted condition (21) becomes
which reduces to
Since (25) implies c - δ is the same for all δ we conclude that indexing real investment accounting will
restore the property of production efficiency to capital market equilibrium. If, furthermore, the nominal
interest adjusts according to (23), (24) implies in turn that
188
189
which is to say the real rental rate and after-tax real interest rate are unaffected by inflation. Note again the
large responsiveness of the interest rate to inflation required to obtain this outcome. The nominal rate must
increase by enough to cover the inflation premium plus the tax due on that premium.
Unfortunately, if marginal tax rates vary across individuals, condition (25) tells us that portfolio distortion is
still a problem. There will be opportunities for tax arbitrage profits, with high bracket taxpayers borrowing to
buy real assets and low bracket taxpayers selling real assets to lend at interest. The result is pressure toward
the sorting of portfolios along lines already discussed.
To sum up, it is possible by reasonably simple methods to correct for the effects of inflation on the tax base
arising from real investment. Furthermore, this would be sufficient in a flat rate tax system, provided nominal
interest rates were sufficiently flexible. However, with graduated rates, there will be portfolio biases unless
the treatment of interest is adjusted as well. This is a much more difficult matter administratively. Current
discussion of S-I incentives emphasizes their potential to offset inflation. One objection to this is obvi-
Â
Page 230
ous, that an incentive designed to offset one inflation rate will be inappropriate at another. Not so widely
recognized is the fact that none of the measures under consideration addresses the need to simulate correction
of interest transactions. Hence as we consider next the properties of S-I incentives under inflation, we know
that his is one problem which they will not solve.
Savings and Investment Incentives and Inflation
The way in which accelerated depreciation or the investment credit fits into this scheme requires no new
analytical materials. From the discussion thus far, we know that these measures can not provide a general
solution to the problems posed by inflation. The situation calls for a second-best analysis, to determine
whether they could effect an improvement on equilibrium with existing rules.
The principal difficulty seems to

be modeling the effect of inflation
on the interest rate. It is tempting
to deal with this by appealing to
the observation that the interest
rate moves roughly
point-for-point with the inflation
rate. However, consideration of
the problem of setting
depreciation allowances to
maintain a fixed rate of return
under this assumption simply
draws attention to the paradox
that it represents, namely, an
inexplicably low nominal interest
rate. For the adjustment required
is a cut in depreciation
allowances.
To see this, let δ* be the depreciation rate allowed for tax purposes. The zero arbitrage profit equilibrium
condition, under the assumption of Fisher's law, is then
189
190
We can see that setting the tax depreciation allowance below the economic level to the extent of the inflation
rate, that is δ* = δ -π (making the tax depreciation rate negative for the most durable assets), leads the
equilibrium condition to reduce to
and since, by assumption, , where is the real interest rate, this in turn implies
To rationalize Fisher's law thus calls for, in effect, taking into the tax
Â
Page 231
base the purely nominal capital gains on real assets, and under this condition, real investment will be
unaffected by inflation.
The difficulty is in explaining

how the assumed relationship
between interest rate and
inflation rate could characterize
equilibrium without the noted
cut in depreciation allowances.
My conjecture is that the
apparent validity of Fisher's law
results from some combination
of failing to account correctly
for the riskiness of debt and
incomplete adjustment of
markets.
The analysis above of equilibrium under C-tax treatment of real investment and graduated rates suggests we
should anticipate an interest rate adjustment close to that required to maintain the after-tax real return to high
bracket taxpayers. Jorgenson has also recommended working with the assumption that the corporate tax rate,
or a marginal tax rate of roughly 50 percent, would dominate the arbitrage between the real capital and
lending markets. Whereas Fisher's law asserts a point-for-point adjustment of nominal interest to inflation,
this assumption involves a 1/(1 - m) point increase in nominal interest rate per point increase in the inflation
rate. We know that indexing real investment returns will justify this result and lead to efficient investment. In
view of the difficulty of implementing indexation however, we may wish to consider acceleration as an
approximation if the rate of inflation is expected to be constant.
It turns out that for the exponential depreciation case this is relatively simple. After integration and some
algebra the zero-arbitrage-profit equilibrium condition (28), when combined with our interest rate
assumption, can be written as
190
191
where
In order that the equilibrium with inflation rate π be the same as that with zero inflation, tax depreciation, δ*,
should be set to render zero the coefficient of δ on the right-hand side of (30). This in turn requires setting δ*
as a simple multiple of δ according to the formula
Â
Page 232
Note that under this interest rate adjustment assumption an increase in depreciation allowances is called for.
The required adjustment is rather sensitive to the assumed real interest rate, . For example, for m = .5,
, and π = .1, (31) implies the tax depreciation rate should be about three times the economic rate; for
, tax depreciation should be at nearly eight times the economic rate. To get a feel for what this might
mean in terms of service lives, we might interpret the ''life" of an exponentially depreciating asset to be the
point at which some specified fraction (e.g. 7/8) is exhausted. This would imply service lives inversely
proportional to δ. Thus (31) would call for assets to be depreciated (on a declining balance basis) assuming
lives roughly one third of those used in the absence of inflation for assumed equal to 9 percent, one eighth
for assumed to be 3 percent.
Consumption Tax Rules and Inflation
One of the virtues of consumption tax rules is their indifference to inflation. A numerical example will make
this clear. Take the case of an individual with a 50 percent marginal rate who wishes to save $100. Under the
standard C-tax rules, he would nominally "save" $200, but would receive a tax rebate of $100 due to the
resulting deduction, thereby foregoing just $100 in consumption. With stable prices and a 10 percent interest
rate, the $200 would increase to $220 by the end of a year. The individual now has the option of "dissaving"
$220, paying the associated tax of $110 and increasing consumption by $110, thus obtaining the full 10
percent return on postponed consumption. By the alternative C-tax rules of exempting the yield on saving, the
10 percent return is obtained directly.
Now suppose the situation is one of 100 percent per annum inflation. If the real interest rate is unchanged,
$100 set aside in the first period will increase to $220 in the second period. The saver subject to the
alternative C-tax rule of yield exemption obviously continues to receive a 10 percent real yield on
consumption foregone. This is so because the $100 given up in the first period is equivalent to the $200
return of principal in the second epriod, while the $20 interest is equivalent to $10 of first period
consumption. The same outcome obtains for the saver subject to standard C-tax treatment. The $200 "saving"
involves foregoing $100 of consumption in the first period, while "dissaving" of the $440 accumulated by the
second period is
Â
191
192
Page 233
divided between $220 in taxes and $220 of increased consumption, exactly the same terms enjoyed by the
conventional saver with exempt returns.
It is thus the case that the real effect of a flat rate C-tax system is independent of the rate of inflation. The
same thing will hold for a graduated rate system in which the rate structure is corrected for inflation. This
adjustment, the easy part of inflation indexing, is sometimes called in the tax policy jargon "type I indexing,"
and it is the only sort of tax correction that seems to have a significant political appeal. Absent type I
indexing there is a tendency for inflation to subject the future dissaving corresponding to current saving to a
higher rate of tax, and thereby inflation may interact even with C-tax rules.
We may easily verify the propositions about flat-rate C-tax rules under inflation for the case of exponential
depreciation. With the interest rate adjusting point for point with the inflation rate, the no-arbitrage-profit
condition with standard C-tax treatment of real investment and no inclusion or deduction of interest becomes
which reduces, as expected, to
While full application of C-tax rules is simultaneously an S-I incentive and a cure for inflation-induced
measurement problems, partial application of C-tax rules, involving partial expensing of new investment and
partial inclusion of interest income in the tax base, only partially solves these problems. The portion of the
investment that is in effect taxed as income is subject to all of the difficulties due to defects in income
measurement, including those associated with inflation. These include increased effective rates of taxation
with inflation and pressure on portfolio composition, with high bracket taxpayers seeking to borrow from low
bracket taxpayers to finance purchases of real assets. This suggests the continued importance of developing
rules to measure income correctly or to impose a tax burden on investment equivalent to that resulting from
correct income measurement.
Â
Page 234
The Auerbach-Jorgenson First-Year Write-Off Scheme
In this connection it is appropriate to digress from the subject of S-I incentives to consider a suggestion put
forward recently by Alan Auerbach and Dale Jorgenson [1980]. 17 They address themselves to the notion that
correcting depreciation allowances for inflation is too complex for practical administration. Under their
proposal, as an alternative to current deduction of inflation-adjusted depreciation allowances, investors are
allowed a single deduction at the time of acquisition of the asset equal to the present value of the appropriate
stream of real depreciation deductions, where a real discount rate is applied in the calculation. This procedure
is intended to accomplish the effect of indexing historical-cost based depreciation allowances for changes in
the general price levels but to be simpler in implementation. Such complexity as there is is embodied in the
derivation of tables by the tax authorities, specifying the allowance for each type of asset.
The characteristics of this scheme emerge immediately if we note that it involves eliminating the current
depreciation allowances, as seen in the term δe(π-δ-(1 - m)i)s in the cash flow stream of expression (24) for the
current-depreciation adjustment case and replacing them with a lump sum initial deduction of
192
193
where the tildes on π and i

indicate the use of forecasted
values of these variables. It
follows immediately that the
two schemes are wholly
equivalent if the term
is equal to π - (1 -
m)i. The practical promise of
the method, insofar as a
reasonably exact substitute for
inflation indexing is desired,
depends upon determining the
correct value of .
This involves two difficulties. First is the necessity to employ a discount factor taking into account the
individual tax bracket of the investor. Second is the necessity to incorporate the relationship between π and i
into the adjustment. As we have seen this is problematical. The scheme works out precisely if the taxation of
interest is also adjusted for inflation. Then the real after-tax interest rate for all taxpayers will be , and
this replaces in calculation of the first-year depreciation allowance. Since is typically
assumed constant, it is reasonably straightforward to adjust the
Â
Page 235
Table 7.5
Illustrative first-year deduction under the Auerbach-Jorgenson approach
Depreciation rate, δ (percentage) Marginal tax rate, m (percentage)

10 20 40 50 70
2 43 45 53 57 69
(20) (22) (27) (31) (43)
5 65 68 74 77 85
(38) (41) (48) (53) (65)
10 79 81 85 87 92
(55) (58) (65) (69) (79)
50 95 95 97 97 98
(86) (87) (90) (92) (95)
Note: Entries (in percentage) show the percentage first-year write-off equivalent to economic depreciation
allowances, calculated according to the formula , where , the real interest rate, is assumed = 03
Figures in parentheses show the allowance assuming .
allowance for differences in m, if m is known. Table 7.5 illustrates under the assumption . To give an
idea of the sensitivity of the scheme to the discount rate assumed, table 7.5 also shows the allowances for
193
194
For both corporate and individual taxpayers there may be a question about what marginal rate to employ in
setting the first-year allowance. This difficulty also arises in the application of the adjustment of the interest
deduction discussed above to coordinate with partial C-tax treatment of real investment. In that context, it
may be acceptable to use the taxpayer's current marginal rate. In the Auerbach-Jorgenson scheme, however,
more may well be riding on the rate chosen, because it affects the taxation of the entire asset purchase price
and not just one year's yield. Thus problems caused by variations over the life cycle and variations which
might be induced by the deduction itself (shifting the taxpayer to a lower bracket) may require resort to
approximations.
While the discussion thus far incorporates the basic principle of the Auerbach-Jorgenson first-year allowance,
their actual proposal does not discriminate according to the investor's marginal tax rate. Allowances are based
instead on the present value of depreciation deductions using the same discount rate in all cases and thus are
uniform for all taxpayers. One way to view this is as an approximation to the theoretical model. If the
allowance corresponds to the
Â
Page 236
ideal for a 50 percent marginal rate investor, for example, we see from table 7.5 the deduction will be "to
large" for a lower bracket taxpayer, and "too small" for a higher bracket taxpayer. Correspondingly there will
be a tendency for arbitrage to move the real assets toward low bracket portfolios, debt toward high bracket
portfolios. Furthermore, if the allowances are "just right" for the 50 percent investor, they will not be
correctly calibrated with respect to durability for the lower bracket investors, providing relatively too much
incentive to purchase short-lived assets.
The first year allowances summarized in table 7.5 are predicated on the assumption of exponential
depreciation. Other depreciation patterns would, strictly speaking, call for different allowances; otherwise
some inefficiency in the composition of investment would be expected. 18 Much the same issue has already
been discussed above in connection with investment credit. In the present state of knowledge about
depreciation there is unlikely to be anything practical to be done about it.
Because under the Auerbach-Jorgenson scheme the full allowance for depreciation is taken in the first year,
its real value is wholly insensitive to the rate of inflation.19 As we have noted, if interest payments and
receipts are adjusted for inflation, and if the correct value for the real interest rate is employed in the formula,
the Auerbach-Jorgenson first-year allowance just duplicates the effect of indexing annual depreciation
deductions. Otherwise, and in particular without correction of interest payments and receipts, the effect will
not be identical. If the first-year allowances are based on an assumed real interest rate , but nominal interest
is fully taxed and deductible, the no arbitrage-profit equilibrium condition is given by
This equilibrium condition is to be compared with (25), the condition when depreciation is actually indexed
but interest is not adjusted. Even with a single flat rate of tax, the two are not identical unless nominal interest
adjusts according to (23). In addition, it should be stressed that under neither approach to insulating the
accounting for income from real assets from inflation can the system reproduce the effect of adjusting interest
payments and receipts when there are different marginal tax rates. This will always result in pressure for
Â
194
195
Page 237
low bracket taxpayers to sell real assets to lend and high bracket taxpayers to borrow to buy real assets.
The Auerbach-Jorgenson first-year allowance is not, strictly, an S-I incentive measure. It is a procedure for
approximating the indexation of depreciation allowances for inflation. It would, however, be a simple matter
to combine the first-year depreciation allowance with a partial first-year expensing of investment or with an
investment credit and thus realize any desired degree of S-I incentive. A flat credit on the difference between
the cost of the new asset and the first-year allowance would maintain production efficiency. 20 In the case of
partial expensing the entire deduction would be available in the first year, but it would consist of the desired
fractional write-off of the investment, with the remaining fraction eligible for the Auerbach-Jorgenson
allowance. As this approach represents simply a combination of two measures already discussed, it requires
no fresh analysis.
IV
Summing Up
In this chapter, I have suggested an eight-way classification of savings and investment incentives. Perhaps the
most fundamental is the division into the class of consumption tax treatments and the class of direct grants.
Roughly speaking, the former includes measures, such as accelerated depreciation and tax sheltered
retirement savings plans, which increase deductions or reduce inclusions in the income tax, while the latter
includes measures, primarily the investment credit, providing an incentive not directly related to the investor's
income tax circumstances. The difference between the two classes is of importance primarily when tax rates
vary in the population of savers and investors.
It proved possible for us to reach some fairly general conclusions about the way incentives of the two classes
operate. As to the broad characteristics of the outcomes under the two approaches, we saw that increasing
levels of consumption tax incentives tend to lead to a convergence of the returns received by savers toward
the social rate of return on investment. At the same time, because the process has the effect of displacing the
income taxation of the returns to saving, income measurement problems, including inflation correction, tend
to diminish in importance. Consumption tax incentives can also be relatively simple to design. While this
does not hold for accelerated
Â
Page 238
depreciation, we saw that expensing immediately a specified fraction of investment outlay, together with a
reduced inclusion of interest receipts, provides an incentive without distorting either investment choice
(among assets of different durabilities) or portfolio composition (between debt and real asset ownership).
The direct grant approach via the investment tax credit, by contrast, leads to an increase, relative to the social
return on investment, in the rates of return received by savers, but does not bring about their convergence.
Provided the basis for depreciation of assets purchased is reduced by the amount of the credit (and provided
the original depreciation rates are accurate), this approach also maintains neutrality with respect to portfolio
composition. On the other hand, designing the credit to take correct account for differences in asset durability
is relatively difficult. Furthermore, the direct grant approach does not share with the consumption tax
approach to savings and investment incentives the tendency to displace imperfect income measurement rules
as the amount of incentive is increased. In particular, it does not to the same degree reduce the sensitivity of
the tax system to the rate of inflation.
195
196
A point that has been emphasized here is the interaction among the consumption tax incentives. In particular,
it has been stressed that applying consumption tax incentives to real investment without implementing a
corresponding exclusion of interest from tax (and corresponding reduction in interest deductibility) tends to
drive real assets into the hands of the highest bracket taxpayers and to negate the potential of the consumption
tax approach to bring about convergence of the yields on saving. What has not, to my knowledge, been
recognized before is that there is an exact and relatively simple degree of interest inclusion appropriate for
each level of write-off of real investment, a fact that would ease a gradual phasing in of a consumption-type
base should this be the objective of policy.
Inflation upsets the measurement of the yield of both real and financial assets. S-I incentives are commonly
viewed as instruments to offset these measurement problems. However none of them addresses the problem
of correcting interest income. As a result it is difficult to reach clear conclusions about their relative
properties, because it is difficult to model equilibrium in a graduated tax rate system. Without any correction
for tax purposes, a change in the nominal interest rate of 1/(1 - m) points per point of inflation is necessary to
maintain the after-tax yield of a lender (or after-tax cost
Â
Page 239
of a borrower) with marginal tax rate m. Thus an interest rate change that just offsets inflation for one
taxpayer will be too large or too small for most others.
Just where the change settles has a bearing on the predicted effect of measures, such as accelerated
depreciation and the ITC, affecting the taxation of real assets. Theory predicts a concentration of real
investment in the hands of relatively high bracket investors and relatively large changes in interest rates with
inflation. To offset the resulting impact on the efficiency and quantity of real investment, accelerated
depreciation or the ITC offer possible alternatives to indexing depreciation, but they do not correct the
portfolio distortions. Furthermore, observed variation in interest rates does not seem large enough to be
consistent with this view. Fully convincing analysis of the alternatives to indexing may have to await the
sorting out of this puzzle.
Notes
Among the many individuals who helped in the development of these ideas, the author would like, in
particular, to thank Alan Auerbach, E. Cary Brown, Dale Jorgen-son, and Alvin Warren.
1. I have addressed these larger issues in Bradford [1980] Readers may also find helpful the discussion there
of the problems created by inconsistent treatment of different forms of savings For further discussion of the
general issues see King [1980]. For empirical analysis of the U.S. experience see, for example, Boskin
[1978], Eisner [1977], Feldstein [1977a,b; 1980], Malkiel [1979].
2. For an examination of selected investment incentive proposals under consideration recently in the United
States see Hendershott and Hu [1980].
3. For further discussion of this issue see Bradford and Toder [1976].
4. For analyses stressing this efficiency problem see Gordon [1980] and Gordon and Malkiel [1980].
5. See, for example, Jorgenson [1963], Hall and Jorgenson [1967].
6. This has been shown by, among others, Samuelson [1964].
7. For an example see Bradford [1980; pp. 42-50].
196
197
8. Auerbach [1979a] confirms this point for the case of exponential depreciation.
9. This was one of the methods proposed by the Meade Committee for implementing a consumption tax in
the U K., where the income tax is imposed at essentially a single flat rate for the great bulk of taxpayers.
[Institute for Fiscal Studies, 1978; ch. 8].
10. An early treatment of this is found in Brown [1962]; for further details see Bradford [1979]
Â
Page 240
11. E. Cary Brown [1962] shows that one way to achieve the correct calibration of the credit is to structure it
as a flat "net credit," that is, a fixed percentage of the difference between the cost of the machine and the
present value (at the net of tax interest for that investor) of economic depreciation allowances. Since this is
equivalent to calibration of the "gross credit" to asset durability, it is subject to the same problems.
12. For examples of the formulas applicable to other patterns of depreciation see Bradford [1979].
13. For a general treatment see Stiglitz [1980].
14. This required adjustment in i to compensate savers for inflation has been stressed by Feldstein [1976]. For
an attempt to rationalize the fact that interest rates have moved with inflation at most according to Fisher's
law (point for point) in the U.S. in recent years see Feldstein and Summers [1978]. Feldstein, Green, and
Sheshinski [1978] explore models of interest determination under conditions of inflation. Stiglitz [1980] has
emphasized that with imperfect indexing the incidence of inflation depends upon the measures taken to
maintain adherence to the government's budget constraint. General conclusions about the effect of inflation
thus require modeling the government's reaction.
15. The original version of this paper contained an erroroneous conclusion that there would be an increase in
the stock of the least durable capital (δ near infinity) as well. The corrected version can be read as agreeing
with Auerbach's [1979a] view that inflation biases the choice of asset life toward assets with greater
durability and conflicting with Fedlstein's [1980] view that inflation biases the pattern of investment in favor
of short-lived assets.
16. For detailed discussions of the problems of indexing the income tax for inflation see Aaron [1976],
Shoven and Bulow [1975; 1976], Fabricant [1978].
17. According to Joseph Pechman [1980] the Auerbach-Jorgenson proposal is a rediscovery of an idea of
Nicholas Kaldor.
18. Different depreciation patterns would also require different "recapture rules" applicable to sale of assets, a
subject we shall not pursue here.
19. There is a frequently overlooked but practically highly significant condition in this statement: the investor
must have sufficient taxable income gross of the depreciation deduction to make use of it.
20. This is thus a "net credit," as discussed in ftp. 11. I would like to thank E. Cary Brown for pointing out to
me the natural adaptivity of the net credit to the Auerbach-Jorgenson proposal.
References
Aaron, Henry J. 1976. ed. Inflation and the Income Tax. Washington, D.C.: The Brookings Institution.
Auerbach, Alan J. 1979a. "Inflation and the Choice of Asset Life." Journal of Political Economy 87, no. 3
(June): 621-638.
197
198
. 1979b. "The Optimal Taxation of Heterogeneous Capital." The Quarterly Journal of Economics 93, no. 4
(November): 589-612.
Â
Page 241
. 1980. "Inflation and the Tax Treatment of Firm Behavior." NBER Working Paper no. 547, September,
Cambridge, Mass.: National Bureau of Economic Research.
. 1981. "A Note on the Efficient Design of Investment Incentives." The Economic Journal 91, no. 361
(March).
., and Jorgenson, Dale. 1980. "Inflation-Proof Depreciation of Assets." Harvard Business Review
(September-October).
Boskin, Michael J. 1978. "Taxation, Saving, and the Rate of Interest." Journal of Political Economy 86, no.
2, part 2 (April): S3-S28.
Bradford, David F. 1979. "The Case for a Personal Consumption Tax." In Joseph A. Pechman, ed. What
Should Be Taxed. Income or Expenditure? Washington, D.C.: The Brookings Institution, pp. 75-113.
. 1980a. "The Economics of Tax Policy Toward Savings" In George M. von Furstenberg, ed. The
Government and Capital Formation. Cambridge, Mass.: Ballinger, pp. 11-71.
. 1980b. "Tax Neutrality and the Investment Tax Credit." In Henry J. Aaron and Michael J. Boskin, eds. The
Economics of Taxation. Washington, D.C.: The Brookings Institution, pp. 281-298.
., and Toder, Eric. 1976. "Consumption vs. Income Base Taxes: The Argument on Grounds of Equity and
Simplicity." Proceedings of the National Tax Association, pp. 25-31.
Eisner, Robert. 1977. "Capital Shortage: Myth and Reality." American Economic Review 67, no. 1
(February): 110-115.
Fabricant, Solomon. 1978. "Accounting for Business Income Under Inflation: Current Issues and Views in
the United States." The Review of Income and Wealth 24, no. 1 (March): 1-24.
Feldstein, Martin S. 1976. "Inflation, Income Taxes and the Rate of Interest." American Economic Review 66,
no. 4 (December): 809-820.
. 1977a. "Does the United States Save Too Little?" American Economic Review 67, no. 1 (February):
116-121.
. 1977b. "National Saving in the United States." In Eli Shapiro and William L. White, eds. Capital for
Productivity and Jobs. Englewood Cliffs, N.J.: Prentice-Hall.
. 1980. "Inflation, Tax Rules and Investment: Some Econometric Evidence." NBER Working Paper 577,
October, Cambridge, Mass.: National Bureau of Economic Research.
. 1981. "Adjusting Depreciation in an Inflationary Economy." National Tax Journal 34, no. 1 (March).
.; Green, Jerry; and Sheshinski, Eytan. 1978. "Inflation and Taxes in a Growing Economy with Debt and
Equity Finance." Journal of Political Economy 86, no. 2, part 2 (April): S53-S70.
., and Summers, Lawrence. 1978. "Inflation, Tax Rules, and the Longer Term Interest Rate." Brookings
Papers on Economic Activity, 1, pp. 61-109.
198
199
Â
Page 242
Gordon, Roger H. 1980.

''Inflation, Taxation, and
Corporate Behavior." NBER
Working Paper 588, December,
Cambridge, Mass.: National
Bureau of Economic Research.
., and Malkiel, Burton G. 1980.

"Taxation and Corporate
Finance." NBER Working
Paper 576, November,
Cambridge, Mass.: National
Bureau of Economic Research.
Hall, Robert, and Jorgenson, Dale. 1967. "Tax Policy and Investment Behavior." American Economic Review
57, no. 3 (June): 391-414.
Harberger, Arnold C. 1980 "Tax Neutrality in Investment Incentives." In Henry J. Aaron and Michael J.
Boskin, eds. The Economics of Taxation. Washington, D.C.: The Brookings Institution, pp. 299-316.
Hendershott, Patric H., and Hu, Sheng-Cheng. 1980. "The Relative Impact of Various Proposals to Stimulate
Business Incentives." In George M. von Furstenberg, ed. The Government and Capital Formation.
Cambridge, Mass.: Ballinger, pp. 321-336.
Institute for Fiscal Studies. 1978. The Structure and Reform of Direct Taxation Report of a Committee
Chaired by Professor J. E. Meade. London: Allen and Unwin.
Jorgenson, Dale W. 1963. "Capital Theory and Investment Behavior." American Economic Review 53, no. 2
(May): 247-259.
King, Mervyn. 1980. "Savings and Taxation." NBER Working Paper 428, January, Cambridge, Mass.:
National Bureau of Economic Research.
Malkiel, Burton G. 1979. "The Capital Formation Problem in the United States." Journal of Finance 34, no. 2
(May): 291-306.
Pechman, Joseph. 1980. "Tax Policies for the 1980's." Tax Notes 11, no. 25 (December 22): 1195-1208.
Samuelson, Paul A. 1964. "Tax Deductibility of Economic Depreciation to Insure Invariant Valuations."
Journal of Political Economy 72 (December): 604-606.
Shoven, John B., and Bulow, Jeremy I. 1975. "Inflation Accounting and Nonfinancial Corporate Profits:
Physical Assets." Brookings Papers on Economic Activity, 3, pp. 557-611.
. 1976. "Inflation Accounting and Nonfinancial Corporate Profits: Financial Assets and Liabilities."
Brookings Papers on Economic Activity, 1, pp. 15-57.
Steuerle, Eugene. 1980. "Is

Income from Capital Subject to
Individual Income Taxation?"
OTA Paper 42, October,
199
200
Washington, D.C.: U.S.

Treasury Department.
Stiglitz, Joseph E. 1980. "On the Almost Neutrality of Inflation: Notes on Taxation and The Welfare Costs of
Inflation." NBER Working Paper 499, July, Cambridge, Mass.: National Bureau of Economic Research.
Â
Page 243
8
The Incidence and Allocation Effects of a Tax on Corporate Distributions
1
Introduction
There has recently developed in the United States an increasing interest in the "integration" of individual and
corporation income tax systems. As seen in the policy discussions of this issue, the central problem with the
existing "classical" system (whereby a tax is levied on corporation profits after payment of interest,
independently of distributions) is the double taxation of dividends. Double taxation arises because the income
generated by corporate investment financed by equity is taxed once at the level of the corporation, currently
at a marginal rate of 46 percent for corporations with significant amounts of income, and then again, at rates
ranging from zero to 70 percent on the income tax return of the individual shareholder on the portion of the
remainder that is distributed in the form of dividends. The resulting penalty on equity finance, it is argued,
distorts the financial and real decisions of corporations. To correct this the usual remedy, which has been
extensively adopted in Europe, is to apply some form of "partial integration," which means to apply a tax
adjustment which is a function of dividend distributions. Most approaches involve either allowing the
corporation to deduct dividend distributions from the corporation income tax base (much as interest payments
are deducted) or allowing the shareholder a tax credit which stands in relation to dividends as the corporation
income tax stands to corporation income net of tax.
Both of these methods have the effect of eliminating the corporation tax on fully distributed earnings, but
they are not otherwise equivalent to the "full" integration of corporation and individual income accounts,
which is commonly regarded as impracticable administratively, even if desirable in principle. 1 The
difference between
Â
Page 244
full and partial integration is in the treatment of retained earnings, which are taxed at a flat rate independently
of the circumstances of the shareholders under partial integration and at the shareholder's marginal income
tax rate under full integration.
The present analysis suggests that the partial integration approach may have gotten the matter backwards.
The problem on this view is not the extra tax imposed on distributions, but the divergence between
shareholder and corporation tax rates on retained earnings. If the market rate of interest is r, an individual
who pays income tax at the marginal rate m earns a rate of return (1 - m)r on his savings in ordinary assets. A
200
201
corporation, on the other hand, is able to accumulate at a rate (1 - c)r on the same investment, where c is the
rate of corporation income tax. The power of compound interest being what it is, we would expect the choice
of asset types to be dominated by the difference in the rates of return, even if a price in the form of a tax on
distributions must be paid to obtain the favorable rate.
In this chapter I consider the incidence and allocation effects of a tax on distributions, as distinguished from a
tax on the income of individuals or corporations. To isolate the issue I abstract from the taxation of income,
as well as from uncertainty, nonlinearity of tax rules, and multiplicity of individual tax regimes. This chapter
analyzes the behavior of a system with a flat rate tax on corporate distributions to shareholders and no other
taxes. It is thus concerned with a kind of polar case of the double taxation of dividends.
In view of the abstraction from uncertainty, it might appear to be an open and shut case that imposing a tax
on distributions will lead to a flight to the corner solution of all-bond finance. Or at least, if this is not
optimal, it would appear obvious that good financial policy toward equity calls for retained earnings only, so
that stockholders can take their returns in untaxed capital gains. Furthermore, one might expect that should
equity finance continue to exist, the investment criterion of the firm would be affected. The conclusion of the
analysis is that none of these results obtains. While the issue of new equity may become unattractive (unless
this form of negative distribution is subsidized at the same rate that positive distributions are taxed), all those
financial plans involving non-negative distributions, over which indifference prevails in the absence of the
tax, continue to be indifferent. The all-debt policy acquires no special place, nor are dividends discouraged in
favor of retained earnings. As for real investment, the criterion used by the firm remains the
Â
Page 245
equating of marginal returns to the rate of interest, regardless of the chosen level of debt finance.
These points emerge from a careful treatment of the determination of the value of equity claims. For if, as is
often assumed, a dollar of retained earnings leads to a one-dollar increase in the market value of equity,
shareholders could obtain their returns free of the tax on distributions. The conclusion of the present analysis,
based on a rational expectations model of asset pricing, is that an extra dollar of retained earnings (with a
corresponding reduction in borrowing) increases the value of equity by one minus the rate of tax on
distributions. This is why shareholders are indifferent between dividends and retained earnings, and why real
investment continues to be determined by an implicit trade-off between it and retiring debt.
As for incidence, an implication of the equity valuation result is that an unanticipated change in the rate of
taxation of distributions affects the market value of existing equity claims and, hence, the wealth of their
holders. For a given rate of interest and for given corporate indebtedness and capital stock, the value of equity
is simply proportional to one minus the rate of tax on distributions. The wealth shifts resulting from a change
in tax rate may constitute the major, even the only, incidence effects. We take up below a particularly striking
case in which an increase in tax rate is accompanied by a one-time government expenditure financed by an
increase in government debt. The incidence of this combination turns out to be confined wholly to the wealth
loss of holders of equity at the time. There is no deadweight efficiency loss, and no burden to be the subject
of subsequent concern.
This explains my earlier remark to

the effect that the partial approach
to integration may be correcting
the wrong problem. While the
present analysis is far too simple
to support policy conclusion, it
does suggest that insofar as partial
201
202
integration amounts to eliminating

a tax on distributions, it may
result primarily in windfall wealth
redistributions, reversing the, by
now, irrelevant wealth changes
that occurred when the tax was
introduced, while leaving the
features of the tax system giving
rise to inefficiency.
The modeling of equity valuation in general equilibrium has some interesting implications beyond those
concerning the neutrality and incidence of the tax on distributions. One is that along an equilibrium path the
total value of the firmthe sum of market values of its debt and equitymay depend on the financial policy of
the firm. In
Â
Page 246
the present model this policy is indeterminate, even though the firm's real capital stock is determinate. The
larger the fraction of equity finance, the lower is the market value of the firm (which is less than or equal to
its replacement cost if the firm is not a net creditor). This contrasts with the well-known Modigliani-Miller
(1958) proposition that the value of the firm is independent of its financial structure, a property holding here
only when the distribution tax rate is zero.
A second interesting feature of the model is the apparently "Ricardian" behavior of government debt. 2 The
inclusion of government debt in the model is not fortuitous. Since corporate financial policy is indeterminate,
so also is the flow of government receipts from the distribution tax. It seemed compelling from the point of
view of both empirical relevance and modeling simplicity to make government expenditure independent of
tax receipts, with variation in government debt issue and retirement taking up the slack. The equilibrium
conditions of the model include a term reflecting the present value of potential receipts from the distribution
tax, comparable and opposite in sign to the quantity of government debt. A cash flow deficit, requiring an
increase in government debt, is exactly offset by an increase in the potential tax receipts term. In effect the
economy does not treat endogenously occurring changes in government debt as wealth changes. This result is
not immediately obvious from the assumptions made. In the model, no one has explicit expectations about
taxes. Furthermore, each generation is concerned only with its own consumption and thus the response to
debt changes does not derive from the combination of extraordinary foresight about the distant future and a
long chain of bequests, a much challenged link in the theoretical argument for the fiscal neutrality of
government deficits [see Barro (1974)]. While the model has been developed primarily to make a particular
point in tax analysis, the lessons here for the analysis of government debt (it may matter whether it changes
endogenously or exogenously) and for the limitations of cash-flow budgeting (it tells us nothing in this
model) seem well worth noting.3
1.1
Intuitive Explanation of the Results
It may assist an understanding of the results to consider the analogy of the tax on distributions with a tax on
all withdrawals from savings accounts opened before some specified date in the past. Were they
Â
202
203
Page 247
bought and sold as are bonds, the market value of such accounts would presumably fall, upon imposition of
the tax, to one minus the tax rate times their face value "replacement cost." Once the tax is in place, there is
no particular incentive for existing account owners to accelerate the pace of withdrawals; a growing value of
balances in such accounts (due to accumulating interest) and, therefore, a growing absolute divergence
between market and replacement values, is fully consistent with optimizing behavior.
In the analysis presented here,

the corporate "vessel" is the
analogue of the savings account,
and the tax on distributions the
analogue of the tax on
withdrawals. The tax is the price
that must be paid to get cash out
of the corporation. Like the
hypothetical tax on bank
accounts, it would be necessary
for a tax on corporate
distributions to be imposed ex
post, or at least as something of
a surprise, for it to find any base
to subject to tax. The usual
analysis of the corporation
income tax, in effect, assumes it
is being imposed before the
system "starts up." The
discussion below illustrates how
profound a difference it may
make to examine instead a tax
on dividends imposed on a
system already in existence.
1.2
Connection with the Existing Literature
The approach taken here bears some similarity to that taken in papers by Stiglitz (1973) and King (1974)
which show that the relationship between income tax rules and optimal corporation financial and investment
policies is more complex than had been previously understood. 4 By analyzing the corporation's choices as
part of a multiperiod optimization problem, these authors are able to treat consistently the interactions
between individual and corporation income tax systems. Both papers conclude that the optimal financial
structure may be indeterminate, given certain relationships among the relative rates of taxation of corporate
income, corporate distributions, interest payments and capital gains, the history of the corporation (in
Stiglitz's analysis), and expected future values of the tax parameters (in King's analysis). For these cases and
for a wide further class as well, the corporation's real investment decisions are unaffected by the taxes.
The present chapter differs from these predecessors in its explicit attention to general equilibrium, including
equilibrium in the asset market in an infinite-horizon world with rational expectations.
Â
203
204
Page 248
Stiglitz devotes little attention to the question of asset valuation. As far as the question of financial structure
is concerned, his analysis is best described as a theory of the small, closely held corporation. His results turn
on the tax-technical matter of whether a corporate distribution is classified as a "return of capital" (and hence
free of individual income taxthe tax on distributions). The corporations accounting for the vast bulk of
corporate assets in the United States rarely approach the condition of having distributions which qualify as
return of capital. As a consequence, the necessary conditions for optimality of financial policy [Stiglitz (1973,
p. 13)] are not fulfilled empirically by the large public corporations, essentially permanent, horizonless
institutions, constituting the major share of the corporate sector. A different theory is needed to explain their
behavior, and it can probably neglect the special treatment of return of capital, as is done here (and in King). 5
Whereas Stiglitz looks at the corporation as the vehicle of an individual investor, King views the problem of
corporate decision-making as that of maximizing the present market value of equity. King's paper can be
regarded as a theory of the market value of a firm setting financial and investment policies optimally with
respect to present and expected tax parameters. Presumably because his analysis is in a partial equilibrium
setting, King does not concern himself with the wealth changes implied by changing tax parameters which
are given special attention here. However, the neutrality results described here can be derived as a special
case of King's model, and it might be interesting to impose his richer structure of tax institutions on the
general equilibrium model examined here.6 For present purposes, however, the single tax on distributions is
sufficient.
1.3
Outline of the Chapter
Section 2 contains the formal description of the model economy, including the equilibrium concept
employed. Section 3 derives a solution of the model. Section 4 contains a discussion of the results
summarized above, while section 5 contains concluding remarks.
2
Formal Model of Rational Expectations Equilibrium with a Tax on Distributions
The model underlying the analysis is in the Samuelson (1958) consumption loan tradition.7 Individuals live
for two periods in an infi-
Â
Page 249
nite horizon world, with L born in the current period. (For any variable for which the current value is X, let
X+, X++, etc. represent its value in succeeding periods; X- its value in the preceding period. We shall also
require a notation for expectations. Let denote the value of X expected to obtain in the next period,
the value expected for next period's expectation, and so on.) In the first life-period each individual works
(offering one unit of labor inelastically), consumes, and saves for retirement. 'Retirement' describes the
second life-period, when each individual consumes his savings, leaving nothing to his heirs.
All production takes place in a single corporation, which may be thought of as a consolidation of the
corporate sector in an actual economy. (It would be a simple extension to allow noncorporate production.)
Production conditions are described by a constant-returns-to-scale production function, F(K, LD), of capital,
K, and labor employed, LD. The capital available to the corporation in any period is inherited from the
previous period, and is thus fixed in amount before the time of actual production. The output of a period may
either be consumed or frozen into an increment of the infinitely durable capital stock. To avoid having to deal
204
205
with corner solutions, investment is assumed to be reversible, i.e. the capital stock can be consumed.
The corporation is assumed to behave as a price taker in output, factor and financial markets. Here output is
taken as numeraire, measured in dollars. The wage rate is denoted by w. Because the firm is the only user of
physical capital, and owns the stock, there is no actual market for capital services. There are, however,
markets for the bonds and equity of the corporation. Bonds are issued on a one-period discount basis. The
total corporate indebtedness, B, inherited from the previous period, must be repaid during the current period.
The current rate of interest is denoted by r, so that the corporation raises B+/(1 + r) dollars by a new issue of
B+ bonds.
The "ex dividend" value of a 100 percent equity interest in the firm is denoted by V. This is the value at the
end of a period, when the equity interest is exchanged, after production is complete, and distribution and
investment decisions have been made. V will be a function, V(K+, B+), determined in the market, of the capital
stock and indebtedness that will be carried forward as a result of those decisions. It is this function that the
firm's owners take as given. It will simplify matters to assume that both sales and purchases of stock by the
corporation itself are prohibited, so that distributions
Â
Page 250
may be equated with ordinary dividends. Alternatively, purchases and sales of equity by the firm may be
allowed (and treated as negative and positive distributions) provided they are subject to the distribution tax (a
subsidy in the case of sales). Note that by definition, distributions, new bond issue proceeds, and real
investment must sum to a constant in any period.
The model also includes a government which ordinarily acts simply as a cash flow manager. It inherits from
the preceding period an obligation to redeem an outstanding stock, Bg, of one-period bonds. It finances this
redemption with the receipts, tD, from the tax at flat rate, t, on corporate distributions, plus the proceeds,
, of the sale of new bonds.
Individual savings may thus be held in three forms: bonds issued by the corporation, shares of its common
stock, or bonds issued by the government. A given generation of individuals acquires these financial assets at
the end of its first life-period, after production for that period has been completed, and after the financial and
investment decisions of the corporation have been fixed. The typical "young" individual is conceived of as
allocating his labor earnings among first-life-period consumption, c1, corporate bond holding, b+, government
1
bond holding, , and a fraction, s, of the ownership in the corporation, to maximize a utility function, u(c ,
2
c ).
The holders of its common stock "own" the corporation. The L- "old" individuals will thus be owners at the
beginning of a period. By choosing LD they control that period's production, and by choosing K+ they
determine the amount of capital which will be available for use at the beginning of the next period. They also
specify the financial policy for the current period, which means they set the amount of funds to be distributed
to themselves as dividends, D, and the amount of corporate borrowing, B+. The objective of the individual
stockholders is to maximize c2, which calls for the maximization of the sum of distributions net of tax and the
residual value of the equity claims, (1 - t)D + V(K+, B+).
Equilibrium Conditions
As we have seen, certain conditions of the model economy are inherited from the past, in particular the
capital stock, the corporate debt obligation, and the government debt obligation. The labour supply is
exogenously determined. I refer to the vector of predetermined and exogenous variables (K, B, Bg, L) as the
"state of the economy."
205
206
Â
Page 251
The endogenous quantities to be determined by the equilibrium conditions include the labor employed by the
corporation, and its capital stock and debt obligation carried forward into the next period. There are also two
prices of a conventional sort, namely the wage rate, w, and the interest rate, r, and a third price-like element,
the "valuation function," V(K+, B+). Finally, expectations about future prices and equity valuation are
determined endogenously. All individuals are assumed to have identical point expectations, represented by
the vector .
Determination of these
endogenous variables involves
clearing five markets, for labor,
corporate bonds, government
bonds, equity, and goods. In each
period the prices of labor and
bonds (government and
corporate bonds are assumed
perfect substitutes), and the
evaluation function relating
equity value to corporate
financial and investment
decisions, adjust to clear these
markets. Naturally, demands and
supplies depend upon
expectations. A temporary
equilibrium, conditional upon
both the state of the economy
and expectations, is a balancing
combination of prices and a
valuation function, together with
a specification of LD, K+, B+ and
, such that all markets clear.
The temporary equilibrium quantities in any period depend on the capital stock, outstanding debt obligations,
and the size of the new generation, in short, on the state of the economy. The evolution of the economy is the
result of the exogenous development of the population and the determination by market clearing of the values
of capital stock and bond obligations carried into the next period. Population growth is assumed describable
by a first-order difference equation. 8 Because of this assumption, it will be taken for granted below that
knowing the relationship between a current endogenous variable and L+ is equivalent to knowing that
between it and L.
As has been mentioned, the

expectations on which temporary
equilibrium depends are
endogenous to the model. It will
be assumed that expectations are
"rational" in the sense introduced
to the economics literature by
Muth (1961). In the present,
nonstochastic model, the notion
of rationality of expectations is
206
207
taken to encompass two

properties. The first is that
rational expectations are correct
in that in an economy evolving
according to a sequence of
temporary equilibria, each
dependent upon that period's
expectations, each period's
expectations will be fulfilled in
the following period. The second
property embodies the idea that
if two economies are identical in
Â
Page 252
structure and have reached the same state (where the calendar date is not considered part of the state
description), their agents should have the same expectations. Expectations satisfying the second property can
be described as stationary functions of state variables. 9 Note that there is no guarantee that rational
expectations exist for an economic system, nor that if they exist they are unique.
3
A Solution
to the
Model
To describe the behavior of the

model economy in an
equilibrium with rational
expectations we must write
down expressions for the
endogenous variables, including
expectations, in terms of the
state variables, with the property
that, given expectations, markets
clear and expectations are
correct.10
The natural way to proceed would be to solve for the temporary equilibrium as a function of the agents'
expectations and then to seek a form of expectations that is self-fulfilling. Given the large set of possible
expectations (recall that the valuation of equity is represented by a function) this is rather complicated.
Instead, I shall take the approach of first developing a reasonable conjecture about the temporary equilibrium
under rational expectations, then describing the associated scheme of expectations, and finally showing that
the optimizing behavior of agents holding those expectations will produce the originally conjectured
temporary equilibrium.
I therefore follow a somewhat artificial sequence in which the first step is to write down equations which are
asserted to determine the endogenous variables other than expectations. I then write down the equations
asserted to determine the expectations from the state variables, and show that when the system evolves as
described in the previous section these expectations are correct. Only then do I go through the steps to show
that if the agents have these expectations, their optimizing behavior will indeed lead to a temporary
equilibrium with the originally asserted relationship between state variables and current prices and quantities.
207
208
3.1
Endogenous Variables in Temporary Equilibrium
The strict logic of the argument

does not call for any discussion
of equilibrium until step three.
However, the asserted
relationships between current
prices and quantities and state
variables in tempo-
Â
Page 253
rary equilibrium with rational expectations will be recognizably conditions of competitive market clearing,
and will be more readily understood if discussed in these terms.
Eqs. (1) and (2) below, with their associated definitions, eqs. (3) and (4), describe wage and interest rate
determination by labor and goods market clearing. The function K* defined in (3) and evaluated at (r, L+)
gives the capital stock which equates the rate of interest to the marginal product of capital at full employment
in the next period. This is the amount of capital which will be carried forward into the next temporary
equilibrium. Problem (4) is a lifetime utility maximization problem of a representative individual with
lifetime consumption preferences described by u, when the rate of return on savings is r and with wealth
equal to w, the wage reward for one unit of labor. Eq. (1) is the condition of equality of wage and marginal
product of labor at full employment. Expression (2) specifies that total savings of the young generation
equals the capital stock carried forward plus the government debt less a term equal to the tax receipts that
would result from liquidation of the firm. The latter term can be loosely interpreted as the present value of all
future distribution tax proceeds, and thus has the effect in the model world of offsetting government debt.
Here Fi denotes the derivative of F with respect to its ith argument; K* (r, L+) is defined by
and c1 (r, w) solves the problem
subject to
The next relationship is based on the market valuation of the equity interest in the firm. Since there is no
uncertainty in production, that value should simply be the discounted sum of next period's
Â
208
209
Page 254
distribution, net of distribution tax, and next period's valuation of the firm after its real investment has been
determined. However, this describes a relationship among endogenous variables over time, whereas our
objective is to describe endogenous variables as functions of current state variables. It is natural to conjecture
that the equilibrium valuation of equity will be related to the potential net-of-tax withdrawals from the firm.
Since all withdrawals are taxed, the expected net withdrawal obtained if no debt is issued in the next period
(in effect ''closing the account") would be the obvious basis for evaluation. This is given by
Since there is no uncertainty, this sum will be discounted at the rate of interest, r, giving
Recall that our objective is to specify the endogenous variables as functions of state variables. Noting that r
has been described as a function of state variables by (1)-(4), we must only anticipate the rational
expectations of wages to write V in the required form. The expected wage level will presumably clear the
labor market when the endogenously determined capital stock is available. Hence, we conjecture, by analogy
with (1):
so that the equity valuation function is itself expressed as a function of state variables by
which says that the value of equity is equal to the discounted value of the after-tax proceeds of distributing in
the next period all of the anticipated resources of the corporation, after paying off the debt. As we shall see
below, this temporary equilibrium relationship among endogenous and state variables implies the desired
intertemporal relationship between one period's equity value and next period's distribution and equity value.
These equations determine the two prices and the equity valuation function. The capital stock carried into the
next period will be given
Â
Page 255
by K* (r, L ). It remains to determine B and . The stock of debt the young will want to hold, with present
+ +
value , will depend on their evaluation of equity. The aggregate portfolio value of the young
is determined by their consumption decision. Therefore, we have
It turns out that any combination of B and satisfying (8) is compatible with equilibrium. Associated with a
+
larger value of B will be lower values of V and . Note that condition (8), together with (7), implies
+
209
210
where the constant depends upon r, w, and L+.
3.2
Rational Expectations
Expression (6) relates the expected wage to the labor force in the next period and the capital stock that would
be optimally employed with the labor if capital services were directly marketed and had a price equal to r, the
rate of interest. The value of r is determined from state variables by (1)-(4). The remaining expectations of
interest rate and equity valuation function are, like (6), obtained by adding the expectation superscript to the
appropriate temporary equilibrium condition and making use of the functional relationship between state
variable and current prices and quantities, as already described.
The value of is obtained by combining (1)-(4), (6), (9) and (10):
Similarly, the expected valuation function is given by
where
Â
Page 256
The Correctness of Expectations
Given our approach to developing the expectations formulae, it will be no surprise that they will be correct if
the next period's equilibrium is described by relationships (1)-(4), (7) and (8), suitably updated. For example,
since K = K*(r, L ), condition (6) implies that . Similar reasoning implies that and .
+ +
3.3
The Temporary Equilibrium is the Right One
It remains to show that if the agents form their expectations in the way described, then eqs. (1)-(4), (7) and
(8) indeed characterize a temporary equilibrium. For this we need to derive the demands and supplies of the
agents on the various markets and test for market clearing.
The budget constraints of the young and old generations together with the behavioral rule ascribed to the
government imply that the excess demands on these five markets are dependent in the usual Walrasian
fashion: clearing of four markets implies clearing of the fifth. This may be most easily seen if we for a
moment regard the old generation as two sets of agents: stockholders and corporation managers. The budget
balance requirement for the young implies that the sum of the values of their excess demands for the five
"commodities" (labor, corporate bonds, government bonds, equity and goods) must be zero. The sum of the
values of excess demands of stockholders will be B + Bg + (1 - t)D as the old generation cashes in its old
bonds (which are not commodities marketed in the current period) and collects the after-tax proceeds from
210
211
the corporate distribution. The sum of values of the excess demands of the corporation will be -B - D, as it
pays off the inherited debt obligation and makes its distribution. For the government, which supplies new
debt to cover the difference between old debt and tax receipts, the sum of excess demands is -Bg + tD. It may
be verified that the four classes of excess demands aggregate to zero. Hence, if market excess demands of
four of the five commodities are zero the fifth market excess demand must be zero as well.
3.3.1
Excess Demands of the Young
Consider first the young generation. They anticipate solving the problem of optimally managing the
corporation as owners of its equity, with capital stock and debt repayment obligation given by (K+, B+). That
is, they anticipate solving
Â
Page 257
where D+ is the distribution they make to themselves from the corporation:
Substitute (11), (12) and (14) into the maximand, thereby eliminating the decision variables B++ and D+, and
calculate first-order conditions for a solution with respect to the remaining two. The anticipated labor demand
satisfies
Using (6) and the assertion (3) that K = K*(r, L ) (used in implicitly determining according to rational
+ +
expectations), we conclude that
The first-order condition for optimality with respect to K++ in (13) is
Euler's theorem applied to the constant-returns-to-scale production function implies that, given the asserted
relationships between state variables and current endogenous variables, the solution of (13) can be expressed
as
In (18) the determination of L and as a function of current state variables is implicitly taken for granted.
+
Now imagine that a Walrasian auctioneer has called off wages, interest rate and valuation function and the
existing owners of the corporation have set K+ and B+. The representative young person now solves the
lifetime consumption maximization problem:
211
212
subject to
Â
Page 258
and
The two constraints in (19) can be combined to imply
Problem (19) is unbounded, and the resulting demands incompatible with equilibrium unless the term in
parentheses is zero:
We may verify from (5) that this condition is satisfied if the wage, interest rate and valuation function are in
the suggested temporary equilibrium relationship.
When condition (21) is satisfied, the young are indifferent about their portfolio composition among the two
types of bonds and corporate equity. Solving problem (19) implies that the value of the aggregate demand for
financial instruments by the young is (w - c1(r, w))L. The aggregate value of their goods demand is c1(r, w)L.
Their labor is supplied inelastically.
3.3.2
Excess Demands of the Old
As owners of the corporation, the old generation must choose labor demand, real investment, and new bond
obligations to maximize the sum of after-tax distributions and proceeds from the resale of the equity interest.
In other words, their problem is:
where
Substitute (7) and (23) into (22) and derive first-order conditions:
212
213
from which
Â
Page 259
Euler's theorem applied again gives us the value of the objective function at an optimum of W(K, B, L, w), as
defined by (18). This is the maximum the owners of the firm can realize, given w and the inherited capital
stock and debt obligation (and given the equity valuation function (7)).
In the solution to the maximization problem the financial policy is indeterminate. Given the other variables, a
unit increase in B+, resulting in a net distribution larger by 1 - t, leads to a reduction of 1 - t in the market
value of equity. The supply of corporate bonds may thus be arbitrarily set by the owners of the corporation.
On the remaining markets they (the old) supply inelastically 100 percent of the ownership interest. They
demand consumption goods amounting to the sum of W, B and Bg. In addition they demand K+ in the goods
market for the corporation to carry into the next period.
3.3.3
Excess Demands of Government
Because I have chosen to consider the markets for labor, corporate bonds, equity, and goods, we do not need
to give more consideration to the demands and supplies of the agent "government" than was involved in
establishing the appropriate version of Walras' law for this model. Ordinarily, the government does not
appear on either side of any market except that for government bonds, the commodity dropped from the
dependent system of market-clearing relationships.
3.3.4
Market Clearing
Having established the excess demands of each of the agents, we can proceed to the analysis of market
clearing. The indifference of the young generation about portfolio composition and that of the corporation's
owners about financial policy assures clearing of the equity and bond markets. The young purchase all of the
equity in the corporation (at the price described by (7)) and all debt offered for sale by the corporation.
This leaves the labor and goods markets. Clearing of the labor market requires Ld = L or, from (24),
Equating demand and supply in the goods market implies:
Â
213
214
Page 260
Using (18), we can rewrite (28) as:
Eqs. (27) and (29) are the same as eqs. (1) and (2), from which we conclude that the latter do, indeed,
describe the determination of wage and interest rate in temporary equilibrium, given expectations formed as
described in the previous subsection. This completes the argument.
3.3.5
Uniqueness of Equilibrium
As noted above, the definition of equilibrium with rational expectations does not imply its uniqueness. We
shall not attempt here to prove the uniqueness of the equilibrium of the model specified above. However, we
can show that what might loosely be described as the "conventional view" of valuation of the firm, in which
each extra dollar of retained earnings is reflected in a dollar increase in equity value, will not be consistent
with rational expectations in this model. For if this valuation rule prevailed, agents would be led to demands
-1
incompatible with equilibrium. To express the conventional view we replace V and above by (1 - t) V and
, cancelling the effect of the tax on equilibrium valuation as seen in (5) and (11). Now the
anticipated valuation problem (13), after the substitution of D into the maximand, continues to have B++ as an
argument, with constant coefficient . The solution value B = - ℜ∞ implies an infinite valuation
of the firm by equity demanders, and an infinite supply of bonds by the++ young (individual borrowing to
purchase stock). By the same reasoning with respect to problem (22), the owners of the corporation are led to
an infinite supply of bonds, ruling out market clearing. The same sort of conclusion follows if V and are
multiplied by any other constant different from unity.
4
Commentary
4.1
Characteristics of Equilibrium Paths
The analysis of section 3 demonstrates the assertion made in the introduction that the indifference between
debt and equity character-
Â
Page 261
izing financial policy in the absence of taxes in this no-uncertainty world continues to hold in the presence of
the tax on distributions. The key to this conclusion is found in the equity valuation function (5), from which it
follows that an extra dollar of retained earnings induces an increase of only 1 - t dollars in the value of equity.
The conclusion that the investment criterion of the firm (K+ = K*(r, L+)) is unaffected by the tax on
distributions is in itself not surprising, since without uncertainty all financing, and in particular marginal
financing, can take the form of debt. Note, however, that this result holds even if the firm's owners do not
consider financing the marginal investment through bonds. It is only required that the firm's potential
shareholders compare the return on a dollar retained by the corporation and invested in capital with that on a
dollar invested in bonds.
214
215
The implication that financial policy is indeterminate follows immediately from (5). It can also be shown,
using (1) and (2), that the path of the economy through time is independent of the choice between debt and
equity finance. From (2) it follows that the determinants of temporary equilibrium values of w and r include
B, the inherited corporate bond obligations, when t is positive. However, equilibrium depends also on Bg, the
inherited government bond obligation. In the evolution of the economy these two types of debt are
interrelated; a larger issue of corporate bonds in a period implies a larger corporate distribution, larger tax
receipts, and smaller issue of government debt. Using the government behavioral equation,
g
, and the definition (23) of D, we conclude that the time path of B + tB is independent
of corporate financial policy (as represented by B+), since
and it is this sum which enters the goods market clearing condition (2).
4.2
The Effects of Changes in Parameters
The equations of temporary equilibrium with rational expectations allow us to analyze the effect of changes
in the tax parameter or in government behavior, provided these are regarded as one-time changes which are
not expected to recur. Thus, an unexpected increase in the tax rate will, according to (5), result in a loss of
wealth (and hence consumption) for the old generation. If there is no change
Â
Page 262
in the rate of interest, a five percentage point increase in the rate results in a loss equal to 5 percent of what
their equity claims would have been worth in the absence of the tax.
In general, we know from (2) that a change in t, given the values of the state variables, will lead to a change
in the equilibrium values of w and r, and consequently, a change in the future course of the economy. In this
sense the tax on distributions is not neutral. However, the changes in the dynamic path derive not from
substitution away from equity finance, but rather from the wealth effects of the tax. An increase in the tax
rate reduces the perceived wealth of the economy by reducing the value of equity. Eq. (2) suggests the wealth
effect may be equivalently viewed as a reduction in the extent to which government debt is regarded as
wealth. An increase in the tax rate results in an increase in the implicitly expected debt retirement (by tax
receipts) in the future.
Paradoxically, the implication that an increase in the rate of tax on distributions has the opposite effect on the
path of the economy from an increase in government debt means that the latter may be the means not for
shifting a burden to the future but for confining the consequences of an unexpected increase in exhaustive
government expenditure to the old generation at the time it occurs. Consider an economy moving along a
rational expectations equilibrium path with tax rate t, arriving at the time of the hypothetical moment of
unexpected government expenditure with outstanding corporate and government debt obligations B and Bg.
The government considers an increase in t to meet the expenditure, which we shall assume has no effects
which interact at any time with ordinary consumption in individual preferences. From (28) we know that an
increase in t of dt, ceteris paribus, creates a goods demand shortfall of dt(F(K, L) - wL - B + K) because it
reduces by this much the value of equity with which the old generation had planned to finance consumption.
This is the extra expenditure which can be undertaken by the government, diverting resources from old
generation consumption, while maintaining goods market equilibrium. If this is financed by issuing the
requisite amount of extra debt, financial market equilibrium will also be maintained. Suppose that in
subsequent periods the government returns to its normal debt-managing role. Then the increased government
215
216
debt will exactly offset the effects of the increased tax rate. In all real respects the path will be unchanged
from the pre-extraordinary events path.
Â
Page 263
To see this note that the original path will continue to be an equilibrium one if in every period (from (2)),
Suppose this relationship holds in some period. Will it hold subsequently under the original rules of the
economy's evolution? We have from the government behavioral equation:
We wish to show that this implies
Substituting from (31) into (32) we have
from which, using the accounting relationship (23),
Now check that
This will hold if
which we know to be true by Euler's theorem along the original path.
Finally, we need to show that the new government debt, created to cover the extraordinary expenditure,
amounts to dBg as defined in (31). Instead of the combination of changes dt, dBg, we make the combination
dt, , where, because of the tax, the latter depends upon the financial policy of the firm such that
Â
216
217
Page 264
From (32) we see that this produces exactly the same situation in the next period as did the originally
described pair of changes. The subsequent path of the economy thus continues to be an equilibrium.
5
Concluding Remarks
The principal purpose of this paper has been to demonstrate that the double taxation of dividends
characteristic of the classical corporation income tax need distort neither corporation financial nor investment
policies. While the model embodies drastic simplifications, these have been designed to isolate the main
point. The usual argument by which it is concluded that double taxation leads to distortion is developed in
essentially the same setting. What is required is an extra tax on distributions; the presence or absence of other
taxes has no bearing on the conclusion. This analysis, therefore, is sufficient to show where the previous
reasoning goes wrong.
When the connection between the tax on distributions and the valuation of equity is taken into account, it is
seen that conclusions about the distorting effects of the corporation income tax based either on simple
compounding of statutory corporation and individual income tax rates or on measured average tax rates using
tax receipts are founded on incorrect premises. Both the continued existence of equity finance and the
practice of making positive dividend payments may be fully consistent with maximization of stockholder
wealth, even though dividends are subject to a tax and even though (in the model) capital gains are not taxed.
The picture of the incidence of the double taxation of dividends that results from spelling out its effects on
asset prices is also somewhat surprising. While the precise incidence in this model is a function of
government debt as well as tax policy, a major, if not the only, burden of a tax on distributions is borne by
equity holders at the time the tax is instituted. A significant aspect of such wealth transfers is that they cannot
be meaningfully reversed at a later date when the equity interests have changed hands. Removing the tax on
dividends cannot restore the wealth of the original losers, but only provide a windfall gain to the new
shareholders. These effects deserve careful consideration in attempting to understand such reforms as full or
partial integration of corporation and individual income taxes.
Â
Page 265
Notes
An early version of this paper was written while I was a visiting fellow at the Center for Operations Research
and Econometrics (CORE), UniversitÃ© Catholique de Louvain, Belgium (January-June 1977), and
appeared as CORE Discussion Paper 7738, August 1977. I would like to express my appreciation for the
exceptionally stimulating research environment at CORE as well as my thanks to UCL and the Commission
for Educational Exchange Between the United States of America, Belgium, and Luxembourg for financial
support. Discussions with William Andrews of Harvard Law School, Maurice Marchand and Henry Tulkens
of CORE and Pierre Pestiau of the UniversitÃ© de Liege, were of great help. Finally, thanks to Martin
Hellwig, then of CORE and Princeton University, who went beyond the call of duty in reading drafts and
educating me on the fine points of the theory of rational expectations.
1. For a discussion of some of the problems and a suggested method to solve them see U.S. Government
(1977).
217
218
2. This is the label attached by Buchanan (1976) to the notion that private demands are independent of the
extent of deficits as a method of financing public expenditure (presumably because taxpayers anticipate the
correspondingly higher future taxes). Both the label and the idea have provoked controversy. For recent
reviews and some evidence see Buiter and Tobin (1979) and Feldstein (1980).
3. Feldstein (1980) seeks to incorporate the relationship between expectations about future taxes and current
deficit-influencing events in his empirical study.
4. See also the follow-on papers by Stiglitz (1976), King (1975) and Asimakopulos and Burbidge (1975), as
well as King (1977).
5. There is a further difficulty with Stiglitz's representation of the provisions in U.S. tax law distinguishing
return of capital from other distributions. Whereas Stiglitz (1973, p. 9) makes this distinction a matter of the
cumulative amount the owner has received from the corporation, it in fact depends upon the relationship
between cumulative earnings and cumulative distributions. Under the Stiglitz version, in effect, the first
distributions represent return of capital while the U.S. law may be described as making the last distributions
return of capital.
6. Using King's notation, the special case is that of mt = 0, zt = 0, tt' = 0, θ = 1 - τ, where τ is the rate of tax on
distributions.
7. Diamond (1965) presents a model which includes government debt. Unlike the Diamond and Samuelson
models, this model does not assume constant population growth, nor is the analysis confined to steady states.
8. If population evolution were describable only by a higher order difference equation, additional state
variables, in the form of a sufficient number of observations of past population levels to determine the
solution of the difference equation, would be required. The basic argument would, however, be unaffected.
9. To express this notion precisely, let Te(a) denote the set of possible successors to a as a result of temporary
equilibrium, given expectations vector e. Formally, expectations will be described as rational if there exists a
mapping, E, from the set of state vectors into the set of expectations vectors, such that for all possible state
vectors, a, and all possible successor states, a+ in TE(a) (a), the expectations are correct, i.e. E(a) = (w+, r+, V+).
Â
Page 266
10. Throughout the following discussion of equilibrium it is assumed that the tax parameter, t, is constant.
One-time policy changes are assumed to be unexpected. Furthermore, non-negativity constraints are ignored.
King's (1974) analysis incorporates expectations concerning tax parameters and makes considerable use of
various non-negativity constraints.
References
Asimakopulos, A. and J. B.
Burbidge, 1975, Corporate
taxation and the optimal
investment decisions of firms,
Journal of Public Economics 4,
281-287.
Barro, Robert J., 1974, Are government bonds net wealth?, Journal of Political Economy 82, 1095-1117.
Buchanan, James M., 1976, Barro on the Ricardian equivalence theorem, Journal of Political Economy 84,
337-342.
218
219
Buiter, Willem H. and James Tobin, 1979, Debt neutrality: A brief review of doctrine and evidence, in:
George M. von Furstenberg, ed., Social security versus private saving (Ballinger, Cambridge, MA) 39-58.
Diamond, Peter A., 1965, National debt in a neoclassical growth model, American Economic Review 55,
1126-1150.
Feldstein, Martin, 1980, Government deficits and aggregate demand, Working Paper no. 435 (National
Bureau of Economic Research, Cambridge, MA).
King, Mervyn A., 1974, Taxation and the cost of capital, Review of Economic Studies 41, 21-35.
King, Mervyn A., 1975,

Taxation, corporate financial
policy, and the cost of capital:
A comment, Journal of Public
Economics 4, 271-279.
King, Mervyn A., 1977, Public policy and the corporation (Chapman and Hall, London).
Modigliani, Franco and Merton H. Miller, 1958, The cost of capital, corporation finance, and the theory of
investment, American Economic Review 48, 261-297.
Muth, J., 1961, Rational expectations and the theory of price movements, Econometrica 29, 315-335.
Samuelson, Paul A., 1958, An exact consumption loan model of interest with or without the social
contrivance of money, Journal of Political Economy 66, 467-482.
Stiglitz, Joseph E., 1973, Taxation, corporate financial policy and the cost of capital, Journal of Public
Economics 2, 1-34.
Stiglitz, Joseph E., 1976, The corporation income tax, Journal of Public Economics 5, 303-311.
United States Government, 1977, Department of the Treasury, Blueprints for Basic Tax Reform (Washington,
DC).
Â
Page 267
9
A Problem of Financial Market Equilibrium When the Timing of Tax Payments is
Indeterminate
1
Introduction
This chapter concerns an aspect of the question: When do government deficits matter? It takes as a starting
point previous work (1981) showing how endogenously generated deficits might have no real effect, a result
obtained under an assumption of perfect substitutability between government and private debt. The extension
to a world of risky debt where the latter no longer holds turns out to involve new elements that may be of
some general interest. In particular, the conditions for neutrality seem less likely to be fulfilled in a practical
context.
219
220
The underlying idea is that it should not matter when taxes are paid provided there is an appropriate
compensating interest element in the postponed liability. This notion conflicts with the assumption often
employed that it is the government's cash flow balance that counts, even though current deficits may be offset
by correspondingly larger liability for future tax payments, and surpluses may reflect drawing down liabilities
for future taxes. This issue arises especially strongly in the context of analysis of proposals for
consumption-type taxes, where there is a choice between a literal consumption tax and a tax on wage and
transfer receipts. Typically, the two approaches generate the same liabilities in a present-value sense but very
different cash flows. In some systems [for example, the Cash Flow Tax analyzed in Bradford et al. (1984)],
the taxpayer has wide latitude to choose between the approaches.
The specific case analyzed here presents the same issues in a particularly clear way. It involves a tax on
distributions by corporations to equity holders, in essence, a dividend tax. If such a tax is assessed at a flat
rate that is not expected to change, and if negative
Â
Page 268
distributions (sales of new equity) are included (i.e., subsidized), the case is quite compelling in a partial
equilibrium setting that the level of the tax should have no influence on real or financial transactions of a
corporation acting in the interest of its stockholders. The reason is simply that the flat tax changes
proportionately the consequences of all decisions as far as stockholder outcomes are concerned. In particular,
the trade-off in after-tax dollars for the shareholders between a larger distribution today and the consequently
smaller distribution at some future time is unaffected by the rate of tax.
In the absence of all taxes (and transactions costs), the various versions of the Modigliani-Miller (1958)
theorem tell us the corporation will be indifferent between debt and equity finance. An implication is that the
timing of dividend payments is a matter of indifference. Since, as I have just argued informally, a flat tax on
dividends has no effect on the optimal financial policy, something like the Modigliani-Miller theorem should
continue to hold. However, the choice of pay-out affects the government's cash flow. Government receipts
will be determined by the whims of corporate managers; private wealth-maximizing calculations are
insufficient to fix the path of revenues. Will the consequent fluctuations in tax receipts have real effects?
In Bradford (1981), I spelled out an overlapping-generations model in which government debt does generally
influence the rational expectations equilibrium path, but in which variations in government debt attributable
to variations in distribution tax receipts do not matter. The indifference about financial policy at the level of
the firm, in spite of the tax on distributions, carries over to neutrality of the economy's path to the choice of
financial policy, even though the flow of tax receipts is affected.
The key question is whether there are general equilibrium effects on the rate of interest. The basis for the
neutrality conclusion may be sketched as follows: The capital investment level chosen by the firm is
governed by the going interest rate. Therefore, a decision to issue an extra dollar of debt implies a decision to
distribute an extra dollar to shareholders. This in turn implies extra tax receipts of t dollars (where t is the rate
of tax). If real government spending is fixed, the extra t dollars are devoted to reducing the public debt. The
result thus far is a net addition of 1 - t dollars to the supply of financial assets. There remains, however, the
equity interest in the corporation, the value of which is predicted to fall, not by one dollar,
Â
220
221
Page 269
as a result of the extra debt cum distribution, but by 1 - t dollars, in reflection of the government's claim to a
fraction t of all cash flows to shareholders. As a consequence, all markets continue to clear. The firm's
decision has no real effect, even though tax receipts are increased.
The foregoing analysis supports two conclusions. First, a tax on corporate distributions may not have the
often-assumed incentive effects with respect to real and financial allocations. And second, variations in
government receipts attributable to changes in corporate distribution policy may have no effect on the real
path of the economy. The present investigation concerns how the neutrality results are affected if the perfect
substitutability among financial assets used in the argument above and attributable to the assumption of
certainty is replaced in the context of an explicit treatment of uncertainty.
As it turns out, the earlier results carry over without significant complication when only equity is risky at the
margin, whereas corporate debt and marginal public debt are risk free and therefore perfect substitutes. The
restriction on government financial choices is a special case of the requirement that applies when the risk
characteristics of corporate debt are unrestricted. In the more general case, the neutrality conclusions require
that the government policy be describable as one of issuing a certain prespecified risky debt together with the
purchase of a fraction of the private debt supply equal to the tax rate.
These conditions on government behavior are not as arbitrary as might appear. As stressed elsewhere
(Bradford 1981), the proportional tax on corporate distributions in effect gives the government a fractional
ownership in the firm. Neutrality will follow if the government uses the tax revenue incident upon an
incremental distribution to purchase bonds of the firm. This is just what is needed to preserve the
predistribution portfolio of the government, taking into account its implicit ownership of corporate equity. In
other words, the government behavior that implies neutrality is just what we wouldin Modigliani-Miller
fashionexpect of a shareholder that receives a dividend.
In Section 2, I review the certainty results as a way of introducing the basic model. The extension to a model
with uncertainty is presented in Section 3. Section 4 contains concluding remarks.
Â
Page 270
2
Review of the Certainty Model
The model underlying the analysis is in the Samuelson (1959) consumption loan tradition. Individuals live for
two periods in an infinite time horizon world. In the first life period, each individual works (offering one unit
of labor inelastically), consumes, and saves for retirement. ''Retirement" describes the second life period,
when each individual dissaves and consumes, leaving nothing to his or her heirs.
All production takes place in the consolidated corporate sector, which is modeled as a single price-taking
firm. Production conditions are described by a neoclassical production function of capital and labor
employed, with constant returns to scale. The capital available to the corporation in any period is inherited
from the previous period and is thus fixed in amount before the time of actual production. The output of a
period may either be consumed or frozen into infinitely durable capital. Investment is regarded as reversible.
Savings may be held in three forms, bonds issued by the corporation, shares of its common stock, or bonds
issued by the government. A given generation of individuals acquires these financial assets at the end of its
first life period, after production for that period has been completed, and after the investment and financial
plans of the corporation have been realized.
221
222
The holders of its common stock "own" the corporation. The owners at the beginning of a period control that
period's production and the real investment that determines the amount of capital that will be available for use
in production at the beginning of the next period. The owners of the firm at the beginning of a period specify
as well the financial policy for that period, which means they set the amount of funds to be distributed to
themselves as dividends and the amount of corporate borrowing. At the end of the period, the current owners
sell the equity to the young generation of savers.
The government's real spending program is assumed fixed, for simplicity at a zero level. The government is
thus modeled as a mechanical cash flow manager: In each period, the inherited debt obligation must be paid
off, with any shortfall of tax receipts covered by the issue of new debt. Since the distribution tax is the only
tax, there is nothing else for the government to do. Note that by allowing the government to make lump-sum
transfers, the model could be
Â
Page 271
used to permit the government to engage in (preannounced) inter-temporal redistributions. This would
involve issue of government debt in amounts larger or smaller than the difference between old debt
obligations and distribution tax receipts.
Notation
To describe the results of the analysis formally, I use the following notation (involving minor changes from
that of the 1981 article to facilitate extension to incorporate uncertainty):
L: Total number of labor units available for application during the period (equals the
number of individuals born in the period, exogenously given)
K: Stock of corporate capital at the beginning of the period (used for production
during the period)
B: Total stock of corporate indebtedness at the beginning of the period, which must be
repaid during the period
F(K, The production function, characterized by constant returns to scale

L):
D: Total distribution made by the corporation during the period
Bg: Total stock of government indebtedness at the beginning of the period, which must
be repaid during the period
t: Rate of tax on corporate distributions (which may be negative) to stockholders
There are two ordinary prices in the model (current output is numÃ©raire) and one pricelike "valuation
function":
222
223
w: Wage
d: The discount factor, the current-period price of a bond paying one dollar
next period
V(K+, B+): A function relating the "ex-dividend" value of equity, that is, the value at
the end of a period, after production is complete, to the financial and
investment decisions of the current owners
For any variable X, let X+, X++,

and so on, represent its value
in succeeding periods; X-
represents its value in the
preceding period. In order to
determine how to value the
firm's equity, individuals
Â
Page 272
must form expectations about the prices and valuation function one period hence. Let denote the value of
w expected to obtain in the next period, and similarly for d and V. It is assumed that everyone agrees about
, , and ; forecasts more than one period into the future are not needed.
There are three classes of agents in the model, two displaying maximizing behavior. The "young" try to
i
maximize (labor is supplied inelastically), where c is consumption during the ith period of the life
cycle. The "old" manage the corporation to maximize c2, which means maximize the sum of after-tax
distributions and the proceeds from sale of the equity, (1 - t)D - V(K+, B+). The third agent is the government,
which simply manages the cash flow according to . (Reminder: d is a price, not the differential
operator.)
Evolution of the Economy
The situation inherited from the past is described by (K, B, Bg, L), with L evolving exogenously along a
known path. The requirements placed on the model world are that the evolution to (K , B , , L ) be
determined by clearing of competitive "spot" markets for labor, corporate bonds, government
+ + bonds,
+
corporate equity, and goods and that the price expectations on which the value of the firm depends be
"rational." The general notion of rational expectations, attributed to Muth (1961), here encompasses two
properties: Expectations are correct, and they are determinate, in the sense that they are governed by
knowledge of the economic structure and the current state of the economy. For rational expectations to make
sense, there must be an appropriate degree of determinacy of the model as a whole, including its expectations
formation. Solving the model involves showing that the endogenous variables, including prices and
expectations, can be expressed as stationary functions of the state variables K, B, and Bg given the known
path of L. Properly, there should also be a demonstration that the proposed equilibrium path of prices is at
least locally uniqueotherwise, why should the economic agents pick the required expectations-forming rule?
My previous paper (1981) described an equilibrium path of the economy sketched out above. (I did not
succeed in demonstrating local uniqueness.) The neutrality result concerning the rate of tax on corporate
distributions followed from the conclusion that the valua-
223
224
Â
Page 273
tion function on the

equilibrium path is
given by
The objective of the owners of the firm is to set employment together with K+, B+, and D to maximize (1 - t)D
+ V(K+, B+). If we substitute for D in the objective function, using accounting relationship (2) among the
outlays by the firm, B+ drops out:
Maximizing values of the other variables are independent of t; the financial structure of the firm is
indeterminate.
Indeterminacy of financial structure corresponds to indeterminacy of government tax receipts. The reason the
equilibrium path of the economy's real variables and prices is nonetheless determinate is suggested by
equilibrium condition (3), which describes equality between the value of demanded and supplied claims to
future consumption:
Here is the
retirement-period consumption
(which will take place next period)
demanded by a representative young
person, the values of K+ and B+ are
set by the current owners of the
firm, and I have taken for granted
. The expression on the left
is the value of claims demanded on
the basis of life cycle optimization
by members of the young
generation. The first two terms on
the right are the values of bonds
supplied by the firm and the
government, and the third term is
the value of corporate equity.
Exploiting the government's budget
constraint and the
already mentioned accounting
relationship (2) between D and the
other variables allows us to write
condition (3) as
224
225
Once again, B+ has been eliminated, so the condition is independent of corporate financial policy.
Equation (5) expresses a further reformulation of the same condition, taking advantage of Euler's theorem:
Â
Page 274
The left side of (5) is, as before, the demand by the young generation for assets. The right side, the economy's
net supply of assets, is the sum of the capital stock and the difference between government bonds carried over
from the past and the tax receipts that would be generated if the corporation were to be liquidated in the
current period. The last term affects the real equilibrium path like an anticipated tax receipt "asset" of the
government, offsetting explicit government debt. Through this relationship, one obtains a clear sense of why
deficits or surpluses due to variations in corporate distributions have no effect on the real path of the
economy, even though government debt does matter. An increase in distributions simultaneously reduces
government debt and the anticipated tax receipt asset by equal amounts.
3
Introducing Uncertainty
There are various ways one might introduce uncertainty to this model. Taking advantage of the framework
pioneered by Kenneth Arrow (see Arrow and Hahn 1971), suppose that future production conditions depend
on the particular state (e.g., weather conditions) occurring. Specifically, assume that the investment decision
is fixed in the current period, but the actual production function is determined in the next period.
Let S be the set of possible states that might obtain in the current period, S+ the set of possible states in the
next period, and so on. The larger dimensionality of the problem requires some new notation. The following
describes my compromise between comprehensiveness and mnemonics. In general, variables are now
understood to have a subscript to designate the state and time with which they are associated. Thus, wi, i ℘
S+, refers to the wage rate realized in state i in the next period. Since the story starts with a known current
state, we can let variables with no subscript refer to the values currently realized.
It will be assumed that S is finite in each period and that an ordering has been agreed upon for the states in
each period, so we can use the subscript + to refer to the vector of values of a variable potentially obtaining in
the next period. (An exception is made for K+ and L+, which keep their previous scalar interpretation.) Thus,
refers to the vector of wage rates presently anticipated for next
Â
225
226
Page 275
period, with components wi, i ℘ S+, and has dimensionality equal to the number of states in S+.
The firm and the government must now specify, instead of the single-dimensional bonds, vectors of
state-specific claims. To simplify matters, assume that markets exist for each of the possible one-period
contingent claims. A unit of type i, i ℘ S+, pays one dollar next period if state i is realized, and zero if
another state is realized. To issue a riskless bond is equivalent to selling one unit each of claims of all types i,
i ℘ S+. The firm's bond financing is described by the vector B+, the ith component of which Bi (i ℘ S+) is
the amount the firm will owe its bondholders if state i is realized next period.
Instead of the single discount factor d, we now have a vector of prices of unit claims contingent on the next
period's state. Like the discount factor d in the certainty case, the vector d is understood as representing prices
actually ruling in financial markets. (The prices of unit contingent claims might be observable only by
appropriately packaging available financial instruments.) If e+ is a vector of 1's of appropriate dimension, the
inner product e+ Â· d is the price of a riskless bond. Since we shall want to continue to use the ordinary
subscript to refer to the state and time in which a particular variable is realized (for example, di, i ℘ S+, is
the vector of discount rates, applicable to claims on output two periods hence, realized if state i occurs next
period), I shall use parentheses when I wish to identify a particular element of d. Thus, (d)i, i ℘ S+, is a
scalar, namely, the present price of a claim to one unit if state i occurs next period. The symbol d+ refers to
the matrix of discount factor vectors, one for each possible state that might be realized next period.
We are now in a position to study the analog in the world of risk to the temporary equilibrium relationships
discussed earlier. Of particular importance is the valuation function for corporate equity corresponding to (1)
given by
This is simply the value of the vector of payoffs, contingent on the state realized, that the purchaser of the
equity expects to obtain next period in the form of distribution plus proceeds from the sale of the equity
interest.
To analyze asset market clearing in this case, it is not sufficient to look at the aggregate value of present
claims, as we did in (3) above.
Â
Page 276
We must now look for the state-by-state equality of supplies of and demands for contingent claims. Recall
that is now to be interpreted as the vector of consumption plans by the representative young person
for the retirement period, contingent on the state realized. Asset market clearing now requires the vector
equation
where, as before, we are taking for granted the determination of K+ and w+ via other equilibrium relationships,
given w and d.
In the certainty case, we were able to use the government budget constraint, , together with
the accounting relationship between D and B, to eliminate both and B from the asset market-clearing
+
226
227
condition. Under uncertainty, the combination of these two relationships is no longer sufficient. Whereas
before, constraining the value of the government's bond sales or purchases determined the quantity (given d),
the government now may choose among various combinations of state-contingent claims (i.e., deal in bonds
of different risk characteristics). Moreover, the same can be said of the firm. Thus, if we take as a starting
point that the government only issues riskless bonds (buys riskless bonds in the case of negative government
debt), we have still not pinned down temporary equilibrium because the risk characteristics of the firm's debt
have not been determined.
If the corporation is restricted to riskless debt and the government is restricted to riskless debt at the margin,
the argument goes through much at it did in the risk-free analysis. In that case, extra bonds issued by the
corporation generate not only the exact withdrawal in value of government bonds (as a consequence of extra
taxes paid) required to maintain financial market equilibrium but also the matching change in the
state-by-state contingent claims. We can see this by inspection of equilibrium condition (7). If the firm issues
another unit of bonds, it adds directly one dollar to each component of the vector B+ of claims supplied on the
right side but subtracts an amount (1 - t) from the anticipated recovery from owning equity in each state that
might occur. If the government reduces its issue of bonds so as to deliver t less in each state, the set of
financial claims supplied will continue to balance the claims demanded.
Â
Page 277
Note that the condition calls for

riskless bonds to be issued by the
corporation but for the
government to issue riskless
bonds "at the margin." The
requirement specifies the way the
government must react to
variations in distribution tax
receipts. The government retains,
however, freedom to set the risk
characteristics of what we may
call its basic debt issue, which we
might specify as the debt issued if
there were no corporate
distributions in the period. The
total value of the debt issue is
determined by the budget
constraint (which could itself be
lifted by addition of, say,
lump-sum taxes to the fiscal
repertoire).
This freedom cannot, however, be unpredictably exercised. In order for the agents of the model to be able to
formulate rational expectations, the risk characteristics of the basic government debt issue must be specified
in advance (e.g., by a formula relating to the characteristics defining the different states of the world). This is
a distinct addition to the model introduced by the extension to uncertainty. In the certainty case, the
government was wholly predictable because it had no degree of freedom within the budget constraint.
Uncertainty brings with it a range of options in each situation. Government policy will influence the course of
the model economy (because the overlapping generations are, by assumption, not linked through bequests).
In this case, the main effect of policy is to influence the choice among risky alternatives. It might be
227
228
interesting to explore the question of optimal fiscal policy in this model; however, for present purposes, the
point to stress is the requirement for predictability, involving prespecification of the characteristics of the
basic debt and the rule for responding at the margin to variations in distribution tax receipts.
The latter requirement has thus far been spelled out for the case in which the corporation issues only riskless
debt. However, we can readily formulate a more general rule relating private and marginal public debt to
preserve the property of the model whereby the firm's financial policy has no real effect.
The investment and employment decisions in the model are essentially determined by the path of w and d,
wages and contingent dollar claim prices. Suppose we had a path of w and d such that (7) is continually
satisfied when the firm is restricted to riskless debt and the government to riskless debt at the margin. Then
equilibrium condition (7) provides us with a general condition on government
Â
Page 278
finance that assures that the path is unaffected by other choices of private financial behavior, namely,
In words, in order for corporate financial policy to have no real effect, government policy must be effective to
issue basic debt, consisting of a predetermined package of contingent claims,
and to purchase tB+, that is, a fraction t of the bonds issued by the firm. The requirement (7) of equilibrium
implies that if any other government policy is followed, private financial policy will have real effects, and
hence further restrictions on behavior are required to close the model.
4
Concluding Remarks
With respect to the narrow question of whether the neutrality results obtained earlier in an all-certainty model
carry over to a world of uncertainty, the analysis yields a clear conclusion. Uncertainty introduces degrees of
freedom to government choices that must be balanced by restrictions. In particular, if corporate distributions
are not to have real effects, the government must use the distribution tax proceeds to purchase the debt of the
distributing firm (or, equivalently, buy back a set of government-issued contingent claims to generate the
same effect on the supply of each type in the market). As suggested in the introductory section, there is an
economic rationale for such a policy on the part of the government. It really is the same policy that we would
expect shareholders to follow in rebalancing their portfolios following a distribution. The distribution by the
corporation changes the characteristics of the equity claim. The portfolio can be restored to its predistribution
characteristics by using the dividend to buy the firm's bonds. When we recognize that the government is an
implicit equity owner, by virtue of the distribution tax, the restriction called for by the theory seems wholly
reasonable.
However, it is hard to imagine any actual government carrying out this program. The systems used to account
for government are typically weak in the dimension of measuring either liabilities to make future payments or
claims to payment of taxes in the future already
Â
228
229
Page 279
established by existing policies. To take an example, existing accounting conventions in the United States
record the loss in tax revenue due to contributions to tax-favored retirement plans and make no allowance for
the resulting increase in the present value of future liabilities implied by the same transaction [Kotlikoff
(1984) develops this general theme more fully].
Unfortunately for the model described here, if the government does not follow the specified behavior,
incremental distribution tax revenues will have real effects. The problem this presents is not simply to
describe these effects but rather to understand how it is that equilibrium is determined at all. For the argument
that corporate distributions are not determined in the model holds generallyit is simply a consequence of price
taking by private agents in financial markets. As a result, corporate distributions cannot be predicted. If an
unpredictable aspect of the model has real consequences, how can agents have rational expectations? Thus I
must conclude this paper on a note of puzzlement.
Note
I would like to express my appreciation for the helpful discussions with Pete Kyle, David Starrett, Joseph
Stiglitz, and seminar participants at the National Bureau of Economic Research and at the Workshop on
Economic Structural Change of the International Institute for Applied Systems Analysis.
References
Arrow, K. J. and F. H. Hahn (1971), General competitive analysis, San Francisco: Holden-Day.
Bradford, D. F. (1981), "The incidence and allocation effects of a tax on corporate distributions," Journal of
Public Economics, XV: 1-22.
Bradford, D. F. and the U.S. Treasury Tax Policy Staff (1984), Blueprints for Basic Tax Reform, 2nd ed.,
Arlington, VA: Tax Analysts.
Kotlikoff, L. J. (1984), "Taxation and savings: A neoclassical perspective," Journal of Economic Literature,
XXII: 1576-1629.
Modigliani, F. and M. H. Miller (1958), "The cost of capital, corporation finance and the theory of
investment," American Economic Review, XLVIII: 261-97.
Muth, J. (1961), "Rational expectations and the theory of price movements," Econometrica, XXIX: 315-35.
Â
Page 281
10
Pitfalls in the
Construction and Use
of Effective Tax Rates
with Don Fullerton
229
230
I
Introduction
The easiest and most common approach to estimating effective tax rates on investment has been to calculate
actual taxes paid as a proportion of capital income. This ''flow of funds" approach is particularly useful for
income effects to capital owners, revenue effects to government, or generally for discussing the relative size
of the public sector. 1 Some have also used this approach to capture the different incentive effects for using
capital in different industries. The implicit assumption is that marginal tax rates in a given industry are not far
from the ratio of actual taxes to capital income in that industry. Harberger (1966) estimated the efficiency
cost of differential capital income taxation using this approach, as did Shoven in his (1976) correction to
Harberger. The approach is still used in recent general equilibrium estimates by Fullerton, King, Shoven, and
Whalley (1981).
A new approach is now emerging, based on the pioneering work of Hall and Jorgenson (1967). Their cost of
capital formulas have long been used to analyze investment and the incentive effects of tax policy changes.
More recently, the formulas have been used to estimate effective marginal tax rates, as in Hall (1981) and
Jorgenson and Sullivan (1981). Tax rates based on the cost of capital approach have been used to recalculate
Harberger-type efficiency costs, as in Gravelle (1981), and to recompute general equilibrium effects, as in
Fullerton and Gordon (1981).
This newer approach considers a "hypothetical project" of a dollar invested in a particular asset to be used in
a particular industry. (Some versions of the approach also assume that the investment is
Â
Page 282
maintained in real or nominal

terms by subsequent
reinvestment.) The view of taxes
is prospective in the sense that the
cost of capital formula looks at the
expected change in future tax
liabilities, usually discounted to
the time that the original
investment takes place. The
method can simultaneously
incorporate actual depreciation
rates, type of finance, eligibility
for investment tax credit,
accelerated depreciation rules, and
depreciation at historical cost. It is
greatly facilitated by the
availability of estimates for
depreciation rates of different
assets, such as those in Hulten and
Wykoff (1981).
The prospective nature of the cost of capital (hypothetical project) approach implies that it is probably more
useful for investigating incentive effects. It measures the expected tax consequences if a given investment is
undertaken. It also concentrates on marginal effects by considering a particular unit of investment. Though
we see many potential benefits of using this approach, the purpose of this paper is to investigate some of its
dangers. In particular, we shall illustrate three points that should be considered by any study which uses the
230
231
cost of capital approach.
First, as mentioned, the cost of capital method considers the expected future tax liabilities associated with a
hypothetical project, discounted to the time that the original investment takes place. Though the investment
tax credit has immediate consequences, other features of tax systems do not. Accelerated depreciation, for
example, has the effect of delaying some tax liability. As a result, effective tax rate estimates will necessarily
depend upon the after-tax interest rate or other rate used for discounting. We shall show this sensitivity below
by plotting a tax rate estimate against the interest rate used to obtain it.
Second, the tax law allows some assets to be depreciated at rates faster than their values decline. With
investment tax credits and with the deductibility of nominal interest payments, the asset need not earn a
positive marginal product for the investor to receive a normal return. Though the implied subsidy might be
measured in a meaningful way, the rate of subsidy might not be. When the investment's return in the
denominator of an effective tax rate formula approaches zero, the rate of subsidy can be arbitrarily high.
Similarly, on an asset with a low real return, a positive tax can be an arbitrarily high portion of it. This
problem can be dealt with by using the numerators of these tax rate estimates alone to describe the effective
tax wedge on a particular asset in a particular industry.
Â
Page 283
Third, effective tax rate estimates depend on assumptions about how inflation affects nominal interest rates.
Hall (1981) and Jorgenson-Sullivan (1981) effectively assume that nominal interest rates increase by the
inflation rate over one minus the corporate tax rate. This increase is just enough to keep the real after-tax
interest rate constant for corporations. If all taxpayers faced the same tax rate as corporations, and if the rules
for measuring the income from real investments were perfectly indexed for inflation, then a strong a priori
case could be made for this behavior of interest rates. The real consequences of given decisions to borrow,
lend, and invest would then be independent of inflation rates. However, historical cost depreciation,
nonuniform tax rates, and other tax features tend to weaken this a priori case. Indeed, Feldstein (1980) has
argued that the monetary authorities have acted so as to impose Fisher's Law, keeping the real before-tax
interest rate invariant with respect to inflation. We show below how tax rate estimates differ according to
whether nominal interest rates increase by just the rate of inflation, or by enough to keep real after-tax rates
constant.
This chapter does not seek to estimate new or better effective tax rates. It only seeks to investigate the
sensitivity of existing estimates to some of the issues just described. These can be clarified adequately within
the context of fairly simple and straightforward cost of capital formulas such as those used by Hall (1981)
and Jorgenson-Sullivan (1981). In order to be particularly careful about the assumptions used in this
procedure, we rederive the cost of capital in section II.1. In order to be particularly careful about what is
being estimated, we describe an array of possible tax rate definitions in section II.2. Then in section II.3, we
state the parameters of the investment and tax systems, chosen for comparability with Hall (1981). We also
display the possible outcomes for savers and investment returns. These constitute the components of effective
tax rates.
In the three parts of section III, we elaborate on each of the three issues raised above. Tax rate estimates are
shown to be sensitive to the interest rate in section III.1. The fact that tax rate definitions are very nonlinear
relationships is illustrated in section III.2. The effect of inflation on nominal interest rates also significantly
affects the estimates, as shown in section III.3. It should be clear by the end of this chapter that the sensitivity
of tax rate estimates implies that one can obtain a wide variety of tax rate estimates with different choices of
parameter values and other assumptions.
231
232

Page 284
The problems emphasized in this paper involve primarily mechanical features of the analysis. That is, they
concern potential misunderstanding of the tax rate formulas and of the assumptions often encountered. We
also touch on some underlying modeling problems through the course of the paper. Foremost among these is
the question of the true relationship between inflation and the interest rate. There are, however, several
additional aspects of the use of effective tax rate estimates which deserve attention. In the concluding
remarks of section IV, we allude to further work we are doing on these problems.
II
Analytical Framework
Because we feel that previous studies have not been explicit enough about what they were estimating, we
devote considerable attention at the outset to deriving and defining different sorts of tax rates. Any of these
might be estimated by a particular study.
II.1
The Cost of Capital
We begin with a simple expression for the cost of capital, that is, the annual market rental price of a unit of
capital, predicted to obtain in a competitive market equilibrium. Although a similar derivation has been
exposited many times, it will be helpful to have a restatement of the underlying assumptions and the
interpretations of different variables.
Define ρ to be the expected real rate of return to the hypothetical project, net of economic depreciation at the
exponential rate δ. 2 In light of our introductory comments, we require a model incorporating the dependence
of ρ on the nominal interest rate i and on the rate of inflation π. However, many complexities can be safely
ignored. Hall (1981), for example, considers (but does not really use) the possibility that a proportion of
accrued capital gains are taxable at the statutory corporate tax rate u. Jorgenson and Sullivan (1981), on the
other hand, allow for (but then abstract from) the possibility that the acquisition cost q and the rental price c
of the asset are arbitrary functions of time, rather than assuming only that they increase with inflation.
Jorgenson and Sullivan also consider the possibility that the rate of economic depreciation is an arbitrary
function of time, rather than using our simpler assumption that true depreciation is at
Â
Page 285
constant exponential rate δ. They allow depreciation deductions as an arbitrary function of time, while we
assume the tax law allows depreciation deductions on a historical cost basis at constant exponential rate δ'.
An investment tax credit at rate k completes the description of our hypothetical real investment project and its
tax consequences. A corporate purchaser of a unit of real capital incurs an immediate after-tax-credit expense
of (1 - k)q, and subsequently obtains a cash inflow, expressed as a function of τ, the time since acquisition of
the asset. This cash inflow includes rental at a rate that starts at c and grows at the rate of inflation π. The
quantity of capital embodied in the investment declines at the depreciation rate δ. At time τ the rental receipts
thus equal (1 - u)ce(π-δ)τ after the corporate income tax. The cash inflow also includes tax reductions due to
depreciation allowances, which at time τ equal uqδ'e-δ'τ (depreciation at rate δ' is allowed on the remaining
basis qe-δ'τ). 3 By changing k, u, and δ', the tax authorities change the attractiveness of these net-of-tax cash
flows, given q, c, and π. Since (at least) q and c are endogenous to the system, changes in the tax rules will
232
233
ultimately be reflected in changes in the values of q, c, or both.
The power of the analysis is based upon the valuation of such cash flows. More specifically, it is based on
valuation relative to available alternatives. Prominent among these alternatives is the purchase or sale of debt.
The analogue of buying a machine is buying debt, or lending. If the market interest rate i is constant, for an
initial outlay of $1, a corporation can accumulate e(1-u)iτ dollars by time τ, where the factor (1 - u) in the
exponent reflects the taxation of interest receipts. Of course, most nonfinancial corporations are sellers of
debt, not buyers. The emphasis therefore is usually on the deduction of interest outlays and not the taxation of
interest receipts. Note that this deduction is a logical extension of the taxation of interest receipts. It is not, as
sometimes made to appear, an explicit subsidy of corporate borrowing.
If borrowing and lending are unconstrained, and if real investment is riskless, it is possible for a corporation
to undertake offsetting transactions, by selling debt and buying an equal amount of real capital. Explicitly or
implicitly, most analyses depend upon the elimination of any possible pure surpluses from such transactions
to determine the equilibrium relationship among i, q, and c (given π and the tax rules). More prosaically,
most analyses represent the
Â
Page 286
corporation as discounting nominal cash flows at the "after-tax nominal interest rate," (1 - u)i.
In equilibrium, then, the present value of the nominal cash flow from a unit of capital, as summarized above,
must just equal the initial outlay. This implies
Explicit integration leads to a relatively simple relationship between the gross of depreciation rental rate, c/q,
and the interest and inflation rates:
This is our basic equation for later computations. Note that this equilibrium condition is independent of the
actual financing method of the corporation; it does not matter whether the source of the investment funds is
debt or equity. The option of arbitrage between debt and real capital implies equation (2). 4
Since the original Hall and Jorgenson treatment of this subject (1967), the notation z has been the
conventional symbol for the discounted sum of depreciation deductions on a one dollar investment. Therefore
(2) can also be written as
where it must be remembered that z depends upon i and u.
For changes in the tax parameters u, k, and δ', it is conceptually straightforward to calculate the effect on the
equilibrium social rate of return, ρ, which equals c/q - τ. This can be done for different combinations of i and
π. Commonly, though, a further simplification is adopted, namely, an assumed relationship between i and π.
This reduces the number of cases to consider, for the whole system described by (2) then has a single
exogenous parameter, π.5 While a number of relationships are possible, two particular assumptions about
inflation and the nominal interest rate are often encountered. The first is a strict version of Fisher's Law. If we
233
234
let i0 represent the interest rate presumed to prevail in the absence of inflation, Strict Fisher's Law says that
Â
Page 287
The argument for this result is simply that this adjustment leaves all real borrowing and lending opportunities
independent of the rate of inflation. Implicit is the absence of taxes on interest. With a tax at a flat rate u on
net interest receipts (which implies a deduction for interest payments), the same argument predicts what
might be called Modified Fisher's Law: 6
Theory does not give us firm

predictions about the relationship
between i and π in a world of
imperfect income measurement
rules, diverse marginal tax rates,
nonlinearities, noise, and other
considerations. Feldstein and
Summers (1978) estimate that i
has varied slightly less than point
for point with π in the U.S. since
World War II. On the other hand,
Hall (1981) explicitly assumes
Modified Fisher's Law. Jorgenson
and Sullivan (1981) postulate
constancy of the real rate of return
on investment after the corporate
tax, citing empirical work by
Fraumeni and Jorgenson (1981).
This procedure is equivalent to
assuming Modified Fisher's Law
when arbitrage with corporate
bonds is encompassed by the
model.7
Later we consider the choice between (4a) and (4b). For our illustrative calculations, we consider three
situations: no inflation, 10 percent inflation with equation (4a), and 10 percent inflation with equation (4b).
The real interest rate to a bond holder with no tax is the same in the first two scenarios, but the real interest
rate after tax at rate u differs. In the first and third scenarios, the real after-corporate-tax interest rate (1 - u)i -
π is the same, but the real interest rate for a nonprofit (nontaxable) institution differs. There is no real interest
rate which is the same in all three cases, and so we studiously avoid defining any parameter as the real
after-tax interest rate. Instead, we take as a basis of comparison the interest rate i0 that would prevail with no
inflation.
II.2
Effective Tax Rates
234
235
The concept of an effective tax rate on capital refers to some measure of the difference between ρ, the real
social rate of return earned on a real asset, and s, defined as the rate of return received by the person
Â
Page 288
or institution financing its purchase. Then we can define the tax "wedge," tw, as
This wedge can be thought of as an annual levy on the specified financer with respect to a dollar's worth of
the asset in question. It may be either positive or negative.
It is usual to express tw as a ratio to either the social return or the saver's return. The first is a tax rate on a base
that includes the tax: a "tax inclusive" rate in the language of the Meade Report (1978). Since the base is
gross-of-tax, we refer to it as a "gross tax rate" tg. The other tax rate is on a "tax exclusive" or net-of-tax basis,
and is referred to here as a "net tax rate,'' tn. These rates are related to ρ and s by
Notice that these rates are nonlinear functions of s and ρ and may behave rather erratically in some
circumstances. Particular care must be used where the denominator of one of these formulas approaches zero,
or passes from positive to negative. In later sections we shall see examples of the practical relevance of this
erratic behavior.
Different values of ρ and s can be derived for
(a) assets with different depreciation rates δ,
(b) assets with different tax rules (δ' and k),
(c) savers with different tax circumstances (u, m),
(d) different types of finance (bonds, stock, direct ownership), and
(e) different i and π combinations.
In particular, we shall focus on

an "effective corporate tax rate,"
a "total effective rate of tax on
bond financed corporate
investment," and a "total
effective rate of tax on equity
financed investment."
In performing this analysis however, there is some question as to what should be taken as constant. Since the
interest rate i is a price established on a market in which all can trade, it is arguably the natural fixed point.
Given this interest rate and a single tax rate u for all traders, the social return ρ would be determined by the
equilib-
235
236
Â
Page 289
rium condition (2) for arbitrage between bonds and real capital. In concluding remarks we touch upon the
possibility of corporate and noncorporate arbitragers with different tax rates. For now, however, we make the
customary assumption that the market is dominated by corporations with tax rate u. Because the corporation
arbitrages between real capital and bonds yielding (1- u)i - π, it is either a borrower or lender at that real
after-tax interest rate. In this sense, we can take (1 - u)i - π as the net return to savings of the corporation, Sc.
Thus, the corporate tax wedge, the effective gross rate, and the effective net rate are given by
Note that the model developed in the previous section implies ρ and i are functionally related to each other,
and i is functionally related to π. Hence the model implies values for the corporate tax wedge and the
corporate effective tax rates as functions of π.
If we imagine a corporation choosing to purchase a dollar's worth of real assets with some funds it has in the
bank, the net of tax real return must be (1 - u)i - π; otherwise (2) would not be satisfied. This can be thought
of as income to the stockholders, who are taxed on it at marginal rate me. This personal rate is designed to
capture the effective personal tax on these earnings when part may be paid as dividends and part retained. It
should also account for the low effective personal rate on accrued capital gains resulting from retentions. The
net return to stockholders on equity is thus Se = (1 - me)[(1- u)i - π]. The tax wedge on equity, the effective
gross rate, and the effective net rate are given by
Â
Page 290
An individual debt holder with marginal tax rate md receives a real return on bonds of (1 - md)i - π after taxes.
Call this return sd. 8 We are entitled to compare sd to the social return on corporate investment, and the
difference is customarily referred to as the effective tax on corporate investment financed by debt. Thus the
tax wedge on debt, the effective gross rate, and the effective net rate are given by
236
237
In two senses, these effective tax rates have nothing to do with whether the investment is actually financed by
issue of debt. First, because i is a market interest rate, any debt holder with tax rate md will earn sd. The
corporation earns ρ on a particular investment. The values ρ and sd are all that is required to define the
effective tax rates, independently of any connection between the two. Second, when the corporation makes its
real investment decisions by comparing the returns on capital and debt, there is a connection between ρ and i.
This relationship depends on potential and not actual arbitrage, however. Thus the td expressions are not only
defined, but relevant for analysis regardless of whether debt finance is actually used.
Two points may be noted from these formulas before we proceed to illustrate them. First, if the personal rate
md happens to equal the corporate rate u, then the total tax on debt (equation 10) is equivalent to the corporate
rate alone (equation 8). When Hall or Jorgenson and Sullivan report effective corporate tax rates , we can
reinterpret them as total tax rates on a debt financed investment where the lender has a tax rate m equal to
u. d
Second, if md is zero, then equation (10) implies
If we use equations (2) and

(4b) to obtain ρ, this is exactly
the tax rate on debt calculated
by Hall (1981). Thus Hall's
tax rates on debt can be
thought of as the lowest
extremes of a spectrum from
md = 0 to md = u. For the other
extreme, we can simply look
at the effective
Â
Page 291
Table 10.1
Definition of different effective tax rates
Superscripts
Subscripts w g n
Wedge Gross Net
c
Corporate
e
Equity
237
238
d
Debt
where s = (1 - u)i - π
c
s = (1 - m )[(1 - u)i - π]
e e
s = (1 - m )i - π
d d
corporate tax rate . Below, we reproduce Hall's results and then recalculate them for different real net of tax
interest rates and for different assumptions about how inflation affects nominal interest rates.
The various effective tax rate expressions are drawn together and summarized in table 10.1.
II.3
Parameter Values
Having specified the

mechanisms determining the
social rate of return and the
saver's rate of return, we can
explore the behavior of the
various effective tax rates under
different assumptions about
parameters. Hall (1981) has
chosen a particular classification
of investment types and saver
types; it will facilitate our
discussion to adopt the same
parameter values.
First, take u to be .46, the marginal tax rate for corporations where nearly all corporate investment takes
place. For md, Hall uses a value of zero, on the assumption that all bonds are held by tax-exempt institutions.
As we have mentioned, we can also consider the case of md = u = .46 with no extra calculations or table
space. Thirdly, let me equal .28, the value chosen by Hall. He assumes that the typical stockholder is in the 40
percent bracket, receiving one-half of corporate equity income as fully taxable dividends and the other half as
Â
Page 292
Table 10.2
Parameters for three asset categories
1 2 3
Definition Parameter Equipment Structures Intangibles
Effective investment tax credit rate k .1 0 0
Economic depreciation rate δ .1 .03 .1
Tax depreciation rate δ' .15 .06 ℜ∝
capital gains. Only 40 percent of the latter are included in the individual income tax base. We regard the
figure of .28 as somewhat high. The value of deferral and of the write-up of capital gains basis at death
probably cut the effective proportion of accrued gains included in taxable income to something like 20
238
239
percent. 9 With this assumption, me would be .24. For purposes of illustrating the characteristics of the tax
system, however, the difference is not of much importance.
Hall identifies three real assets. "Equipment" depreciates at 10 percent per year, receives a 10 percent
investment tax credit, and is allowed depreciation deductions at 15 percent per year. "Structures" depreciate
at an annual rate of 3 percent, receive no investment tax credit, but are allowed accelerated depreciation
deductions at 6 percent per year. Finally, "intangibles" (e.g. advertising or R&D) are assumed to depreciate at
10 percent, receive no tax credit, but may be written off immediately. These asset characteristics are
summarized in table 10.2. Tax rate estimates will be sensitive to these assumed parameters, but they do
represent plausible examples of real asset characteristics. Notice that equipment and intangibles are
technologically identical (have the same depreciation rate). We can thus attribute their different results purely
to differences in tax treatment.
It remains to specify the interest rate. Hall chooses as his starting point an assumed real after-corporate-tax
interest rate of .04. In the absence of inflation, this is our (1 - u)i0. We also consider .02 and .06 as alternative
assumed values of (1- u)i0. Whereas Hall takes Modified Fisher's Law (4b) as given, we want to look at the
effect of varying this assumption. We take Strict Fisher's Law (4a) as an alternative. In each case we display
the results for (1 - u)i0 equal to .02, .04, and .06.
Â
Page 293
As mentioned above, we consider 10 percent inflation with (4a), 10 percent inflation with (4b), and zero
inflation, a rate at which the two versions of Fisher's Law imply the same interest. We thus consider three
inflation assumptions, three distinct saver types, three asset types, and three values of (1- u)i0. Any single
interest-inflation combination can be used in one direction to determine the social return on each investment,
ρ, or in the other direction to determine the return to each saver, s. The values of ρ and s for all of these
combinations are displayed in table 10.3.
Readers are advised to spend a few minutes absorbing table 10.3. Notice, for example, that because
intangibles are expensed, this form of investment is effectively untaxed at the corporate level (ρ = sc). The
column of real returns on intangibles just shows the behavior of the real after-corporate-tax interest rate under
the various assumptions. With Strict Fisher's Law and low values of (1 - u)i0, this interest rate is negative. The
column of real returns to tax-exempt debt holding savers, sd, shows what happens to i - π under the various
assumptions. With Modified Fisher's Law, the real rate of return to tax-exempt debt holders rises sharply with
inflation.
Readers can now construct their own effective tax rates. First, choose a row of table 10.3. Second, subtract
from any real social rate of return ρ, any saver's real return s in that row. Third, decide whether to divide by
the former, the latter, or not at all. In the remainder of this chapter, we discuss some of the issues to keep in
mind during this exercise.
III
Three Caveats
III.1
Tax Rates are Sensitive to the Interest Rate
The first of our three points is very simple, now that the apparatus of section II is available: effective tax rate
estimates depend on the assumed interest rate. Given π, the interest rate determines ρ. Thus each tax rate,
such as , , or , is a function of the interest rate. Different tax estimates will result from different interest
rates used as input.
239
240
Later, we shall develop the point that tax rates are also sensitive to how inflation affects nominal interest
rates. To abstract from that point here, consider the simple case with no inflation. Table 10.4, part A, displays
the various gross tax rates under these circumstances.
Â
Page 294
Table 10.3
Social rate of return and return to savers under various assumptions
Real rate of return to savers, s (in %)
Real social rate of return, Corporations alone, or debt holders Equity holders with me =
ρ (in %) with .28 se
md = .46 sc.
(1 - u)i0 Equip. Struc. Intang.
A. With no inflation .02 1.0 3.1 2.0 2.0 1.4

.04 3.9 6.4 4.0 4.0 2.9
.06 6.9 9.8 6.0 6.0 4.3
B. With 10% inflation and Modified Fisher's .02 4.3 4.8 2.0 2.0 1.4
Law (eq. 4b)
.04 7.2 8.2 4.0 4.0 2.9
.06 10.1 11.6 6.0 6.0 4.3
C. With 10% inflation and Strict Fisher's Law .02 -1.9 -2.4 -2.6 -2.6 -1.9
(eq. 4a)
.04 0.7 0.6 -0.6 -0.6 -0.4
.06 3.5 3.9 1.4 1.4 1.0
Â
Page 295
Table 10.4
Effective tax rates as percent of capital income gross of tax
A. Effective tax rates with no inflation
= corporate rate only, or total rate on debt if = total rate on equity if m = .28 = total rate on debt if m
e
md = .46 d
(1 - u)i0 Equip. Struc. Intang. Equip. Struc. Intang. Equip. Struc.

.02 -104 35 0 -47 53 28 -278 -21
.04 -2 37 0 26 55 28 -89 -16
.06 13 39 0 38 56 28 -60 -13
B. With π = .1 and equation (4b): inflation adds more than point-for-point to nominal interest
= corporate rate only, or total rate on debt if = total rate on equity if m = .28 = total rate on debt if m
e d
md = .46
240
241
(i - u)i0 Equip. Struc. Intang. Equip. Struc. Intang. Equip. Struc.

.02 54 59 0 67 70 28 -183 -153
.04 44 51 0 60 65 28 -122 -95
.06 40 48 0 57 63 28 -95 -70
C. With π = .1 and equation (4a): inflation adds point-for-point to nominal interest
= corporate rate only, or total rate on debt if = total rate on equity if m = .28 = total rate on
e
md = .46
(i- u)i0 Equip. Struc. Intang. Equip. Struc. Intang. Equip. Struc.
.02 -38* -8* 0 1* 22* 28* 296* 254*
.04 181 193 0 158 167 28* -896 -1043
.06 60 64 0 71 74 28 -219 -188
* denotes tax rates with anomalous signs as described in the text.
Â
Page 296
Let us pause to study the numbers in table 10.4A. Looking across any row, say for (1- u)i0 = .04, we see the
expected wide range of effective rates applicable to different holders of different types of claims on different
forms of real capital. The effective corporate tax rate on intangibles is zero, because this asset receives
immediate expensing, which is equivalent to eliminating the tax. The corporation equates ρ on this
investment to the after-tax return it can earn on other assets, (1 - u)i. Since i is the rate of return received by
tax-exempt bondholders, their implied effective tax rate is negative:
percent. The holder of equity, on the other hand, pays a tax of 28 percent (the assumed value of me) on ρ = (1
- u)i.
Still for (1- u)i0 = .04, the effective corporate rate on structures is 37 percent, below the statutory rate of 46
percent. This difference reflects depreciation allowances in excess of economic depreciation. The higher 55
percent total tax rate on equity simply reflects the "double taxation" of corporate income. Finally for
structures, the rate on debt is -16 percent. This subsidy is less than the subsidy for intangibles because
structures do not receive immediate expensing. This asset has a higher marginal social rate of return while the
return to debt-holding savers is the same.
The three tax rates for

equipment follow a similar
pattern. Investment tax credits
and accelerated depreciation
imply a near zero corporate rate,
while the rate on equity is
higher and debt is lower.
Now we turn to examination of the columns of table 10.4A, that is, to the effect of varying the assumed
interest rate. We note immediately that the effective taxes on intangibles are unaffected while those on
equipment move rather dramatically. The former result follows from the fact that the various tax rates are
simple multiples of (1 - u), independent of i. Put another way, there are no delayed taxes or benefits with
immediate expensing, so the discount rate does not matter.
241
242
The behavior of the effective taxes on equipment can be understood by reexamining three aspects of equation
(2). First, the return on the investment is indeed taxed at rate u. Second, it receives investment tax credit at
rate k. Third; it receives accelerated depreciation since δ' > δ (we can ignore historical cost problems here
since π is zero). When the discount rate (1 - u)i0 is low, the future depreciation advantages are relatively more
important. Together with the investment tax credit, they outweigh the corporate tax, and a net
Â
Page 297
subsidy results. As the discount

rate increases, accelerated
depreciation becomes less and
less important until the
corporate tax outweighs the
credits and deductions, so a net
tax results.
The effective tax rates for structures turn out to be less sensitive to the interest rates in the range considered
here. Because structures do not qualify for the investment credit, the effective corporate tax rate must be at
least zero.
The sensitivity to the interest rate remains when there is inflation. Let us continue to delay the issue of how
inflation affects nominal interest rates. For now, just consider the case of equation (4b), where the real
after-corporate-tax interest rate (1 - u)i - π is constant. The nominal interest rate starts with no inflation at i =
i0 and increases to i = i0 + π/(1 - u) with inflation at rate π of 10 percent. Resulting effective tax rate estimates
are shown in table 10.4B.
Again the tax rates with (1 - u)i0 = .04 reproduce estimates from Hall's paper. Again the tax rates on
intangibles are insensitive to this interest rate, except for . The insensitive tax rates result from the fact that
the real after-tax interest rate is constant. For tax-exempt bondholders, however, the real return rises because
the equilibrium market interest rate increases by more than the inflation rate. The higher is the inflation rate
(relative to the rate of return) the larger is this subsidy.
Tax rates for equipment and structures in table 10.4B may appear to be fairly stable, but only because of the
range for i0. Hall reports a 44 percent tax rate at (1 - u)i0 = .04, while the table shows a lower rate (40 percent)
at .06 and a higher rate (54 percent) at .02. In fact, as (1- u)i0 is reduced further, the tax rate gets even higher.
This sensitivity is displayed dramatically in figure 10.1, which plots for equipment against the real after-tax
interest rate (1 - u)i0 = (1- u)i - π with 10 percent inflation. With (1 - u)i0 between zero and .06, one obtains
tax rates anywhere between 40 percent and 100 percent.
III.2
The Denominator can be Zero
Our second caveat concerns the manner in which a given tax wedge is expressed. Even if we agree on an
interest rate (in order to set aside the problems of the previous section), the resulting tax rate estimate might
imply that the entire return to investors is financed by government through investment tax credits and
accelerated
Â
242
243
Page 298
FigureÂ 10.1
CorporateÂ taxÂ rateÂ onÂ equipmentÂ withÂ πÂ =Â .1Â andÂ equationÂ (10Â 4b)
(inflationÂ addsÂ moreÂ thanÂ point-for-pointÂ toÂ nominalÂ interest)
depreciation allowances that

outweigh the corporate tax. The
required return on the
investment, ρ, may be zero. In
this case the rate of tax tg = (ρ -
s)/ρ is not defined. If the saver
obtains a positive rate of return,
it is paid entirely by a tax
subsidy. While the gross tax
rate is undefined, the net tax
rate is defined and equals -100
percent. However, there is no
insurance against s going to
zero either. The real net of tax
return of savers has even been
negative in riskless terms with
inflation. The remaining
alternative is merely to report
the total wedge tw = ρ - s. This
value can be interpreted as a
''property tax" rate, the
percentage of asset-value paid
in tax each year.
To illustrate the relevance of this problem, consider the corporate tax rates on equipment without inflation.
These tax rates are reported for selected interest rates on the left side of table 10.4A. They are also
reproduced in figure 10.2 for all interest rates between zero and (1- u)i0 = .06. With high interest rates (.06),
accelerated depreciation has a low present value, and a small net tax results. At (1- u)i0 = .04, as reported in
Hall, tax credit and depreciation advantages just about balance the tax at rate u, with an effective corporate
tax rate of nearly zero. With lower interest rates, the depreciation advantage is more important, ρ is always
less than (i - u)i , and (the numerator of ) is always negative. Near (1 - u)i = .014, the subsidy rate can be
0 0
as high as desired, since ρ approaches zero. With interest rates below .014, however, ρ in the denominator is
also negative, with the anomalous result that is
Â
243
244
Page 299
Figure 10.2
CorporateÂ taxÂ rateÂ onÂ equipmentÂ withÂ noÂ inflation
Table 10.5
Alternative expressions for the effective corporate rate (equipment, no inflation)
(1 - u)i0
.01 -1.45 322 -145
.02 -1.02 -104 -51
.04 -.08 -2 0
.06 .93 13 16
.08 2.00 20 25
Note: is a percent of asset-value paid in tax each year. The other tax rates are expressed as a percent of the
capital income flow.
positive. In no sense is there a positive tax rate in this region, since ρ < (1 - u)i , yet will be positive.
0
Table 10.5 summarizes the possibilities for this example of the effective corporate tax on equipment with no
inflation. The use of is not really acceptable, because the subsidy at .01 appears as a +322 percent tax. The
net rate seems to make more sense, since the sign is correct here. However, table 10.3 reveals that s could
c
also be negative in the denominator of . In such a case, a subsidy would again appear as a positive net tax
rate. Furthermore, both gross and net tax rates are subject to misleadingly wide variation when their
denominators are close to zero.
These considerations seem to point toward the use of tw alone; we do not need a denominator. This effective
tax always has the right
Â
244
245
Page 300
sign and is not so sensitive to small changes in the assumption about interest rates. However, the problem of
the assumed interest rate does not vanish. The tax wedge tw does become larger as the rate of return increases,
even though the rate of tax levels off as seen in figure 10.2.
The moral seems to be that analysts should report the underlying components ρ and s, as well as summary
figures such as tax wedges, gross tax rates, or net tax rates. They should also include a discussion of the
sensitivity of these figures to the underlying assumptions. To follow our own advice, we report tax wedges in
table 10.6 for each asset and inflation scenario. This table has the same format as table 10.4.
III.3
The Inflation Assumption Matters
Fisher's original law predicted that i would increase by π in a world with no taxes. Since most investment
takes place in corporations with tax rate u, however, one is tempted to adopt the modified view that i should
increase by π/(1 - u), keeping constant the real interest rate after corporate taxes. This is the basis for Hall's
assumption.
There are, however, influences which weaken the a priori case for this outcome. First, not all investors have
the same marginal rate. If the system is dominated by nonprofit institutions with md = 0, then the argument
behind Fisher's Law would imply that i increase only by π, to keep their real (nontaxed) interest rate constant.
Second, even if all tax rates did equal u, historical cost depreciation and taxation of nominal capital gains will
tend to reduce the real net return on investments when there is inflation. These features imply that the interest
rate would tend to rise by less than π/(1 - u) with inflation.
Feldstein and Summers (1978) have estimated that inflation adds approximately point-for-point to nominal
interest rates. This is not the result of a simple Fisher's Law without taxes. Rather, it is the result of two
countervailing forces within the tax system: taxation of nominal interest at some average rate, call it m, tends
to raise i by π/(1 - m), while historical cost depreciation and taxation of nominal capital gains tend to pull the
adjustment below this level.
Jorgenson and Sullivan also use empirical work to support their assumption of a constant real after-tax rate of
return on corporate investment. This estimated constancy would appear to contradict the results of Feldstein
and Summers. The difference can be reconciled
Â
Page 301
Table 10.6
Effective tax rates as percentage of asset-value per year
A. Effective tax rates with no inflation
= corporate rate only, or total rate = total rate on equity if me = .28 = total rate on debt if md = 0
on debt if md = .46
(1 - u)i0 Equip. Struc. Intang. Equip. Struc. Intang. Equip. Struc. Intang.
.02 -1.02 1.06 0.00 -.46 1.62 .56 -2.72 -.64 -1.70
.04 -.08 2.39 0.00 1.04 3.51 1.12 -3.49 -1.02 -3.41
.06 .93 3.83 0.00 2.61 5.51 1.68 -4.18 -1.28 -5.11
B. With π = .1 and equation (4b): inflation adds more than point-for-point to nominal interest
245
246
= corporate rate only, or total rate = total rate on equity if me = .28 = total rate on debt if m = 0
d
on debt if md = .46
.02 2.32 2.84 0.00 2.88 3.40 .56 -7.90 -7.38 -10.22
.04 3.16 4.17 0.00 4.28 5.29 1.12 -8.76 -7.75 -11.93
.06 4.07 5.58 0.00 5.75 7.26 1.68 -9.56 -8.05 -13.63
C. With π = .1 and equation (4a): inflation adds point-for-point to nominal interest
= corporate rate only, or total rate on = total rate on equity if me = .28 = total rate on debt if m = 0
d
debt if md = .46
.02 .71 .19 0.00 -.02 -.54 -.73 -5.60 -6.12 -6.30
.04 1.34 1.25 0.00 1.18 1.08 -.17 -6.67 -6.76 -8.01
.06 2.08 2.46 0.00 2.47 2.85 .39 -7.63 -7.26 -9.71
Â
Page 302
by a limitation on arbitrage between bonds and real capital, implicitly invoked by Jorgenson and Sullivan.
With such a limitation, real after-tax interest rates could fall with inflation (Feldstein and Summers), while
real after-tax rates of return are constant (Jorgenson and Sullivan). In terms of our analysis, the Jorgenson and
Sullivan procedure is equivalent to assuming Modified Fisher's Law, as far as effective corporate and equity
tax rates are concerned. Calculating effective tax rates applicable to bond holders would require a separate
model of interest determination.
The choice between Strict and Modified Fisher's Laws does affect tax rate estimates. As is clear from table
10.3, the required real social returns on investment, ρ, and the real net return to savers, s, depend critically on
how inflation affects the nominal interest rate.
Consider first the Strict Fisher's Law of equation (4a). With 10 percent inflation, table 10.3 shows low and
even negative required real rates of return on investment. If the demand for capital is inversely related to this
required return, (4a) implies an increased capital stock. At the same time, table 10.3 shows lower real
rewards to saving under Strict Fisher's Law. If the supply of savings is positively related to its real net return,
this would imply a decreased capital stock. The nominal interest rate would have to increase by more than π
to encourage savers to supply enough capital to meet investment demand. The figures in table 10.3 by
themselves cannot represent an equilibrium.
Now consider Modified Fisher's Law (4b). The large nominal interest increase associated with inflation
implies a higher required real return on two of the three illustrative asset categories of table 10.3. The higher
required ρ would suggest lower incentives to invest. At the same time, however, most savers are receiving
higher real returns with inflation. These table 10.3 figures cannot be in equilibrium, either. Only if interest
rates increased by something less than π/(1 - u) would savers' desired wealth match the producers' desired
capital.
Thus equations (4a) and (4b) represent two logical possibilities for an unknown relationship: the truth is
likely to lie somewhere in between. If the assumption of Hall and Jorgenson and Sullivan is correct, then
gross tax rates with π = .1 will look like part B of table 10.4. If, on the other hand, estimates from Feldstein
and Summers (1978) are correct, gross tax rates with π = .1 will look like part C of table 10.4.
246
247
Â
Page 303
Table 10.4C illustrates some of the difficulties to which effective tax rates are subject. The corporate rates on
equipment and structures are negative when (1 - u)i0 is .02. As can be confirmed by reference to table 10.3,
these are cases of negative social rates of return. Since the return received by a 46 percent bond holder is
even more negative, these negative numbers in 10.4C reflect a positive tax. At higher values of (1 - u)i0,
relatively large positive corporate tax rates are shown. If we compare the tax rates here to those with no
inflation in 10.4A, we might draw the conclusion that inflation effects a disincentive to invest. Again, a
glance at the social rates of return in table 10.3 will confirm that ρ is sharply lower with inflation and Strict
Fisher's Law. This can only be the result of increased investment, which pushes down the marginal rate of
return. Such an outcome is to be expected, since inflation lowers real corporate borrowing costs under this
assumption. The relative increase in effective tax rates is simply the result of much smaller denominators in
the gross tax rate formula.
One last anomaly will complete the catalog of illustrations. Look at intangibles in the bottom right-hand
corner of table 10.4. We now have outrageous tax rates between -694 percent and +1335 percent. Table
10.3 or 10.6 reveals that the wedge (ρ - sd) is in fact negative for all three values of (1 - u)i0. For .02 and .04,
however, this subsidy is larger than the saver's return, and ρ is negative in the denominator. We thus have
negative numbers in table 10.4 reflecting positive taxes, and positive numbers in the table reflecting negative
taxes.
IV
Concluding Remarks
Table 10.4 exemplifies three separate conclusions. Tax rates are sensitive to the interest rate (.02, .04, .06),
tax rates are sensitive to the assumed effect of inflation on nominal interest (part B vs. part C), and tax rates
are not best expressed as a percent of gross capital income ρ.
Beyond the three major caveats discussed in this paper, there exist other more subtle problems. We are forced
by the limits of this paper to abstract from them, as have other studies. However, we might take a few
paragraphs to outline these problems for future work.
Different effective tax rates are useful for different purposes. First, one may want an average tax rate or "flow
of funds" rate to capture
Â
Page 304
income effects, as described in our introduction. Second, there are different types of marginal tax rates.
Capital income can increase because of higher rates of return or because of more investment. Since only the
latter induces an investment credit, for example, the effect on taxes is not the same. Feldstein and Summers
(1979) are not interested in the additional tax associated with another unit of investment. Instead, they seek to
measure the effects of inflation on taxes. They need to specify the effect of inflation on nominal interest rates
as well as the effect of nominal interest rates on real taxes paid.
Third, even if we agreed on

marginal tax rates for additional
247
248
investment, we might want an

effective corporate rate, an
effective personal rate, or the total
effective wedge between the
marginal product and the saver's
rate of time preference. Consider
for a moment the use of each such
rate. The assumptions of our
investment model rely heavily on
a single market interest rate. Given
this baseline for all corporate
investment, the "effective
corporate tax rates" can be used to
measure the misallocations of
capital among assets in the
corporate sector. They cannot be
used, for example, to capture
misallocations between the
corporate and noncorporate
sectors. Given the same market
interest rate as a baseline, different
savers earn different net of tax
returns. Thus "personal effective
tax rates" can be used to measure a
''misallocation" of savings in the
personal sector: not all marginal
rates of time preference are
identical. Finally, only the total
effective wedge between the
marginal social rate of return ρ
(averaged over different assets),
and the marginal rate of time
preference s (averaged over
different savers), can be used to
measure the misallocation of
consumption between present and
future periods, as caused by
capital income taxation.
Analyses of allocative and distributive effects are hampered by the questionable consistency of the tax rates
estimated with the assumption of overall equilibrium in capital markets. 10 Condition (2) expresses the
requirement that the corporation should have maximized its profits in equilibrium. There can then be no
potential for the corporation to gain by arbitrage between bonds and real capital. There are, however, other
conditions one might wish to hold. For example, it might be required that profits of noncorporate investors
from the same sort of arbitrage be eliminated. Equilibrium would then call for the analogue of condition (2),
but with the proprietor's marginal rate m replacing the corporate marginal rate u. But (2) cannot hold for both
m and u (unless they are equal).
Â
248
249
Page 305
For the model to tell a consistent

story about the effect of taxes, it is
necessary to find a way for
corporations, individuals, and
tax-exempt institutions to be in
equilibrium simultaneously. There
are two basic ways such a
reconciliation might be
accomplished. The first is to adopt
assumptions that constrain the
agents of the model. For example,
it seems natural to impose
borrowing limits on individuals,
and to limit negative positions in
(short sales of) real capital. One
might simultaneously assume that
corporate and noncorporate
technologies are distinct, so that
investment opportunities available
to corporations are not available to
other firms.
By a careful combination of such restrictions, a consistent model should be feasible. At this stage, the point to
emphasize is that the particular constraints imposed are likely to have a significant bearing on the distorting
consequences attributable to taxes. Take as an example the constraint that tax-exempt savings by individuals
are subject to fixed ceilings. This assumption is likely to eliminate any allocative effects of the sort of subsidy
to zero bracket bond holders that is apparent from table 10.4 or table 10.6.
The method of imposing constraints is likely to imply extreme specialization of portfolios. Individuals will
hold only stock or bonds, for example, not both. The rates of return in the analysis above are treated as
certain, and hence no one would hold assets generating different yields. Actual assets, however, are risky.
Thus a second approach to resolving the problem of inconsistency is to attempt an explicit treatment of risk.
This undertaking would clearly be difficult, but it is important to explore. The effect of taxing the return to
saving may be quite different from the effects of taxing risk premia. As shown in Gordon (1981), and as
estimated in Fullerton and Gordon (1981), a tax on risk premia may constitute a simple risk sharing by
government, with no distorting effect at all.
Pending modeling advances along the lines described here, we would urge those who construct and use
effective tax rates to exercise appropriate caution.
Notes
This material is based upon work supported by NBER, Princeton University, and the National Science
Foundation under grant SES 8025404. The authors are grateful to Thomas Kronmiller for research assistance.
For general discussions on these issues, the
Â
249
250
Page 306
authors are grateful to Robert E. Hall, Dale W. Jorgenson, and Mervyn A. King. The research reported here is
part of the NBER's research program in taxation. Any opinions expressed are those of the authors and not
those of the NBER or the National Science Foundation.
1. For a stimulating treatment of the problems of measuring both the numerator and denominator in such an
approach, see Fiekowsky (1977). We recalled Fiekowsky's paper after completing our own but should
probably acknowledge an unconscious debt to him for our title.
2. By the real rate of return ρ we mean the internal rate of return of the project's real cash flow, gross of taxes
and subsidies. Because we confine our attention to simple cases, this rate is always well defined.
3. Note we have assumed that the corporation will actually manage to use its investment credit and that it will
benefit from the subsequent depreciation allowances. This assumption is far from innocuous because the
actual income tax is nonlinear. This nonlinearity is obvious for the case of most individuals, but it also holds
for corporations risking low or negative taxable income. We have also assumed that depreciation allowances
are based on the historical cost of the asset gross of the investment credit. Finally, note that the cash flow to
an individual asset owner with marginal tax rate m is obtained by substituting m for u in these expressions.
4. One way of modeling the imperfect substitutability of debt and real capital is to regard the corporation as
subject to constraints on this arbitrage, e.g., the outstanding debt cannot exceed some fraction of the value of
real assets owned. In this case the relationship (1) will not generally hold for firms where the constraint binds.
Other relationships must then determine c/q. See, for example, King (1977). However, the assumption (if
only implicit) of unconstrained debt-real capital arbitrage is frequently encountered.
5. Stiglitz [1980] has emphasized that if we really did the analysis "right," π and i would be simultaneously
determined as endogenous variables.
6. To our knowledge the first published appearance of Modified Fisher's Law was in Feldstein (1976).
7. If we use Modified Fisher's Law (4b) to eliminate the current interest rate from equilibrium conditions (2)
and (3), we obtain
and
Equation (2b) is equivalent to Hall's crucial equation, except for notational differences. Hall's d and d'
correspond to our δ and δ', and his taxable proportion of capital gains, g, is set to zero. Hall's "real
after-tax interest rate," r (assumed constant) is what we have called (1 - u)i0. Equation (3b) is the basic
equilibrium condition of Jorgenson-Sullivan, with their "rate of return," r (assumed constant) equal to our
i0.
8. There is a question of consistency or existence of equilibrium if the same taxpayer were to earn different
after-tax rates of return on different assets. The concluding section of this paper touches on possible
resolutions of this problem.
Â
250
251
Page 307
9. See Bailey
(1969) for more
discussion on this
point.
10. For analyses stressing this problem, see Bradford (1980, 1981).
References
Bailey, Martin J. 1969. "Capital Prices and Income Taxation." In Arnold C. Harberger and Martin J. Bailey
(eds) The Taxation of Income from Capital Washington, D.C.: The Brookings Institution.
Bradford, David F. 1980. "The Economics of Tax Policy Toward Savings." In George von Furstenberg, ed.
The Government and Capital Formation Cambridge: Ballinger, pp. 11-71.
. 1981. "Issues in the Design of Savings and Investment Incentives." In Charles R. Hulten, ed. Depreciation,
Inflation, and the Taxation of Income from Capital. The Urban Institute Press, Washington, D.C.
Feldstein, Martin. December 1976. "Inflation, Income Taxes, and the Rate of Interest: A Theoretical
Analysis." American Economic Review 66: 809-820.
. May 1980. "Tax Rules and the Mismanagement of Monetary Policy." American Economic Review 70:
182-186.
, and Lawrence Summers. 1978. "Inflation, Tax Rules, and the Long-Term Interest Rate." Brookings Papers
on Economic Activity 1, pp. 61-109.
. December 1979. "Inflation and the Taxation of Capital Income in the Corporate Sector." National Tax
Journal 32: 445-470.
Fiekowsky, Seymour. 1977. "Pitfalls in the Computation of 'Effective Tax Rates' Paid by Corporations."
OTA Paper 23, Department of the Treasury, Washington, D.C.
Fraumeni, Barbara M., and Dale W. Jorgenson. 1980. "The Role of Capital in U.S. Economic Growth,
1948-1976." In George von Furstenberg, ed. Capital, Efficiency and Growth, Cambridge: Ballinger, pp.
9-250.
Fullerton, Don, and Roger H. Gordon. 1981. "A Re-examination of Tax Distortions in General Equilibrium
Models." Forthcoming in a National Bureau of Economic Research conference volume on tax simulations.
, Thomas King, John B. Shoven, and John Whalley. 1981. "Corporate Tax Integration in the United States: A
General Equilibrium Approach." American Economic Review, Forthcoming.
Gordon, Roger H. 1981. "Taxation of Corporate Capital Income: Tax Revenues vs. Tax Distortions." Mimeo.
Bell Labs, Murray Hill, N.J.
Gravelle, Jane G. 1981. "The Social Cost of Nonneutral Taxation: Estimates for Non-Residential Capital." In
Charles R. Hulten, ed. Depreciation, Inflation, and the Taxation of Income from Capital, Washington, D.C.:
The Urban Institute Press.
Hall, Robert E. 1981. "Tax Treatment of Depreciation, Capital Gains and Interest in an Inflationary
Economy." In Charles R. Hulten, ed. Depreciation, Inflation, and the Taxation of Income from Capital,
Washington, D.C.: The Urban Institute Press.
Â
251
252
Page 308
, and Dale W. Jorgenson. June 1967. "Tax Policy and Investment Behavior." American Economic Review 57:
391-414.
Harberger, Arnold C. 1966.

"Efficiency Effects of Taxes on
Income from Capital." In M.
Krzyzaniak, ed. Effects of
Corporation Income Tax.
Detroit: Wayne State University
Press.
Hulten, Charles R., and Frank C. Wykoff. 1981. "The Measurement of Economic Depreciation." In Charles
R. Hulten, ed. Depreciation, Inflation, and Taxation of Income from Capital, Washington, D.C.: The Urban
Institute Press.
Jorgenson, Dale W., and Martin A. Sullivan. 1981. "Inflation and Capital Recovery in the United States." In
Charles R. Hulten, ed. Depreciation, Inflation, and the Taxation of Income from Capital, Washington, D.C.:
The Urban Institute Press.
King, Mervyn A. 1977. Public Policy and the Corporation, London: Chapman and Hall.
Meade, James E. 1978. The Structure and Reform of Direct Taxation: Report of a Committee Chaired by
Professor J. E. Meade. London: Allen and Unwin for the Institute for Fiscal Studies.
Shoven, John B. December

1976. "Incidence and
Efficiency Effects of Taxes on
Income from Capital." Journal
of Political Economy 84:
1261-1283.
Stiglitz, Joseph E. 1980. "On the Almost Neutrality of Inflation: Notes on Taxation and the Welfare Costs of
Inflation." National Bureau of Economic Research Working Paper 499, Cambridge, Mass.
Â
Page 309
III
INCOME AND CONSUMPTION TAXATION: TRANSITION, COMPLEXITY,
AND IMPLEMENTATION
Â
252
253
Page 311
11
Transition to and Tax-Rate Flexibility in a Cash-Flow-Type Tax
1
Introduction
This paper concerns a class of problems of implementation of and transition to consumption-type taxes. The
specific issue that motivated this paper is transition to what I have called (Bradford, 1986) a two-tiered cash
flow tax. The term refers to a two-component system. A single-rate business-level tax applies to the real cash
flow of business firms (so investment is expensed), net of payments to workers. A personal-level
graduated-rate tax applies to the workers' compensation. A good example is the flat tax, pioneered by Robert
Hall and Alvin Rabushka (1983, 1995), which first brought this type of tax to my attention. In the flat-tax
system, graduation in the compensation tax takes the particularly simple form of a tax-free allowance, based
on family composition, together with application of the same rate paid by businesses to all amounts in excess
of the exempt amount. 1
Discussions of actual flat-tax proposals typically take for granted that a new system would replace the old as
of some transition date. It seems much more likely, however, that if such a major change were to be
undertaken a new system would be phased in over time. At the price of postponing the full achievement of
whatever might be seen as the policy gains from the shift, the myriad transition incidence effects might
thereby be adequately muted.2 Phase-ins of tax changes are notorious sources of complexity and
opportunities for political machinations. I have argued in the past, however, that there is a way of phasing in a
two-tiered cash-flow tax that would minimize these problems (Bradford, 1986, pp. 329-334). The simplicity
of two-tiered cash-flow tax systems makes the apparently unthinkable quite rea-
Â
Page 312
sonable, namely, running totally separate parallel tax systems for a period of years. 3 The flat tax, for
example, requires little, if any, information not required for the present-day income tax on individuals
(including the tax on proprietors and partners) and corporations. It could easily be incorporated as an
additional schedule on the existing individual and corporate tax returns. One possibility is to calculate tax
under both systems. For an initial period, say two years, pay 80 percent of the bottom line of the present-day
income tax and 20 percent of the bottom line of the new tax. During a second period, say the next two years,
pay 60 percent of the bottom line of the present-day income tax and 40 percent of the bottom line of the new
tax. At the end of this process, the tax is based 100 percent on the new system and the old system can be
discarded.4
Whatever the virtues of this adjustment process, it suffers from at least one clear disadvantage in the form of
more or less (depending on the speed of the phase-in and the durability of the investment in question) severe
disincentives for new investment. By the same token, disinvestment would be encouraged. New investment is
immediately deducted in the calculation of business income under the new system. If the rate of tax were
constant over time, by a familiar argument, the expensed deduction would just offset the taxation of
subsequent returns for an investment that would barely break even in the absence of tax. The tax would thus
be neutral with respect to the investment decision. But since the tax rate at which the deduction is made
during the phase-in period is lower than the rate applied subsequently when the investment's payoff comes in,
253
254
a break even investment becomes a loser.5
It is readily seen that the investment-discouraging effect of a rising rate of tax during the phase-in to a cash
flow business income base would characterize an increase in the rate for any other reason, as well. It is likely
that policymakers will insist on having the option to change the rate of tax. They may, however, want to
avoid the windfall gain and loss aspects of such changes. Taxpayers' anticipations of such changes will
furthermore generate potentially large incentive effects. The problem that I describe in connection with a
phase-in to such a tax therefore generalizes to a potentially serious problem of implementing a consumption
tax on an ongoing basis. This paper is devoted to a discussion of methods of dealing with this problem.
Â
Page 313
In an appendix, I provide a formal model of the propositions developed verbally, but more generally, in the
body of the paper. Throughout, I confine my attention to the partial equilibrium assumption that interest rates,
and their generalization in a world with risk, are not affected by changes I consider. I believe that a more
sophisticated modeling taking into account general equilibrium repercussions of changes in tax policy would
not change the conclusions in any major way.
2
Tax-Rate Variation in a Two-Tiered Cash-Flow Tax
As described, for example, in Bradford (1996), a useful way to think about a two-tiered cash-flow tax is to
start with a subtraction-style value-added tax. This is simply a tax at a single rate levied on all businesses, of
whatever legal form. The base of the tax is the non-financial cash flow of the firm, that is, the difference
between receipts from sales of goods and services of all kinds and purchases from other business firms. A
two-tiered cash-flow tax simply modifies such a subtraction-style value-added tax by permitting firms to
deduct as well their payments to workers (leaving the business tax). The payments to workers are, in turn,
subject to the compensation tax.
The intertemporal rate variation problem is starkly manifested by the case of introduction of a
subtraction-style value-added tax (the same would hold for a European-style invoice and credit value-added
tax). I use a canonical example to convey the nature of the problem: a retail store owner who buys a stock of
canned tomato juice for $10,000 the day before the tax goes into effect, with a rate, say, of 20 percent. If the
tomato juice is sold the day after the introduction of the tax, for roughly $10,000 (I assume a hotly
competitive retail sector), the owner of the inventory will get to keep only $8,000 after tax. This is because
the cost of the goods sold is not allowed as a deduction. Rather, a deduction is allowed only for current
purchases by the business.
The short-term inventory example, for which the element of waiting is negligible in the business calculation,
gives a vivid instance of the impact of a rate change. The effect is the same for assets other than inventory,
however. A person who bought a building the day before the introduction of the tax will suffer exactly the
same loss, amounting to the new tax rate times the amount paid for the build-
Â
254
255
Page 314
ing, even though the cash payoff from holding the building may be years in the future. If that person were to
sell the building for its current market value (presumably the same as the day before) the proceeds would be
subject to the value-added tax. That is the justification for the commonly held view that imposition of a
consumption tax imposes a one-time levy on "old wealth."
A typical statement of this position is my own: "In general, imposition of a consumption-type tax will cause a
one-time loss to owners of certain assets." 6 I would note in passing that the words, "owners of certain
assets," were carefully chosen. The loser in the tomato juice example is the owner of the inventory, who may
not be wealthy at all; the inventory might have been financed by the issue of debt. That is why I went on to
elaborate, "The loss will be spread over all wealth-owners to the extent the transition is accompanied by an
unanticipated increase in the price level.'' An unanticipated general price level change, which is not a
necessary concomitant of the introduction of a value-added tax, would have the usual effect of penalizing net
nominal creditors. A price level change (unanticipated) in exactly the amount of the tax, often taken for
granted by commentators, would effectively spread the transition tax over wealthholders in general. (So the
illustrative holder of debt-financed inventory would not suffer any transition tax burden.)
Although my words were carefully chosen, they were, perhaps, not sufficiently so. As has been pointed out to
me by Daniel Shaviro, I could equally well have written, "In general, imposition of an income-type tax using
a present-value depreciation scheme will cause a one-time loss to owners of certain assets." An example of a
present-value depreciation scheme is the proposal by Auerbach and Jorgen-son (1980) to substitute an
up-front deduction at the time of acquisition of an asset equal to the discounted value of future depreciation
allowances, in place of the normal year-by-year depreciation deductions. Upon introduction of such a system
(or upon transition to it from the usual income accounting) there would be a one-time loss to owners of
business assets, exactly like the tax on the inventory-holder in the example of a subtraction-style value-added
tax. Similarly, changes in the rate of tax in a system with Auerbach-Jorgenson present-value depreciation
would result in losses (for a rate increase) or gains (for a rate cut) to owners of existing assets. The transition
effects are not due to the economic character of the tax but to the method of its implementation.
Â
Page 315
2.1
Insulating Against Tax-Rate Changes: Grandfathering
It seems evident in the case of a switch in methods of accounting for depreciation in an income tax that the
way to eliminate the transition-incidence effect would be to do something like "grandfather" the depreciation
allowances of assets acquired before the new system was introduced. The same holds for the
consumption-type tax. In the example of the tomato juice inventory, grandfathering means looking back to
the purchase of the asset in the pre-transition period and allowing a current deduction in the new system.
Generalized to the broader system of income accounting, grandfathering calls for allowing the owner a
deduction for the current market value of business assets. In a well-constructed system of income accounting
(including adjustment for inflation) this would be called in income-tax jargon the "basis" (acquisition value
less cumulative depreciation allowances, for example) of business assets.
As I have emphasized, changes in the rate of a subtraction-type value-added tax (or the business tax rate in a
two-tiered cash-flow tax) generate the same incidence and incentive effects as the transition from the no-tax
situation. Eliminating these transition effects can be accomplished by the same grandfathering method as just
described. Upon a change in the rate, the owner would be allowed a credit in the amount of the product of the
basis of business assets and the increase in the rate of tax. So, for example, an increase in the rate of tax from
255
256
20 to 50 percent would give rise to a credit of 5 percent of basis. 7
To serve the function of eliminating transition incentives due to tax-rate changes, "economic" depreciationthe
change in an asset's market value during the periodis required. That is, there will be transition incentives
unless the basis of assets equals their market values. To implement the required grandfathering policy
perfectly would require true, inflation-corrected depreciation adjustments. Indeed, the accounting really
required is mark-to-market valuation. Suppose, for example, the illustrative tomato juice inventory is
purchased on day -2. On day -1, a disastrous frost wipes out the tomato-juice crop, so the canned stock jumps
in market value from $10,000 to $20,000. On day 0, the new tax goes into effect, and on day +1, inventory is
sold for $20,000. To make the juice-holder whole requires a credit of 20 percent of $20,000, not 20 percent of
$10,000 as historical-cost accounting would suggest.8
Â
Page 316
2.2
Insulating Against Tax-Rate Changes: Depreciation Plus Interest on Basis
The requirement of good economic income accounting is clearly an obstacle to practical implementation of
insulating the system from rate-change effects, on which I have a bit more to say below. Another obstacle
seen by many commentators who have considered a switch from an income- to a consumption-based tax
system is the huge onetime revenue cost due to the write-off of the entire depreciation basis at the time of
transition. Viewed in terms of the long-term budget constraint, this revenue cost is simply the requirement of
avoiding the one-time extra tax on those affected by the transition over what might, arguably, be considered
the intent of introducing the new tax. 9 Because, however, the revenue cost occurs in a lump at the time of
introducing the consumption-based tax, it is generally seen as posing a large fiscal challenge.
The solution to this problem is simple enough: The tax allowance could be given in the form of government
IOUs. That is, instead of allowing an immediate write-off of basis at the time of transition, the transition rule
could provide a path of tax rebates or other transfers over time with the same discounted value.10 This would
be the equivalent of allowing the write-off and issuing debt, but would show up differently in the fiscal
bookkeeping.11
A particular form of this alternative policy turns out to have a very convenient property of eliminating the
need for specific adjustments to take account of changes in the rate of tax. Under the alternative policy, the
taxpayer is allowed a deduction for the decline over time in the value of business assets12 plus a deduction for
the cost of carrying the capital reflected in the value of those assets. Practically speaking, this implies using
income accounting for business income with the addition of a capital cost allowance equal to the going rate of
interest times the tax basis in the business (including basis in inventory). (Since the objective is a real-income
measure, the depreciation and similar allowances would be adjusted for inflation. Similarly, the interest rate
applied to basis would be a real, inflation-adjusted measure.13
If the rate of tax is constant over the life of the investment, the suggested policy is evidently equivalent to the
expensing characteristically associated with consumption taxation. The difference is that the tax benefit
ordinarily obtained due to expensing is received over
Â
256
257
Page 317
time, with interest. "Basis" is simply the part of the value of assets that has not yet been taken as a deduction
that "should" have been allowed, so the taxpayer is compensated with an interest allowance.
With a constant rate of tax, all schemes that provide an interest compensation for delayed receipt of benefits
will be equivalent from the point of view of the taxpayer. It may be less clear, but is true, that the specific
form, deduction of economic depreciation plus the interest rate times the basis in business assets, is unique in
incorporating just the right adjustment to achieve neutrality in the face of inter-temporal variation in the rate
of tax. This point is perhaps best established by considering the alternative and most easily understood in the
context of introduction of a new tax. Suppose the basis, reflecting the amount that has not yet been allowed as
a deduction (from the zero-rate tax), is different from the market value of the asset. That is, suppose
something other than economic depreciation has been used in determining the amount subject to the new
depreciation cum carrying-cost deduction. To be specific, suppose depreciation is accelerated, relative to
economic depreciation, so that the basis is below market value. Now a tax is imposed at a positive rate, which
will stay constant in-the future. The taxpayer will obtain the equivalent, in present-value terms, of the
deduction of basis at the transition date. But we know that deduction of the market value of the asset is
required to eliminate a loss in asset value at the transition date, and therefore to eliminate the incentive to
disinvest prior to the transition date.
By a similar argument, if depreciation allowances for the pre-transition period are below economic
depreciation, the basis in the asset will exceed the market value at the transition point. Then the allowance at
the transition date, economically equivalent to write-off of basis at the new tax rate, will be too high, in the
sense that the taxpayer will experience a jump in asset value and an extra incentive to invest prior to an
anticipated transition. Economic depreciation, coupled with deduction of interest on basis, is "just right" in a
world of changing tax rates.
3
Commentary
3.1
Measuring Depreciation
A very big drawback of both approaches is their requirement for well-measured, inflation-corrected
depreciation allowances. Currently, in
Â
Page 318
the United States there is no correcting for inflation in either tax or financial accounts, in spite of arguably
significant mismeasurement as a result. One of the most serious obstacles to inflation adjustment is the
appropriate treatment of interest. Fortunately for the feasibility of the schemes described in this paper,
interest payments and receipts are not included in the tax base. Adjustment of the purely real side of the
accounts is likely to be more manageable. 14
Even inflation-corrected historical-cost accounts may still produce rather poor approximations to current
market values. In Bradford (1991) I presented data on the ratio of the aggregate market value of equity in
U.S. corporations to the net worth of the U.S. nonfinancial corporate business sector to the consolidated
financial accounting measure of net worth (incorporating Commerce Department data on the corresponding
inflation-adjusted capital stocks). Over the period 1948 to 1987 the ratio varied widely, with a high (in 1968)
of 110 percent and a low (in 1974) of 37 percent. Although one can reasonably quibble with the details of this
257
258
calculation, it does suggest that historical-cost accounting is unlikely to be a particularly accurate measure of
current value.
One important source of divergence between any measure of the tangible assets of a business and the market
value of that business are the intangible assets with which they are combined. In the "information age" it
seems likely that intangible assets, such as trade marks, marketing skills, software copyrights, and so on,
account for an increasing fraction of wealth. Consider, for example, an investment in a marketing campaign
to promote the popularity of a brand of running shoes. Under present tax (and financial) accounting
procedures, the outlays on the campaign would be deducted currently, even though they are likely to produce
sufficient extra future profits to justify the expense. In tax jargon, the basis in such an intangible asset is zero.
Presumably, however, the anticipation that a tax would be introduced between the time of making the outlay
and the realization of the payback would have the effect of discouraging the investment. Short of marking the
value of the firm to market, there does not seem to be any practical way to avoid such mismeasurements.
As in the income tax, there is a remedy available to the taxpayer with basis different from market value:
realize any gain or loss by selling the asset. (I neglect here niceties of the limits placed by actual income-tax
rules on such transactions.) In a world without transac-
Â
Page 319
tion costs, this option (since it is an option) can only work to the taxpayer's advantage. Such realization
transactions have been sufficiently attractive in the past in the case of residential real estate (where a
differential between capital gain and ordinary income tax rates was involved) to generate significant activity.
It can probably be taken for granted, however, that the tax stakes are unlikely to motivate significant asset
trading in the case of the transition situations envisioned in this analysis.
The key requirement is to measure correctly the cumulative total in real terms (hence the importance of
inflation correction). If the cumulative total of allowances adds up the cost of the asset, the problems are due
to timing effects, which are important in an income tax but not in a constant-rate cash-flow tax. In the context
of a cash-flow company tax, timing matters only when there is intertemporal rate variation. For tax-rate
changes of the sort one is likely to anticipate (compare, say, the rate changes enacted in typical value-added
tax systems), the distorting effects of mismeasurement should be minor. As is touched upon below, in the
case of the transition effects upon introduction of a two-tiered cash-flow tax, there would, in addition, be
offsetting incentives from the residual income tax (which is on a realization basis).
3.2
Special Aspects of the Transition from an Income Tax
The phase-in transition to a cash-flow tax discussed in this paper would be accompanied by a corresponding
phase-out transition from the income tax. If the existing tax system were well modeled by an accrual income
tax, the latter process should be free of the transition incidence and incentive effects examined here. This is
because neutrality with respect to intertemporal variation in the tax rate is a (not very well-known) property
of an ideal income tax. "Ideal" here means a mark-to-market accrual tax, which perforce implies economic
depreciation. 15
The actual income tax differs, however, from the accrual ideal. Separate taxes at the individual and corporate
level and the intricate rules relating to the financial structure of corporations greatly complicate the problem
of sorting out transition effects. This is why the gradual process of transition might be attractive. The central
phenomenon considered in this paper is the incentive to alter investment owing to the interaction with a
changing rate of tax and the recovery
258
259
Â
Page 320
of basis in business investments. The main "defect" of the income-measurement system in this regard is
accelerated depreciation of depreciable assets and complete write-off of investment in "self-constructed"
intangible property. Ordinarily, a declining rate of income tax works to increase the attractiveness of such
investments (and, in effect, to forgive taxes that would have been expected on the return flow from past
breakeven propositions). It is a convenient property of the simultaneous phasing out of one tax and phasing in
of the other that this class of distortions under the imperfect income tax would be roughly neutralized.
3.3
Concluding Comments
Three factors favor the depreciation-with-interest method over the grandfathering method of dealing with
transition (and with the ongoing problem of rate changes). First, the former method would be effective in
neutralizing investment incentive effects in a situation in which the marginal rate of tax varies over time for
an individual taxpayer. Grandfathering requires keeping track of the time path of marginal rates, a possible
complication. To be sure, the theoretical two-tiered cash-flow tax has a single business tax rate applicable to
all taxpayers at any time. In a real-world application, however, there might well be limits on loss offsets and
similar features that would render the tax schedule nonlinear.
Second, grandfathering carries with it a large revenue cash-flow shortfall at the time of introduction of a
two-tiered cash-flow tax, with similar lumpy effects at the time of any rate change. As discussed above, these
effects could be offset by granting a rebate in the form of a smoothed stream of payments with the same
present value. This would, however, add a nontrivial complication to the operation of the system.
Third, because depreciation-with-interest-on-basis

results in neutrality toward business investment
under a constant tax rate, regardless of whether
the depreciation allowances match the economic
levels, this method takes some of the pressure off
of this difficult administrative problem. 16 Getting
the timing of allowances wrong affects only the
transition incentives. If the rate changes are small
in the cash-flow tax (typical rate changes in
European value-added taxes might serve as a basis
for comparison), transition incentives
Â
Page 321
would also be small unless the timing deviation were very large. Adjusting allowances for inflation would
remain critical, but since the appropriate adjustments are conceptually fairly simple and would be the same
for all assets, this may be less of a problem.
The main disadvantage of the depreciation-with-interest method is its requirement to identify the appropriate
real discount rate. Conceptually, this choice is clear enough in the world without transaction costs that
underlie the investment model used in deriving the scheme. In that world, the investor always has the option
of financing with risk-free debt. Everyone is indifferent between $1 now and $(1 + i) a year from now, where
259
260
i is the risk-free interest rate. Since the government guarantees the interest on postponed depreciation
deductions, the appropriate interest rate to use is the short-term government borrowing rate
(inflation-indexed).
To determine how this conclusion would be modified in the light of more realistic assumptions about the
financial environment would merit exploration. Transaction costs could imply that an investment that would
be attractive in the absence of taxes, or with immediate expensing in a cash-flow tax framework, could not be
financed under the depreciation-with-interest method. On the other hand, the value of the future tax
deductions would seem to be comparable as collateral to the asset itself. More difficult to sort out is the
influence of uncertainty about the reliability of the government's commitment not to change the policy. 17
(Since the policy could become more favorable, rather than less, the influence of this form of risk on the
investment decision is, perhaps, ambiguous.)
It is appropriate to conclude with this hint of the political process. This chapter concerns the possibility of
carrying out a smooth transition from the existing, messy income tax to a two-tiered cash-flow tax. The main
reasons one might want to undertake this transition are the superior simplicity and neutrality properties of the
new system. The warning may bear repeating, therefore, that these attractive properties, which result from the
ability to exclude financial instruments from the tax calculation and from the relatively simple business
accounting, would not necessarily be proof against a complex political process.18 Adding, as the suggested
transition scheme would, a requirement for inflation-corrected depreciation and an allowance of interest on
undepreciated basis solves some problems but introduces others of a political nature, since someone has to
specify the depreciation rules and interest rate.
Â
Page 322
Appendix: Formal Modeling
In the body of this chapter I have used verbal, although I believe general and rigorous, arguments. In this
appendix I present a mathematical formulation of the results, using the exponential-decay model of capital
familiar to economists since at least the famous Hall and Jorgenson paper (1967). A typical productive asset
("machines") is assumed, in effect, to shrink over time to successively smaller replications of itself. A unit
machine of durability δ is transformed by the passage of time into c-δs units of the same machine, where s is
the time since putting the machine in service. The set of available technologies is assumed to be represented
by machines of durabilities 0 δℜ∞. The value of the output, net of other input costs, of machines of each
durability is subject to diminishing returns relative to the cost of the machine (which might be the
consequence of a diminishing price at which the particular good produced can be sold) and is determined in
equilibrium.
The exponential-decay, or proportional-depreciation, model is commonly used for such analyses primarily
because it can be manipulated easily. It should be kept in mind, however, that there is no particular reason to
believe that it fits the facts of investment opportunities. As discussed in the body of the paper, however, the
conclusions reached here with regard to the exponential-decay model carry over to a consistent
mark-to-market income measurement system. 19
As is also typical, I assume a one-good world, in the sense that a unit machine is assumed costlessly
convertible into a unit of the single good. It follows that the value of one unit of any machine depreciates at a
rate δe-δs at age s.
A.1
Rate Changes
Unimportant under
260
261
an Income Tax
A.1.1
The Analysis without Taxes
Suppose a unit of asset of type δ throws off an annual gross rental (gross of any depreciation in the machine
but net of payments to any collaborating factors) of c(δ). Then, in the absence of taxes and risk, and with the
option to borrow and lend at interest rate i, a capitalist will value a unit asset of type δ at
Note that this is an arbitrage

argument. The arbitrage is with
the alternative opportunity by
which the interest rate is defined,
which I have here represented as
interest-bearing debt. If the asset
could be purchased for less than
this demand price, by borrowing at
the going interest rate, a capitalist
could purchase a machine and
arrange for a net cash flow that is
positive over some time interval
and never negative. Likewise, if
the going price of the asset is less
than the demand price, the owner
of a machine could produce a
strictly positive net cash flow over
some interval by a combination of
selling the machine and lending
the proceeds.
Â
Page 323
Since a unit machine costs 1, then for demand and supply for machines to be equated it must be true for each
durability used,
The before-tax or social rate of return, r(δ), is defined implicitly by the internal rate of return on investment
in a machine of the given durability:
That is,
So with no taxes, the social rate of return on all machines of all durabilities in positive use will be equal in
asset market equilibrium, and equal to the interest rate,
261
262
If the (instantaneous) interest rate is a known function of time, i(t), instead of a constant, these expressions
and ideas need to be generalized. Then, the equilibrium net rental thrown off by a unit machine will also
typically be a function of time, which I write as c(t) (changing the argument from δ), and the demand price
for an asset of durability δ is
In general, the demand price of an asset now depends on the anticipated time path of future net rentals and
interest rates. If, however, it is anticipated that the net rental rates and interest rates will be related by
then the demand price will equal
This integral is of the general form
In the present case -e-f(ℜ∞) = 0 and -e-f(0) = 1; the demand price equals the supply price, validating the fact that
c(t) = i(t) + δ will equate demand and supply for machines. In this context we need also to substitute the
instantaneous social or before-tax yield,
for the yield-to-maturity formulation that is appropriate for the constant-stationary-state story (where I have
added the durability arguments to the expression for gross rental to emphasize that at any given time there
will be many margins of return to investment).
Â
Page 324
A.1.2
Adding Income Taxes
As background for the analysis of consumption taxes, it is useful to review the equilibrium conditions when
an income tax at rate m applies to gross rent, with an allowance for actual (''economic") depreciation and a
deduction for interest (which is, in turn, taxed to the recipient). By the same arbitrage argument as used
before (trading off the after-tax consequences of buying a machine with the after-tax consequences of lending
at taxable interest), in the stationary state, with constant interest and gross rental rates, the demand price for a
unit asset of durability δ will be
Equilibrium in asset markets requires that this demand price be equated to the supply price (which is 1),
262
263
which implies further
for all δ. [Reminder: In equilibrium, c = c(δ).] The implication that, provided a deduction for economic
depreciation is allowed, the choice among assets according to durability is undistorted by an income tax is
called by Sinn (1987) the Johansson-Samuelson theorem.
It is important to be clear about what it is that is not distorted. In the equilibrium with the income tax, the
social rate of return, r, is equated on all assets. This follows immediately from the observation that
equilibrium requires that r(δ)= i for machines of all durabilities employed in positive amount. Given the
interest rate i, the tax rate m has no influence on the level of investment in machines of any durability. (As
has been mentioned, this result is general, and not dependent on the exponential depreciation structure,
although defining economic depreciation is not so easy once one leaves the simple model. In particular, if
investment is irreversible, the time path of interest rates may matter. Risk is also likely to play a more
important role in the analysis. So long, however, as true market value is used as the basis for the depreciation
allowances, the propositions discussed in this appendix will hold.) There is, however, a distortion in the
model due to the deviation between the after-tax yield on savings, (1 - m)i, and the common before-tax yield
on investment, r. A change in the income tax rate will, in general, influence investment levels via its general
equilibrium impact on the desired stocks of wealth and, hence, the interest rate. More important, there will, in
equilibrium, be a difference between savers' marginal rate of time preference and the social rate of return on
investment.
A convenient property of an income tax is that these neutrality results carry over to the case of time-varying
tax rates. Given the path of interest rates, anticipated changes in the rate of a true income tax (with economic
depreciation) have no effect on the current level of investment. (In general equilibrium one would expect
tax-rate changes to produce changes in the
Â
Page 325
path of the interest rate, which would have an impact on asset prices if investment is not reversible.)
To see how this conclusion emerges from the formal model, we examine the breakeven requirement for the
case of a time-varying tax rate (for simplicity, keeping the assumption of a constant interest rate). The
demand price for a unit asset of durability δ will then be
As in the no-tax case with varying interest rate, the current demand price depends upon current and future
gross rental and net tax rates. We can check that the condition i = c(s) - δ will be an equilibrium relationship;
in this case, the breakeven condition would imply
As before, the integral is of the general form
263
264
and, as before, -e-f(ℜ∞) = 0 and

-e-f(0) = 1; the demand price
equals the supply price,
validating the fact that c(t) =
i(t) + δ will equate demand
and supply for machines.
A.2
Rate Changes May Matter under a Business Level Cash-Flow Tax
With a proportional cash-flow tax on business investment at rate m, constant through time, the net cash flow
of a capitalist who purchases a machine is simply a proportion 1 - m of what it would be in the absence of the
tax. The net-of-tax gross rental from a unit machine ("gross" here refers to the treatment of depreciation) is
reduced by the fraction m, but so is the cost to the capitalist of buying the asset. By the same arbitrage (with
debt) arguments as we used in the no-tax environment, the capitalist's demand price for an asset is given by
but because the net-of-tax cost to the capitalist of acquiring a machine is 1 - m instead of 1, equilibrium still
implies
So, given the interest rate, the tax rate has no effect on investment. [Sinn (1987) calls this result the Brown
theorem, referring to E. Cary Brown (1948).] Here the neutrality carries over to the saving decision as well as
the durability of machines in equilibrium. That is, in equilibrium, the intertemporal marginal rate of
substitution of consumers equals the social rate of return.
Â
Page 326
Intertemporal variation in the

interest rate affects this argument
in exactly the same way as it
affected the story in the case
without taxes. Intertemporal
variation in the tax rate is,
however, another matter. The tax
rate determines the fraction of a
newly purchased machine that is
financed by tax savings due to
the expensing of the purchase. If
that fraction is not matched by
the fraction of the future cash
flow that is taken away by the
tax collector, an investment that
breaks even in the absence of
taxes will no longer break even
with taxes. If the future tax rate
is higher than the present rate, a
breakeven investment becomes a
loser and it becomes a winner if
264
265
the future tax rate is lower than

the present rate.
To take a simple example, suppose the business tax rate is rising over time in such a way that the fraction of
the payoff of an investment retained by the owner declines at some multiple of the rate of interest:
Then the breakeven condition for an investment would be
which reduces to
The effect of the rising business tax rate is equivalent to an increase in the interest rate. Its impact is identical
to that of a uniform income tax.
Introduction of a cash-flow tax, from no tax, corresponds to intertemporal variation in the business tax rate
from zero to some positive amount. If the introduction is unanticipated, owners of machines in effect give up
a fraction m of their assets. If the tax-rate change is anticipated, there are incentive effects on investment.
A.3
Neutralizing Transition Effects in a Business Cash-Flow Tax Regime
We can use the exponential decay case to model the two methods of neutralizing the transition effects
described in the body of the paper:
1. Investment expensed and tax treatment "grandfathered."
2. Only economic depreciation allowed, instead of expensing, but with an additional deduction equal to the
rate of interest times the undepreciated basis of business assets.
In both cases, there is assumed to be no taxation of interest received or deduction of interest paid.
A.3.1
Method (1): Grandfather Tax Treatment
The idea is to compensate the investor for an increase in the tax rate, or to extract a payment for a cut in the
tax rate. In a discrete model, to grandfather the tax treatment of past investment if the tax rate goes up from
m(t - h) to
Â
Page 327
m(t), the owner of the asset is provided a tax rebate of m(t) - m(t - h) times the "unused basis" in the asset.
The latter will (in equilibrium) be the amount of the asset remaining after deducting economic depreciation
since acquisition. [This is speaking loosely. In this case, the basis is actually zero, since the asset is expensed
in method (1).]
In the instantaneous version of this method, with tax rate as a function of time, m(t), there is a rate of rebate
equal to m'(t) times the undepreciated basis. If the rental rate c is constant, the basic economics of the
investment is described by
265
266
rent after tax: {[1 - m(s)]c + m'(s)}e-δs;
machine costs: 1 - m(0).
Eliminating the opportunity for arbitrage profit requires
We know that with a constant tax rate the breakeven rental rate is given by
We need to check that this will still present a breakeven investment opportunity when the tax rate is varying
over time.
Integrating the left-hand side by parts,
Q.E.D.
A.3.2
Method (2): Economic Depreciation Plus Interest on Basis
In method (2) the investor pays income tax, using economic depreciation rather than expensing of new
investment, but there is allowed in addition a deduction for the cost of carrying the capital, in the form of the
rate of interest times the remaining basis. If the rental cost c is constant, the cash flows are described by
rent after tax: {[1

- m(s)]c + m(s)(i
+ δ)}e-δs;
machine costs: 1.
Eliminating the opportunity for arbitrage profit requires
As before, use the fact that with a constant tax rate the breakeven rental rate is given by
Â
266
267
Page 328
If this continues to hold, we can substitute into the equilibrium condition, which becomes
which we know to hold.
Notes
The research reported here has benefited from the financial support of the Woodrow Wilson School at
Princeton and the John M. Olin Foundation and from the stimulating environment at the Center for Economic
Studies of the University of Munich. Thanks also to Daniel Shaviro and Jane Gravelle, and to James Poterba,
Michael Boskin, and other participants in the NBER teleconference on Asset Price and Transition Effects of
Consumption Tax Reform, January 27, 1997.
1. Other examples are my own loosely specified X tax (Bradford, 1982) and Charles McLure and George
Zodrow's simplified alternative tax, described in McLure, Mutti, Thuronyi, and Zodrow (1990) and in
McLure and Zodrow (1991).
2. For a careful effort to model the trade-offs in alternative methods of introducing a reformed tax, see
Zodrow (1981, 1985). Louis Kaplow (1986) and Daniel Shaviro (1997) develop the (sometimes unexpected)
pros and cons of providing protection against statutory change.
3. Lyon (1992) notes the

ubiquity of parallel systems. A
good, although perhaps not very
happy, example is the
alternative minimum tax for
businesses and high-income
individual taxpayers. Michael
Graetz (1983) even argued that
the minimum tax might serve
the function of a bridge to a
reformed income tax, along the
lines of the process envisioned
here.
4. The same transition plan could be effected by appropriate adjustments of all the rate and credit parameters
of the old and new systems. The plan description has, however, the advantage of transparency.
5. An early discussion of the transition incentive problem is by Hall (1971). Hall has returned to the subject
in (1996) and (1997). The problem is also discussed in Sinn (1987) and Howitt and Sinn (1989).
6. Other examples include Gravelle (1995); Altig, Auerbach, Kotlikoff, Smetters, and Walliser (1997) show,
in simulations of transitions to VAT, flat, and X taxes, how important the issue may be.
7. English and Poddar (1995) have developed a scheme along these lines to deal with rate changes in a
value-added tax that includes financial institutions.
8. An instance of a somewhat
similar transition phenomenon
was the impact of lower rates of
tax, enacted in the Tax Reform
267
268
Act of 1986, on the return flow

from assets for which
accelerated depreciation had
been taken in earlier years. In
that case, a windfall gain was
involved. Zodrow (1988)
presents an analysis of the effort
to neutralize this change.
Â
Page 329
9. The idea that consumption out

of past accumulation "ought" not
to be taxed is perhaps most
plausible in connection with a
shift from an income tax, under
which the past accumulation has
been taxed already. To be sure,
some argue that an extra tax
based on past accumulation is a
desirable policy objective (see,
for example, Kotlikoff, 1992).
10. Boadway and Bruce (1984) exploit this equivalence in their suggested design of a neutral business tax.
11. Kotlikoff (1992) has emphasized the arbitrary nature of fiscal accounting as a system of "labels" on cash
flows.
12. I apply the term "economic depreciation" to this amount, even though the adjustment may involve
something more than is suggested by this accounting terminology. In particular, the decline could be
negative.
13. Essentially this method has been implemented in Croatia, as described by Rose and Wiswesser (1997),
who played a role in designing the new system. An important intellectual predecessor was Wenger (1983).
14. For a discussion of the issues, see Shoven and Bulow (1975).
15. The proposition that, provided a deduction for economic depreciation is allowed, the choice among assets
according to durability is undistorted by a flat-rate income tax is called by Sinn (1987) the
Johansson-Samuelson theorem, referring to Johansson (1969) and Samuelson (1964). Although Johansson
(1969, p. 110) seems to have anticipated the fact that these neutrality results carry over to the case of
time-varying tax rates, Samuelson (1964) did not. It was spelled out by Sandmo (1979) and Lyon (1990).
16. This is the point of Boadway and Bruce (1984).
17. For a study that demonstrates both the reality of risk of changes in policy and the potential effect on
investment incentives, see Auerbach and Hines (1987).
18. For an extended discussion of the political forces at work in alternative tax regimes see Paul (1997) and
my commentary (Bradford, 1997).
19. For proper income measurement, both the interest rate and any depreciation or similar adjustments should
be on an inflation-corrected basis.
268
269
References
Altig, David, Alan Auerbach, Laurence Kotlikoff, Kent Smetters, and Jan Walliser (1997). "Simulating U.S.
Tax Reform." Washington, DC: U.S. Congress, Congressional Budget Office. September.
Auerbach, Alan J., and James Hines (1987). "Anticipated Tax Changes and the Timing of Investment." In
The Effects of Taxation on Capital Accumulation, Martin Feldstein (ed.). Chicago: University of Chicago
Press.
, and Dale Jorgenson (1980). "Inflation-Proof Depreciation of Assets." Harvard Business Review,
September/October.
Boadway, Robin, and Neil Bruce (1984). "A General Proposition on the Design of a Neutral Business Tax."
Journal of Public Economics, XXIV: 231-239.
Â
Page 330
Bradford, David F. (1982). "The Choice between Income and Consumption Taxes." Tax Notes, XVI (8,
August 23): 715-723. Revised version in New Directions in Federal Tax Policy for the 1980s, Charls E.
Walker and Mark A. Bloomfield (eds.). Cambridge, MA: Ballinger.
(1986). Untangling the Income Tax, Cambridge, MA: Harvard University Press.
(1991). "Market Value versus Financial Accounting Measures of National Saving." In National Saving and
Economic Performance, B. Douglas Bernheim and John B. Shoven (eds.). Chicago University of Chicago
Press.
(1996) "Consumption Taxes: Some Fundamental Transition Issues." In Frontiers of Tax Reform, Michael J.
Boskin (ed.). Stanford, CA: Hoover Institution Press.
(1997). "What's in a Name? Income, Consumption, and the Sources of Tax Complexity." Forthcoming in
University of North Carolina Law Review.
Brown, E. Cary (1948). "Business Income Taxation and Investment Incentives." In Income, Employment and
Public Policy Essays in Honor of A H Hanson, L. A. Metzler, E. D. Domar, et al. (eds.). New York: W. W.
Norton.
English, Morley D, and Satya Poddar (1995). "Taxation of Financial Services under a VAT: Applying the
Cash-Flow Approach." October. Manuscript.
Graetz, Michael J. (1983). "The 1982 Minimum Tax Amendments as a First Step in the Transition to a 'Flat
Rate' Tax." Southern California Law Review LVI (January): 527-571.
Gravelle, Jane G. (1995). "The Flat Tax and Other Proposals: Who Will Bear the Tax Burden?" Washington,
DC: U.S. Congress, Congressional Research Service. November 29. Report 95-1141 E.
Hall, Robert E. (1971). "The Dynamic Effects of Fiscal Policy in an Economy with Foresight." Review of
Economic Studies XXXVIII (April 1971): 229-244.
(1996). "The Effects of Tax Reform on Prices and Asset Values." In Tax Policy and the Economy (X), James
Poterba (ed.). National Bureau of Economic Research, MIT Press.
269
270
(1997). "Potential Disruption from the Move to a Consumption Tax." American Economic Review, Papers
and Proceedings, LXXXVII (May): 147-150.
, and Dale W. Jorgenson (1967). "Tax Policy and Investment Behavior." American Economic Review LVII
(3, June): 391-414.
, and Alvin Rabushka (1983). Low Tax, Simple Tax, Flat Tax. New York: McGraw-Hill.
, and (1995). The Flat Tax, 2nd ed. Stanford, CA: Hoover Institution Press.
Howitt, Peter, and Hans-Werner Sinn (1989). Gradual Reforms of Capital Income Taxation. American
Economic Review LXXIX: 106-124.
Johansson, Sven-Erik (1969). "Income Taxes and Investment Decisions." Swedish Journal of Economics
LXXI: 104-110.
Kaplow, Louis (1986). "An Economic Analysis of Legal Transitions." Harvard Law Review (January).
Â
Page 331
Kotlikoff, Laurence J. (1992). Generational Accounting: Knowing Who Pays, and When, for What We Spend.
New York: The Free Press.
Lyon, Andrew B. (1990). "Invariant Valuation When Tax Rates Change over Time." Journal of Political
Economy XCVIII (2, April 1990): 433-437.
(1992). "Tax Neutrality under Parallel Tax Systems." Public Finance Quarterly XX (3, July): 338-358.
McLure, Charles E., Jr., Jack Mutti, Victor Thuronyi, and George Zodrow (1990). The Taxation of Income
from Business and Capital in Colombia. Duke University Press.
, and George R. Zodrow (1991). "Implementing Direct Consumption Taxes in Developing Countries." Tax
Law Review XLVI (4, Summer): 405-487.
Paul, Deborah L. (1997). "The Sources of Tax Complexity" Forthcoming in North Carolina Law Review.
Rose, Manfred, and Rolf Wiswesser (1997). "Tax Reform in Transition Economies: Experiences from
Participating in the Croation Tax Reform Process of the 1990s." Forthcoming in Public Finance in a
Changing World, Peter Birch Soerensen (ed.). London: Macmillan Press.
Samuelson, Paul A. (1964). "Tax Deductibility of Economic Depreciation to Insure Invariant Valuations."
Journal of Political Economy LXXII: 604-606.
Sandmo, Agnar (1979). "A Note on the Neutrality of the Cash Flow Corporate Tax." Economic Letters IV
173-176.
Sinn, Hans-Werner (1987). Capital Income Taxation and Resource Allocation. New York: North-Holland.
Shaviro, Daniel (1997). Tax Transitions and Tax Politics. Forthcoming. Chicago: University of Chicago
Press.
Shoven, John B., and Jeremy I. Bulow (1975). "Inflation Accounting and Non financial Corporate Profits:
Physical Assets." Brookings Papers on Economic Activity 3: 557-611.
270
271
Wenger, Ekkehard (1983). "GleichmÃ¤ssigkeit der Besteuerung von Arbeits- und

VermÃ¶genseinkÃ¼nften." Finanzarchiv, 207-252.
Zodrow, George R. (1981). "Implementing Tax Reform." National Tax Journal XXXIV (4, December):
401-418.
(1985). "Optimal Tax Reform in the Presence of Adjustment Costs." Journal of Public Economics, XXVII (2,
July): 211-230.
(1988). "The Windfall Recapture Tax: Issues of Theory and Design." Public Finance Quarterly XVI (4,
October): 387-424.
Â
Page 333
12
What's in a Name? Income, Consumption, and the Sources of Tax Complexity
Professor Paul has provided a nuanced analysis of complexity in the tax law. 1 She may well be right about
the tangles that would emerge, should the country seek to travel the consumption tax route. I fear, however,
that her treatment of problems that could arise under a variety of circumstances may distract from the
potential for simplification offered by the consumptionand not by the incomeapproach to tax design. After
reading her article, I remain convinced that the consumption approach wins the simplicity contest.
Professor Paul identifies three sources of complexity: the quest for equity, the financial stakes in reducing
uncertainty, and interest group politics.2 My basic position is that, unless one defines equity in terms of
Haig-Simons income, consumption tax approaches are superior on all three counts. To be sure, any system
can be made complex (in all three dimensions that Professor Paul identifies). In our political process,
complexity is a likely outcome in any regulatory regime (need I mention pensions, financial institutions,
environmental regulation?). The consumption tax approach does not promise nirvana. But I think that once
the approach is understood, it is possible and even likely that simplicity relative to current law would result.
In my view, the theoretical contrast between consumption and income as guiding ideas for a tax base has
diverted attention from how actual taxing schemes that fit under one or the other rubric might actually work.
Take as an example a value-added tax of the consumption type implemented by the subtraction method.3 In
such a system, each business is taxed on the difference between gross sales and purchases from other
businesses. If all businesses are taxed at the same rate, interbusiness transactions net out, so the tax base in
the aggregate equals the sales by the consolidated business sector to
Â
271
272
Page 334
households. This is a measure of national consumptionone reason for applying the term "consumption type"
to this tax. A well-known feature of such a tax is that purchases of capital goods from other businesses are
immediately deducted (that is, expensed) by the acquiring business. Changing the accounting for such
purchases to an income basissubstituting depreciation allowances for expensingconverts a consumption tax
into an income tax. It seems to me that this is not a very dramatic difference.
One reason why this is made out to be a dramatic difference is that economists have taught us that under an
income tax, "capital income" is fully taxed, whereas under the consumption tax, it is completely free of tax.
Exempting income from capital would seem, surely, to be a regressive approach. Yet if we study the actual
systems I described, we see that the difference is that the taxpaying business gets an acceleration of a certain
deduction. The great fortunes are not made on early versus late deductions. They are made by inventing
Microsoft DOS or monopolizing the cigarette industry. These sources of great wealth are taxed alike under
either the consumption-type or the income-type value added tax. The labels, "consumption" and "capital
income" (the rest being, presumably, "labor income''), divert commentators from looking at what actually is
occurring. The valid point is that in principle, the difference between income and consumption taxes is the
treatment of the risk-free reward to waiting. 4 Whether or not one agrees with me that the timing of
consumption should not affect a person's discounted tax burden, this treatment does not amount to much.5
There are, to be sure,

consumption type systems that
would markedly change the
distribution of the tax burdens.
Replacing the income tax with a
proportional sales tax, with no
adjustment in transfers or
programs such as the earned
income tax credit, would be just
such a major change. But the
same comment would apply if
the replacement were an
income-type value-added tax.
I agree with Professor Paul that the equity issue drives the debate. As we both see it, the equity challenge to
all tax systems is to attach the right burdens to the right people. I have generally argued that the right people
to bear relatively more of the tax burden are those who are fortunately endowed with skills and opportunities.
Identifying those people is very difficult, but it also is my view that income and consumption taxes
discriminate among people in about the same way in the most important dimensions. Where they differ, in
differ-
Â
Page 335
ently discriminating among people who differ in their taste for the timing of consumption or the timing of
their endowed earnings, it seems to me the income tax is inferior. 6
Professor Paul gives a good example of the failure of a consumption tax to discriminate correctly between
two people who have the same opportunities when she notes that "a person who has the opportunity to earn
$100 but chooses not to earn the $100 has the ability to pay tax on $100 (because she has the ability to earn
the $100), but she does not have any consumption because she does not in fact earn the $100 and consume
272
273
it."7 It is unclear why Professor Paul does not emphasize that consumption and income-type taxes are equally
deficient on this score because the hypothetical person also would have no income. To me, this is the central
defect of both income and consumption taxes but hardly a particular shortcoming of consumption taxes
relative to income taxes, the bone of contention here.
In my view, if one asks not whether consumption taxes will produce greater simplicity than income taxes, but
instead whether the consumption approach offers significant advantages if one is setting out to design a fair
and simpler system, however one may choose to label it, the answer remains, "yes."
The distractions in this quest are, by and large, of two types. One set of distractions relates to adjustments that
are often made within proportional taxes to make them more progressive. For example, Professor Paul notes
that a "common approach in the case of a retail sales tax or a value added tax is to exclude necessities, such
as food and prescription drugs, from the tax base."8 It is hard to understand why this admittedly common
source of complexity is necessary in the first place in a system that permits, through transfers, a wide variety
of superior methods of vertical adjustment.9 As a replacement for the existing income tax, an indirect income
tax would suffer from the same shortcomings and attract the same sorts of remedies.
As pointed out by Professor Paul,10 two consumption type reforms are capable of satisfying the
not-much-change-in-overall-progressivity requirement: a direct tax (epitomized by the Cash Flow Tax
proposal spelled out in Blueprints for Basic Tax Reform11, and a proportional indirect consumption tax (such
as a value-added tax) accompanied by a major change in the transfer programs, such as expansion of the
earned income credit. In the latter category, I put the Flat Tax or the (more progressive) variant that I have
dubbed the
Â
Page 336
"X Tax." 12 Either of these approaches would permit progressivity similar to that of the present system.
Whether they would satisfy other criteria, including other equity criteria is, to be sure, debatable. But they
should indicate the direction in which to look when encountering progressivity concerns while pursuing the
consumption approach.13
The second set of distractions

relates to various measurement
problems that might arise under a
consumption approach that
satisfies the requirement of
progressivity. Examples in
Professor Paul's article include the
treatment of owner-occupied
housing and consumer durables,14
measurement of consumer surplus
and nonmarket consumption,
including leisure,15 distinctions
based on the timing of cash
flows,16 distinctions between real
and financial transactions,17
distinctions between domestic and
foreign location of consumption,18
and distinctions between services
(including education) and goods.19
273
274
I believe that for these, and for most of the other examples raised in the article, there are relatively simple
responses. Rather than take them up seriatim, I would emphasize two general points to keep in mind while
thinking about such issues. First, if, as is generally accepted, an income tax is supposed to be based on the
sum of a person's consumption and change in wealth during a period, any question about how to interpret the
consumption concept should be about the same in both systems. So, for example, an ideal income tax would
measure the yield obtained from an owner-occupied home, in the form of consumption services and accruing
wealth. An ideal consumption tax would measure only the consumption piece.20
Second, and more important, the object

of the game is not to measure and tax
consumption, or income, for that
matter. The object is to impose tax
burdens on the right peopleto achieve
equityin a simple manner. That is why
I usually speak of consumption
approaches or the consumption
approach rather than consumption
taxes. Thus, if one accepts as a good
outcome the result of treating
owner-occupied housing services as
consumption and taxing them
currently, a system thatwhile not
measuring and taxing such
consumption currentlynevertheless
imposed the same burden, understood
as the same discounted tax liability, is
fine.21
Suppose, for the sake of argument, that my equity argument is accepted and that substance, rather than form,
is to guide our design
Â
Page 337
of instruments. (So, for example, we are not put off if someone who receives a payment of interest does not at
that point send a check to the Treasury, provided we are convinced that person bears an appropriate share of
the tax burden.) Since it is conceded that many sources of complexity are the same in the income and
consumption approaches (such as dealing with the family and the business-personal boundary), what are the
relative advantages of the consumption approach?
I believe that the main advantages arise from (1) the possibility of virtually eliminating the taxation of
financial transactions from the personal tax and greatly simplifying their treatment in the business tax; and (2)
making feasible the correction of the tax base for inflation. Although these may seem to be minor, technical
matters, consider just three examples of matters in which these advantages arise: new financial instruments,
retirement savings, and capital gains.
New Financial Instruments
The U.S. economy is extraordinarily dynamic. Innovation and flexibility are its hallmarks. Nowhere has
innovation proceeded more rapidly than in financial markets. The ingenuity of Wall Street's rocket scientists
is legendary. The tax system relates to these developments in two ways. First, much of the ingenuity is
directed toward obtaining the best tax results in connection with any particular economic activity. Indeed,
274
275
some of the ingenuity is directed to making profit at the expense of the U.S. Treasury. The tax rules also
obstruct the accomplishment of new financial arrangements. 22
Retirement Savings
The second example is retirement savings.23 Practitioners know how difficult it is to assure compliance with
the law, and how much talent is devoted to designing plans that fit within the regulatory constraints. Yet the
contractual relations between employer and employees serve a large variety of economic functions. As with
financial innovation we are, on the one hand, wasting resources figuring out how to make the arrangements
that most exploit the tax advantages, while staying within the law, and on the other hand, tying the hands of
our employers and employees in reaching the most effective compensation arrangements.
Â
Page 338
Capital Gains
It used to be said that half of the business of a tax accountant or lawyer was converting ordinary income to
capital gains. The other half was converting capital losses to ordinary income losses. This was especially true
when the difference in the rate of tax applied to the two types of income was larger than it has been until
recently. But the difference has widened again. And in any case, because of the necessity to limit capital
losses, the rules that distinguish the income according to its ordinary or capital gain "character" have
continued in force and account for many inches of tax law books. In view of the interminable controversy
about capital gains taxation, it need hardly be said that the rules involve much more than fees for accountants
and lawyers, but enter into many aspects of the business and financial life of the country, with costly effect.
All three of these examples have in common that they are sources of great complexity and impose costs on
the U.S. economy. These costs are borne not only by the businesses and individuals who must directly
comply with the rules in question. Like the costs of banking laws or securities laws or telecommunications
regulations, the costs that they impose are spread throughout the economy, striking workers and consumers at
all income levels. These three examples also have in common that they are hard to fix within an income
approach, given the lack of acceptance and impracticality of universal mark-to-market accounting.
A third common element shared by these three examples is that they are easily dealt with in a consumption
tax approach.
Not just any old consumption tax will translate into a simpler (or otherwise better) tax law. An ill-informed
application of the ideas can lead to something like the European value-added taxes or the state sales taxes. 24
These may be simpler than the U.S. income tax, but they are hardly encouraging models for us to follow. On
the other hand, a consumption approach to tax design grounded in a clear grasp of the underlying logic does,
indeed, hold the potential for a much simpler law. Professor Paul has added to our knowledge of the
complexity-producing processes. As far as the consumption-income base choice is concerned, I hope that
readers will take away from her account a catalogue of traps for the unwary rather than a conclusion that the
consumption approach cannot be deployed to advantage in the quest for a simpler tax system.
Â
275
276
Page 339
Notes
1. See Deborah L. Paul, "The Sources of Tax Complexity: How Much Simplicity Can Fundamental Tax
Reform Achieve?," 76 N.C.L. Rev. 151 (1997).
2. See id. at 163-80.
3. See id. at 185.
4. I say "in principle" because in practice, income taxes are such bad approximations of the ideal. Consider,
for example, the taxation of risk-free interest with even very modest inflation. Professor Paul notes that the
federal income tax (along with the state income taxes) does tax inflationary gains. See id. at 191 & nn.
146-47. The inflation problem is much more problematic than is suggested by Professor Paul's discussion.
The federal tax also allows deduction of the inflation premium in interest paid. As a result, in a time of
inflation, the inconsistency between the treatment of different forms of financial returns is likely to permit
sufficiently clever taxpayers to arbitrage away their "capital income." For example, they can borrow to buy
assets that appreciate. It is true that the appreciation is overtaxed (because of failure to correct the basis for
inflation) but the interest deduction may be so exaggerated that it overcompensates. Inflation is a killer of an
income tax that does not use mark-to-market accounting.
5. This point is slowly being absorbed into the conventional wisdom. See Joseph Bankman & Thomas
Griffith, "Is the Debate Between an Income Tax and a Consumption Tax a Debate About Risk? Does It
Matter?," 47 Tax L. Rev. 377, 407 (1992), David F. Bradford, "Consumption Taxes: Some Fundamental
Transition Issues," in Frontiers of Tax Reform 123, 128 (Michael J. Boskin ed., 1996); William M. Gentry &
R. Glenn Hubbard, "Distributional Implications of Introducing a Broad-Based Consumption Tax," in 11 Tax
Policy and the Economy 1, 1-2 (1997).
6. As Professor Kaplow has pointed out, the discrimination under a conventional Haig-Simons income tax
between two individuals, alike in the discounted value but differing in the timing of their earnings, is due to
the failure to account consistently for human and other capital. See Louis Kaplow, "Human Capital Under an
Ideal Income Tax," 80 Va. L. Rev. 1477, 1490-94 (1994).
7. Paul, supra note 1, at 194.
8. Id. at 195.
9. Professor Paul applies the positive-sounding adjective "flexible" to the option to impose different taxes on
different goods under retail sales and credit-and-invoice value-added taxes. See id. at 186. Note that this
flexibility, which may have something to do with the political success of these forms of tax, is arguably
undesirable on all the criteria usually accepted by policy analysts. It is ineffective as a means of adjusting for
equity, and utterly unnecessary when there is a personal tax and transfer system. It is a huge source of
complexity and compliance costs. It is wonderful for playing interest group games. Least certainly, but
probably, it is a source of extra deadweight loss relative to a uniform rate. (Whether this is so depends on a
lot of cross-elasticities about which we have virtually no evidence.)
10. See id. at 181.
11. Department of the Treasury.

Blueprints for Basic Tax
Reform (1997). This source is
also available, in convenient
format, with index and with a
preface by me containing
276
277
Â
Page 340
commentary that benefits from hindsight, as David F. Bradford et al., Blueprints for Basic Tax Reform
(2d ed. 1984). For further discussion of the cash flow tax, see generally David F. Bradford, Untangling
the Income Tax 75-99, 316-20 (1986) [hereinafter Bradford, Untangling the Income Tax].
12. The "X tax" scheme is in the class of "two-tiered cash-flow taxes" described in Bradford, Untangling the
Income Tax, supra note 11, at 59-74, 329-34.
13. A brief comment, cutting across the income-consumption issue, on simplicity aspects of the choice
between the indirect-with-adjustment and the direct approaches: as Professor Paul notes, in this choice there
are forces working in both directions. See Paul, supra note 1, at 181-87. There exists a striking example,
however, of the potential for simplicity and low compliance costs in a system with sufficiently clear
conceptual underpinning: the social security payroll tax. See 42 U.S.C. Â§Â§ 302-433 (1994). It is
interesting that an instrument second in revenue only to the individual income tax, with a rate of over 15%,
seems so relatively free of complexity and low in compliance cost.
14. See Paul, supra note 1, at 199.
15. See id. at 200.
16. See id. at 202.
17. See id. at 205-06.
18. See id. at 206-07. In a personal consumption (or income) tax framework, the distinction would never
arise. In an indirect tax framework, the matter is a little subtler. Ignoring for the moment cross-border
shopping (tourism), enforcement problems, and transition, the two approaches (origin versus destination
principles) have to be essentially equivalent, since trade surpluses equal trade deficits in discounted value.
Aside from appearances, which are important politically, the border adjustment question, as I see it, involves
transition and a trade-off between two contending administrative approaches. First, with no adjustment (sales
to abroad are taxed, purchases from abroad are deducted), there is no need to police the border, but transfer
pricing problems are a nightmare. Second, with adjustment, there is no transfer pricing problem, but you have
to check at the border, and goods purchased abroad as a tourist escape U.S. tax The latter problems are
diminished if tax regimes and rates are similar in different countries.
In a true individual Haig-Simons income tax, the situation would be comparable to the Blueprints Cash
Flow Tax. See Department of the Treasury, supra note 11, at 9. If an entity-level tax (e.g., value-added
tax or corporation income tax) is employed, I do not think that one can seriously argue that the treatment
of international transactions is not much easier under a consumption tax than under an income tax.
19. See Paul, supra note 1, at 207-08.
20. It is ironic that Professor Paul singles out as a potential source of complexity in a consumption tax the
need to measure the consumption element of owner-occupied housing, and the similar consumption yield of
other consumer durables. See id. at 199. Whereas exactly the same problem exists in an income tax but is
insoluble, one of the nice things about consumption approaches is that they get this right, in the sense of
consistency with the treatment of other assets. This (especially owner-occupied housing) is generally taken as
one of the major advantages of the consumption over the income approach. For an extensive discussion of
this point, see, for example, Bradford, Untangling the Income Tax, supra note 11, at 85-86.
Â
277
278
Page 341
21. This is the effect of a value-added tax applied to the purchase of housing and of the disallowance of a
deduction for purchase of a house in the Blueprints Cash Flow Tax. Similarly, in the Blueprints Cash Flow
Tax, the taxpayer is allowed to choose between qualified account and tax pre-payment treatments of financial
assets and liabilities. See Department of the Treasury, supra note 11, at 18, 121-22. The choice affects the
timing of tax payments but not their discounted value. If the tax rate varies over time, this present value
equivalence breaks down, but Blueprints argued that the self-averaging effect of allowing the taxpayer to
choose between treatments was one of the strengths of the proposed system. See id. at 123-24.
22. David Hariton provides a superb illustration of the problems, taking as an illustration of the state of the
law the steps involved in advising a taxpayer who owns low-basis stock and borrows identical stock and sells
it short. See David P. Hariton, "The Tax Treatment of Hedged Positions in Stock: What Hath Technical
Analysis Wrought?," 50 Tax L. Rev. 803, 804-09 (1995). This would seem to be a pretty straightforward, if
not exactly everyday transaction. Having first established that any number of possible complicating
characteristics of the taxpayer do not apply (the taxpayer is not a controlled foreign corporation, for
example), Hariton goes on for many pages discussing options that the tax advisor must resolve, rules that
may or may not apply, apparent inconsistencies, and so on, and concludes there is no unambiguously correct
advice. See id.
23. When I started as an assistant professor at Princeton University a long time ago, there was a retirement
plan that called for contributions that were graded according to age, one rate up to age 30, another to age 40,
and the highest rate over age 50. At some point Princeton changed its plan, but let existing faculty elect to
continue on the "old plan." The new plan was more generous than the old plan up to age 50, and less
generous after age 50. I chose to stay on the old plan. By the time I reached age 50, most faculty members
were on the new plan. Those of us who had survived so long at the University were, perforce, in the senior
and hence better-paid ranks of the faculty. At that point it was determined that the old plan discriminated in
favor of better-paid employees, and was therefore ineligible for favorable tax treatment. We had to switch to
the new plan, notwithstanding the fact that we had just reached the point where our persistence in the old plan
was about to pay off! Believe it or not, my objective here is not to plead for relief from this injustice, but
simply to call attention to one of the most complex areas of the tax law.
24. Cf. Paul, supra note 1, at 195 nn. 163-64 (comparing complexity under a U.S. consumption tax to
complexity under European sales taxes); id. at 195 n. 165 (comparing complexity under a U.S. consumption
tax to complexity under state sales tax regimes); id. at 206-07 (describing difficulties that state sales tax
regimes encounter regarding taxation of goods purchased outside but consumed within the state); id. at
207-08 (suggesting that the difficulties that states have experienced with teasing apart goods and services also
would be problematic for a federal consumption tax).
Â
Page 343
13
Treatment of Financial Services under Income and Consumption Taxes
Why are financial institutions singled out for special attention in discussions of "fundamental" tax reform?
Why not farms? Pharmaceutical manufacturers? State and local governments? Multinational companies? The
278
279
economic activities of each of these would be profoundly affected by fundamental tax reform. Yet they do
not receive special attention. The reason, I believe, is that three of the reforms now widely discussed include
a business tax much like a value-added tax. Experts on the value-added tax have long recognized that it is
difficult to apply to financial institutions the same tax system applicable to most other companies.
Four plans for restructuring the U.S. tax system have been the focus of discussion: institution of a federal
retail sales tax, a value-added tax, a "flat" tax, and the USA (unlimited savings allowance) tax. Representative
Bill Archer (Republican of Texas), chairman of the House Committee on Ways and Means, and Senator
Richard Lugar (Republican of Indiana) are supporters of a sales tax, although no legislative proposal has yet
been introduced. There have been a number of value-added tax proposals, including a highly detailed plan
introduced by Representative Samuel Gibbons of Florida, ranking Democrat on the Committee on Ways and
Means. 1 Representative Richard Armey (Republican of Texas) is a particularly well-known advocate of a
flat tax; others have advanced similar proposals. All are modeled on the flat tax developed by Robert Hall
and Alvin Rabushka. Senators Pete Domenici (Republican of New Mexico) and Sam Nunn (Democrat of
Georgia) developed and introduced the USA tax in April 1995.2
At the heart of all four plans is a tax paid by businesses.3 A retail sales tax requires businesses to pay tax
equal to a percentage of the value of their sales to nonbusiness customers.4 In the other three
Â
Page 344
plans the business tax is a variant of a consumption-type value-added tax implemented by the subtraction
method. 5
Two features of a value-added tax are responsible for the problem in taxing financial services. First, financial
transactions, such as the payment or receipt of interest or dividends or the sale or purchase of securities, are
not counted in determining a business's value-added tax liability. Second, although a value-added tax is
levied on all business sales, not just sales to nonbusiness customers, in one way or another a business
customer gets a rebate of tax paid by its suppliers. Sales from one business to another generate no net revenue
to the government because the tax paid by the seller just equals the reduction in tax for the business buyer.
Final consumers get no such tax reduction. The net effect, therefore, is an indirect tax on individuals collected
from businesses. This indirect measurement fails for some financial services.
The Problem
It is a matter of conventional wisdom that financial services create headaches for consumption-type taxes,
which many people equate with value-added taxes. It is less widely recognized that the difficulties are not
confined to consumption taxes. Hall and Rabushka assert that "banks, insurance companies, and other
businesses that bundle services with financial products present a challenge to any tax system."6 I shall
develop this point at some length later. But first, I should state my thesis: it is the visibility of the problem
under value-added taxes (whether or not they are consumption taxes, as economists understand this term) that
accounts for the attention paid to it. The economically equivalent difficulty shows up in any tax system, but it
generally lacks political prominence.
The problem is created by the fact that banks, insurance companies, and other financial institutions receive at
least some of their income from the difference between the rate of interest they earn on loans and the rate
they pay to depositors and other providers of their funds. But financial transactions are normally not included
in the base for determining a company's value-added tax. Value-added tax is not charged on interest received,
nor is there any credit or deduction for interest paid. If a financial institution earned all its profits from the
spread on financial transactions and had any deductible or creditable expenses, the business would be in a
perennial negative tax
279
280
Â
Page 345
position. If the tax were ''refundable," even a profitable financial institution might forever qualify for a net
refund. And if a company had enough ordinary sales to pay some tax or were consolidated with a company
that had a positive tax liability, it would be obvious that financial institutions were paying less tax than
apparently similar nonfinancial enterprises.
It may not be immediately obvious that the problem has nothing inherently to do with consumption rather
than income as the basis for taxation. Consumption taxes differ from income taxes in the treatment of
investment. Investment outlays such as the purchase of equipment or accumulation of inventory are deducted
immediately under consumption taxes. Under an income tax they are deducted as income is earned. Current
proposals are consumption taxes because investment expenditures are written off at once rather than
depreciated over time.
Value-added taxes would become income taxes rather than consumption taxes if the costs of investment were
deducted as investments depreciate instead of expensed. From this perspective, one can see that a financial
institution can run a perennial loss for tax purposes as readily under an income-type value-added tax as under
a consumption-type. The method of collecting the tax at the business level is what causes the problem. The
problem exists under other approaches, but it is invisible and so of less political concern.
Solutions
Most countries that have a value-added tax exempt financial services. 7 Under an exemption, a financial
institution's value added from financial services provided to customers is untaxed, but the institution is also
denied any credit for taxes paid on purchased goods and services it uses in producing those financial services.
For sales to households, as opposed to sales to businesses, exemption results in a somewhat lower tax than
would be due if the financial services were treated like other sales. But because no credit is provided for taxes
paid on inputs, all that is left out is the value added by the financial organization. Exemption in an
invoice-and-credit value-added tax comes close to taxing financial services "correctly." The more the costs of
the services derive from inputs purchased by the financial business from other companies (and the less from,
for example, the financial company's own employees), the closer the approximation.
Â
Page 346
Exemption is no simple cure, however. For one thing, if financial services provided to households are
undertaxed, financial services to other businesses are overtaxed because the businesses receive no credit for
inputs used to produce the services. Furthermore, exemption requires distinctions among types of purchases
and sales. The result is complexity and the unwanted incentives typical of line-drawing in a tax system.
Merrill and Adrion observe that "the VAT rules applicable to financial services are among the most complex
in the entire VAT system." 8
Countries exempt financial services in invoice-and-credit value-added taxes to approximate the tax rate
applied to other goods and services. But even if the result is close to "right," many countries impose separate
taxes on financial products to offset the value-added tax exemption. Merrill and Adrion suggest that the
combination probably taxes financial services more heavily than other goods and services.9 The imposition of
280
281
extra taxes to offset a tax advantage that may not really exist illustrates the political importance of
appearances.
The Proposals' Approaches
U.S. retail sales taxes typically exclude financial services, whether the customer pays for them explicitly or
accepts reduced interest. Even if explicit charges for financial services were subject to retail sales tax, a
functionally similar charge imposed by paying below-market interest rates might attract little attention,
because it would not put the financial company in a negative tax position. The value-added tax elements of
the other three plans do, however, create this possibility, and they do incorporate special rules for financial
institutions and services or both.
The USA tax is much the most explicit in this respect, as well as in others. For most businesses, the tax base
would consist of the difference between sales and purchases from other businesses, with no account taken of
financial flows. Banks and bank customers, insurance companies and their products, and investment conduits,
such as mutual funds and real estate investment trusts, would all have special rules that resemble a
conventional income calculation.10
Congressman Gibbons's value-added tax proposal is not specific on the details but asserts that procedures
would be developed to tax
Â
Page 347
financial services. It appears the intent is to use methods like those proposed in the USA tax. 11
The flat tax would also provide special rules for banks and insurance companies. Banks would have to report
the price of the services they provide to depositors, measured as "the difference between the market interest
rate and the lower rate that the bank pays on accounts that have bundled services."12 Similarly, the service
element in mortgage interest charges, in the form of higher interest charged than the market interest rate,
would be added to the tax base of the bank. An analogous procedure is envisioned for insurance companies.
Taxation of Financial Institutions under Existing Law
The simplified examples of the tax treatment of various financial transactions presented below do not
describe how financial institutions are now taxed. These details would be important in thinking through the
consequences of major changes in the tax law for existing financial businesses. To compute corporation
income tax liability, life insurance companies, for example, have to determine the policyholders' share of
tax-exempt interest; and mutual insurance companies are subject to provisions designed to place them in a
particular relationship to stockholder-owned companies.13
In spite of such complexities, however, two basic tax regimes apply to financial institutions. Some
institutions such as mutual funds are treated as pass-through entities. Provided various requirements
concerning distributions are satisfied, income from the business's portfolio is allocated to shareholders and
taxed to them. Because income is net of administrative and management costs, services provided by such
companies are excluded from the income tax base. The other regime, applicable to banks and insurance
companies, applies the same income-measurement rules as the ones nonfinancial businesses face. The main
difference, important primarily for insurance companies, is that they may set up deductible reserves for future
losses (payouts on the policies for insurance companies, bad debts for banks). That is, they can deduct
currently the value of expected future losses. Such a deduction results in the conceptually correct measure of
the change in net worth of the business, but ordinary companies may not take such deductions. In this respect
banks and
281
282
Â
Page 348
insurance companies enjoy a tax advantage relative to ordinary companies.
Finally, life insurance offers individual policyholders a well-known tax advantage. The liability for death
benefits or for future annuity that insurance companies may deduct corresponds to accruing income of the
policyholder. The failure to impose income tax on policyholders for this "inside build-up" reduced federal
revenues by an estimated $10 billion for fiscal 1995. 14
This chapter recapitulates the challenge posed by the taxation of financial services and examines options for
confronting it. A principal theme is that in most instances in which financial services might go untaxed under
the reform plans (absent special, and complicating, rules), they are also free of tax under the current income
tax. The locus of the nontaxation is shifted, however, from the individual to the business, and the problem
thereby becomes more obvious. A second theme is that relatively simple techniques of cash-flow accounting
would tax financial services consistently with other goods and services in consumption tax systems.
Nevertheless, cash-flow accounting approaches have not yet proven attractive to policymakers in value-added
tax countries, in part because cash-flow approaches raise transition issues that would need to be addressed in
the design of tax rules.
The Canonical Case: Demand Deposits
Demand deposits illustrate the problem of taxing financial services.15 Typically, banks pay little or no interest
on demand deposit balances, but they provide checking services "free." More accurately, the depositor pays
for the checking services by accepting a lower rate of interest than could be earned from other equally risky
and liquid assets. For example, suppose the going interest rate is 10 percent, and the bank pays 4 percent on
demand deposits, provided the depositor maintains an average monthly balance of at least $1,000. A carefully
calculating depositor will keep, let us say, an average balance of $1,500, so the bank pays $60 a year in
interest. In the meantime the funds are invested at 10 percent, yielding $150. The $90 difference covers the
bank's noninterest costs; any net surplus accrues as profit. I shall assume that the $90 exactly covers the
bank's costs, so that there is no profit and that the costs consist entirely of payments to companies that
provide computer services.16
Â
Page 349
Table 13.1
Treatment of a Demand Deposit under the Income Tax
Item Value
Information
Account average balance $1,500
Cost of service 90
Service charge 0
Interest paid on account 60
Bank's taxes
282
283
Interest received 150

Interest paid on deposit (60)
Cost of providing service (90)
Taxable income 0
Summing up
Interest bank received 150
Tax bank pays 0
Interest paid on deposit (60)
Costs (90)
Net 0
Value of service in forgone after-tax interest in 30 percent bracket $63
The Income Tax Problem
The income tax problem is that the depositor is not taxed on the service yield from the funds on deposit. If the
bank paid the full 10 percent interest rate and charged for the services, the depositor would have an additional
$90 in taxable income and have to pay an explicit charge of $90 for checking. A depositor in a hypothetical
30 percent tax bracket would owe $27 in tax. Table 13.1 summarizes the situation.
Under the familiar

Schanz-Haig-Simons standard, income
for tax purposes is the sum of the
taxpayer's consumption and increase in
wealth during the period. If one
examines the typical depositor's
accounts, to say that leaving out the
financial services paid for by forgone
interest is to say one is omitting some
consumption. In effect, under existing
rules the depositor obtains a deduction
for the cost of the checking acount,
which would (or might) be regarded as
a personal expenditure and therefore
not eligible for deduction in normal
income tax usage. The effect is to
subsidize checking account services at
the depositor's marginal tax ratea
zero-bracket taxpayer gets no subsidy.
Â
283
284
Page 350
If checking account services are

properly deductible, there is no
economically interesting
problem. Implicit interest that
takes the form of services is not
taxed, but neither would explicit
interest payments associated
with offsetting deductible
charges. Since a large proportion
of demand deposits, and of
financial services in general, is
for business customers, the
income measurement difficulty
is of less quantitative
significance than might be
thought. For the individual
customers for whom a deduction
would not be allowed, the
problem is measured by their
marginal tax rates and stock of
deposits.
Exclusion of consumption of financial services is a problem for two reasons. First, the subsidy of
demand-deposit services causes inefficiently large use of these services relative to other consumption goods
and services. And the failure to tax income that takes the form of financial services necessitates higher tax
rates to raise a given amount of revenue, thereby increasing the efficiency cost of the tax system.
Second, mismeasuring income from financial services raises the usual vertical and horizontal equity issues.
The effect of mismeasurement on the average progressivity of the tax system could be corrected by lowering
individual tax rates a little. The result would be no change in overall progressivity, but a redistribution from
demand-deposit lovers to demand-deposit avoiders. If there is close correlation between income and tastes for
demand deposits, the change in the distribution would be small. In that case the only issue would be the size
of any efficiency gain from confronting individual taxpayer depositors with the true marginal cost of the
services they consume. 17
One may get a very rough idea of the efficiency stake from the figures compiled by the Commerce
Department on "imputed interest" received by persons from banks, credit agencies, investment companies,
life insurance carriers, and private noninsured pension plans. The estimate for 1993 of $350 billion includes
$204 billion from life insurance and pension funds and $146 billion identified as "services furnished without
payment" from the other financial intermediaries. The $204 billion presumably consists predominantly of the
accrual of inside build-up and only secondarily of financial services. That would leave about $200 billion of
the Commerce Department's 1993 estimate that might measure the phenomenon discussed in this chapter.
Some, perhaps significant, part of this total may well be overstatement of nondeductible interest paid by
house-
Â
284
285
Page 351
holdsmortgage interest of nonitemizers or interest on credit card debt, for example (see the discussion of
mortgage interest later). If, however, all of it is now excluded from the individual base and ought to be
included, adding it to the individual income tax base and taxing it at an average marginal rate of 20 percent
would add more than 6 percent to aggregate federal personal and corporate taxes of $661 billion. The $200
billion may also be compared with aggregate personal consumption in 1993 of $4,454 billion, so it would
amount to more than 4 percent of consumption. 18 The figure seems implausibly large as a measure of
excluded income in the form of financial services.
Is a Fix Needed?
The significance of such income mismeasurement depends on the extent of differences in the use of demand
deposits among taxpayers within income classes and on the extent of overconsumption of demand-deposit
services (the elasticity of demand for demand deposits).19 If members of each income class make similar use
of demand deposits, the issue is wholly one of efficiency rather than equity as these terms are generally
understood. In turn, efficiency needs to be assessed in relation to the difficulty of correcting the income
measurement.
Before turning to possible fixes, one should consider whether demand-deposit services should be taxed. Are
the services consumption? This question is not simply a technical one to which there is a correct technical
answer.20 It is instead a normative question about who should bear the tax burdens. Suppose banks are
required to pay market interest on demand deposits and to charge fees that cover their costs. Should a person
who has $90 of demand deposit fees bear the same tax burden as someone who has no demand deposit fees
but instead $90 extra of other goods and services? The answer might be no if fees are regarded as a cost of
earning a living. This answer may seem fanciful in the case of demand deposits, but not for itemized
deductions for the cost of advice on how to manage a stock portfolio.21 The use of demand deposits may
increase income net of fees by economizing on other more costly methods of settling payments.22
Alternative Fixes: Imputing Income
Methods of imputing in-kind financial services to the income tax base fall into two classes: depositor-side
corrections and institution-
Â
Page 352
side corrections. Within the latter, corrections may be institution specificfor example, applied to companies
identified as banksor transaction specificfor example, applied to demand-deposit services, by whomever
provided.
Income mismeasurement occurs at the depositor level. If depositors face different tax rates, as they do under
the current individual income tax, a precise correction would need to be made at the depositor level.
Depositor income is understated by the explicit fee the bank would charge as an alternative to providing a
lower financial return. To correct the error in income measurement, this forgone payoff to the depositor
would be imputed as income subject to tax.
The problem is that this fee is

not observed. Rather, it is
inferred from the difference
between the actual financial
payoff to the depositor and the
285
286
payoff if no banking services

were involved. The benchmark
payoff is presumably that of an
asset as safe and liquid as a
demand deposit but without its
convenient services. Hall and
Rabushka suggest the required
amount is easily measured as
the difference between the
interest paid on the demand
deposit and the Treasury bill
rate. 23 This simple standard
might work for demand
deposits, whose security and
liquidity are easy to observe.
How this "interest imputation"
approach would work in more
complex cases is less clear.
Since income is imputed to the depositor, there would be no need to distinguish business from nonbusiness
customers. A business customer would deduct the imputed service fee, just offsetting the imputed income.
Rules for taxing financial services often single out particular types of financial institutions as opposed to
particular types of transactions. But business-level adjustments could be based on the nature of the
transaction, whether carried out by a bank or by some other enterprise. Income would be imputed to the
business instead of to the customer who actually enjoys it. This approach resembles the proposal to tax fringe
benefits by denying a deduction to businesses that offer them. The transaction is taxed at the bank's marginal
tax rate. If the bank's marginal tax rate approximates the highest depositor rate, this correction would overtax
service income of all taxpayers in lower brackets. A distribution- and revenue-neutral correction requires
reduced marginal tax rates on individuals at all levels. The choice between banking and other services would
be unchanged for high-bracket depositors, but low-bracket depos-
Â
Page 353
itors would have an incentive to substitute away from banking services.
One problem is that the institution-side correction is wrong for business depositors. An additional correction
would be necessary for business customers. Business deposits could be identified and excluded from the
adjustment, although mismeasurement would persist if businesses differ in their marginal tax rates. The
administrative complications are obvious.
Imputing income to the institutions based on demand-deposit balances is a transaction-specific correction,

triggered whenever the tax authorities observe the bundling of demand-deposit services with a financial
return. The relevant rule could be stated in functional terms so that the same correction applies for a nonbank
institution that provides a similar service. (A depositor-level correction would be necessarily specific to each
transaction.)
Banks could be subjected to tax rules different from those applied to other businesses. The USA tax takes this
approach. Current income tax law also applies special rules to particular classes of institutions, in part to
permit deductions that are not allowed to ordinary businesses to handle future contingencies.
The Consumption Tax Problem: VAT Systems
286
287
Under the consumption tax, more than under the income tax, a variety of rules can achieve economically
equivalent results. Furthermore, the rules may be mixed and matched if reasonable care is taken. 24 I focus on
two main approaches to consumption taxation. Business-level taxes include the value-added tax and retail
sales tax, the Hall-Rabushka flat tax, and the business tax component of the USA tax. The other approach is
an individual tax on all cash inflow less cash outflow for the acquisition of assetsthat is, saving. The
individual tax component of the USA tax is in this category.25
The value-added tax base consists of a business's receipts from sales of goods and services less the purchases
of goods and services from other businessesa tax on real transactions. Such financial transactions as
borrowing and lending, issue and repurchase of stock, and payments and receipts of dividends do not enter
the tax base. What, exactly, is a real transaction? When tax liability turns on a distinction, that distinction will
come under pressure. Consider an automobile dealer who sells a luxury car for only $5,000 but requires
Â
Page 354
installment payments calculated at a very high rate of interest. The sale of the real product, the car, is bundled
with a financial transaction, the installment loan. If the financial transaction is untaxed, the business can
avoid tax by arranging for buyers to pay for purchases largely through untaxed interest on the installment
loan.
One way to protect against tax avoidance using such bundled transactions would be to require separation of
real and financial transactions, perhaps by mandating that the automobile seller and the lender be separately
owned businesses. This requirement would go some way to ensuring the arm's-length quality of the financial
transaction. Or tax law might require that the buyer always have the option to buy the car for cash for the
price quoted in the installment contract.
Still easier would be to require that a bundled transaction be accounted for on a bundled, cash-flow basis. The
seller of the car would treat as taxable income all cash paid by the buyer, whether characterized as principal
or interest. This rule would not require adopting cash-flow accounting in any other respect. It would render
the seller indifferent among payment schedules with the same discounted value and eliminate incentives to
inflate the interest rate.
The Demand Deposit under the VAT
"Real" bank financial services, such as checking accounts and safe deposit boxes, are bundled with the
financial return, the payment of interest. Table 13.2 shows the break-even situation corresponding to the
income tax example. The bank has no identified real sales, but does have real expenses of $90.
Under the VAT the extra $1,500 in demand deposits gives rise to a $90 loss. If the tax is refundable or the
bank owes enough tax on other transactions, its taxes are reduced by its marginal tax rate times this loss. If
the tax rate is 30 percent, the saving is $27. The break-even interest payable on the demand deposit is
increased, relative to the income tax case, by the tax saving: $87 instead of $60. Because the individual does
not pay tax (or deduct expenses), the depositor nets $87 plus demand deposit services by forgoing interest,
which I assume to be 10 percent, on $1,500. The implied value of services to the depositor is the difference
between 10 percent of $1,500 and $87, or $63.
This example illustrates several points. Value-added tax accounting excludes the financial-service value of
the demand deposit,
Â
287
288
Page 355
Table 13.2
Treatment of a Demand Deposit under a Subtraction-Type Consumption Tax
Item Value
Information
Cost of service 90
Service charge 0
Interest paid on account 87
Bank's taxes
Sales 0
Purchases (costs of services) 90
Net tax base -90
Rebate of tax (assuming 30 percent tax rate) 27
Summing up
Interest bank receives 150
Tax rebate to bank 27
Interest paid on deposit -87
Costs -90
Net 0
Value of service to client (forgone interest of $150 less interest of $87 received on the deposit) $63
which is implicit in the reduced rate of return paid to the depositor. For a taxpayer with the same marginal tax
rate as the bank, the outcome is the same under uncorrected value-added tax and income tax regimes.
The effect of the tax rule on the cost of financial services to the depositor is independent of the depositor's
marginal tax rate. That means, for example, that a shift from an uncorrected income tax regime to an
uncorrected value-added tax regime would in this respect benefit low-bracket taxpayers. The implicit subsidy
received by low-bracket depositors is lower under the income tax. A zero-rate depositor obtains no subsidy.
Under the value-added tax, the implicit subsidy is the same for all and is at the value-added tax rate. The key
point here is that the undertaxation of financial services is more or less the same in both direct income taxes
and consumption taxes implemented at the company level. The principal difference is visibility. The
undertaxation of financial services under the consumption tax shows up as negative taxable income. Because
business depositors would be allowed to deduct any service fees that were assessed, the problem has to do
only with services provided to households.
Â
288
289
Page 356
Transaction-Specific Corrections under the VAT
If a VAT has a uniform tax rate, correcting the tax base to include the value of financial services is easier
than it is in a graduated tax regime. If labor earnings, as in the flat tax, are the measure of ability to pay, it is
not necessary to allocate services to particular depositors. The transaction-specific, institution-side
corrections described in connection with the income tax could be extended to the value-added tax.
For example, the bank could be obliged to report as taxable sales some interest rate times deposits. Any
payment made to depositors (whether labeled interest or toasters) would be taken as a deduction and treated
as a rebate on sales. Alternatively, the estimated cost of providing deposit services could be imputed as sales
by the bank (with payments to depositors netted). As with the income tax, these imputations would produce
the wrong result for business depositors. Either the imputation would be omitted for them, or the imputed
amounts would be reported to business depositors to be taken as deductions against their sales.
It is instructive to consider how the various methods for dealing with the automobile installment sale situation
could be adapted to the treatment of demand deposits. 26 The simplest approach, as in the case of the car
dealer, would be to require cash-flow accounting. The bank would treat all incoming cash as taxable sales
and all outgoing cash as purchases. The depositor would be treated symmetrically. A business would take a
deduction for all deposits and include all withdrawals.
I did not mention cash-flow accounting as a possible method for correcting the income tax base, although by
adding imputed interest to balances it might be feasible. It is, however, a natural approach in a
consumption-tax world. The transaction-specific approach would be identified with any bundled transaction
involving a financial product and financial service, not with institutions. Any business providing a bundled
service could be obliged to use cash-flow reporting.
Table 13.3 illustrates the cash-flow approach. The bank's tax rate is assumed to be 30 percent. Accordingly, it
owes $450 in tax on the original deposit of $1,500. As before, the depositor receives $60 in interest, relabeled
a premium payment. This premium, along with the costs of servicing the account, is deductible. The result is
a current year's loss of $150 (larger than in the conventional case), which
Â
Page 357
Table 13.3
Cash-Flow Treatment of a Demand Deposit
Item Value
Information
Up-front tax on deposit 450
Yield on remainder at 10 percent 105
Cost of service 90
Service charge 0
Premium paid on account 60
Bank's taxes
289
290
Sales -60
Less purchases (costs of services) 90
Net tax base -150
Summing up
Interest bank receives 150
Expenses -90
Premium on account -60
Net 0
Value of service to client (forgone interest of $150 less premium of $60 on the deposit) 90
qualifies for $45 in tax rebate at the bank's 30 percent marginal rate. The key to that result is that, although it
is carrying a liability of $1,500 in demand deposits, the bank is able to earn interest on only the net $1,050
left after the deposit tax. The break-even requirement that interest earned less net taxes equals premiums paid
to depositors plus the costs of serving the account results in a net payoff to depositors of the excess of interest
over costs. As a consequence, the depositor pays the full cost of serving the account in the form of forgone
interest, net of any premium earned on the account.
Cash-Flow Treatment: Perception, Cash-Flow, and Timing Problems
The cash-flow method precisely measures income but raises perception problems. In the example the current
loss to the bank is even greater than it is with no adjustment. Could policymakers grasp the effect of the
front-loaded tax paid at the time a deposit liability is created?
Â
Page 358
The initial tax payment is a hidden asset that purchases future tax deductions. In the example the deduction
for liquidating the deposit currently would be $450. The bank loses the income$45 at an assumed interest rate
of 10 percenton this asset, which accrues to the government, exactly offsetting the rebate claimed by the
bank, which is breaking even.
One objection to this approach is

the requirement for a large tax
payment by the bank when an
account is created, coupled with a
large deduction from the base of
a business depositor. The effect is
reversed when the deposit is
drawn down. Apart from start-up
situations, balances would
typically be sufficiently stable to
moderate this problem.
Alternatively both sides of the tax
transactions could be treated as
compulsory loans to and
290
291
borrowing from the government.

The bank would not actually pay
tax on net deposits. Instead, it
would obtain a balance of tax due
on the tax authorities' books, on
which it would pay
nondeductible interest annually.
I have not mentioned a necessary condition for the efficacy of the cash-flow method: the same rate must
apply to deposits and withdrawals. If rates change, powerful incentives arise to manipulate balances. An
anticipated rate hike would boost the value of future tax deductions from reductions in balances. The same
incentive effects of anticipated changes in tax rates apply more broadly to investment decisions in a
value-added tax. 27
To avoid the incentive effect of an anticipated change in rates, an account built up of inflows when the tax
rate is 25 percent would have to be segregated from accounts built up when the rate is 30 percent. This
solution brings problems of its ownthe difficulty of policing the segregation of different vintages. It is not
clear whether this problem is solvable.
Morley English and Satya Poddar describe an analogous method for dealing with time-varying tax rates.28 A
tax calculation account (TCA) would be established to serve as a clearing device to which taxes due on cash
inflows are credited and taxes refundable on cash outflows would be debited. The taxpayer would be charged
interest on a positive balance and receive interest on a negative balance. (To assist with the perception
problem, these payments might be called taxes.) Otherwise no taxes would be paid except for periodic
settling up. They deal with tax rate changes by adjusting balances in the TCA. If rates change, balances
would be adjusted so that the bank's
Â
Page 359
position on reaching a zero deposit balance would be unchanged. In the example given in table 13.3, the bank
starts with a balance of $450 after receiving the demand deposit. If the tax rate is cut to 20 percent, the
balance would be multiplied by 0.20/0.30, so that upon liquidation of the deposit, the bank would owe no tax.
The effect is to neutralize the incentives (and redistributions) that would otherwise be set up by anticipated
changes in tax rates.
Institution-Specific Corrections
Countries using credit-and-invoice value-added taxes have generally chosen a transaction-specific,

institution-side correction. Designated financial services are exempted from the value-added tax, while the
credit otherwise available for value-added tax paid on purchased inputs is denied. Value-added tax is paid on
the inputs to financial services that are purchased from other businesses, but not on the labor and perhaps
profits of the financial institution. This approach has a number of more or less serious drawbacks, particularly
in its failure to distinguish between business and nonbusiness clients. Since it is virtually impossible to
identify financial services rendered within a business (to itself), this role creates an incentive to shift such
activities away from specialized financial institutions.
The Consumption Tax Problem: Individual Cash-Flow Taxes
Individual-level cash-flow taxes raise other problems of measuring financial services. Individuals would pay
tax on all cash receipts and deduct saving. 29 Capital purchases are immediately expensed. Financial assets
would be treated in a way analogous to today's individual retirement account. Flows into so-called qualified
accounts are deductible; flows from the account to the taxpayer are included in taxable income. This
291
292
framework indirectly measures consumption. Any cash that comes in that is not saved must be consumed.
In some cases the exact opposite treatment of saving and its future payoff may be desirable. Financial assets
acquired outside of qualified accounts are ignored for tax purposes. No deduction is allowed, and no
inclusion of any return flow in future taxable income is required. Called the prepayment approach, this
method produces the equivalent result.
Â
Page 360
Qualified Account
Treatment of
Demand Deposits
At the individual level the prepayment method is likely to be applied to demand deposits. An effective tax
would be imposed on financial services paid for in the form of reduced interest received by the depositor. 30
Cash-flow accounting for a demand deposit, which implements a tax on services when it is applied to the
bank, gives the opposite result when it is applied to individuals. Consider a depositor who is allowed a
deduction for a deposit. Subsequent withdrawals are included in the depositor's tax base. The withdrawals
are, however, net of the bank's implicit charges that are netted from the interest that would otherwise be
credited. Consequently, the return to the depositor consists of a financial component that is included in the
depositor's tax base and a service component that is not, just as in the present income tax. The depositor-side
corrections discussed in connection with the income tax can counter the same problem in an individual
cash-flow tax. The tax prepayment treatment of demand deposits deals with the problem more simply,
however. Since the return is not taxed, the depositor will view financial and service returns as equally
valuable.
Other Financial Services
Problems with respect to other financial services tend to crop up in both individual and business taxes, but in
different places.
Mortgage Loans
The value of financial services associated with a mortgage loan can easily be missed under a value-added tax.
Mortgage loans provide an opportunity to charge for financial services in the guise of interest. Hall and
Rabushka note that there is about a 3 percentage point spread between the ''pure" interest rate and the lending
rate for bank loans.31 This differential seems rather high for bookkeeping and similar services and may
incorporate default risk. If it is possible to identify the value of the financial service, and if one decides to tax
it, the alternatives are clear-cut.
Â
292
293
Page 361
Current Law
In deducting interest, an itemizer deducts any embedded financial service charges. So the consumption
element is excluded from the tax base. For nonitemizers, implicit charges for financial services remain in the
tax base and are taxed like other consumption.
Subtraction VAT
If the bank treats mortgage payments as purely financial, they are not included in the bank's tax base. The
nonfinancial costs of providing those services, however, are deductible. The analogy with the demand deposit
is clear.
Without a correction, VAT rules would extend the exclusion from itemizers to all taxpayers. Although the
result would have the appearance of a tax break for wealthy banks, under competitive conditions, relative to
the preexisting income tax, the benefit would flow mostly to lower-income individuals and tax-exempt
borrowers. Consequently, any correction (such as imputing a service charge) would work to the disadvantage
of lower-income individuals and tax-exempt borrowers.
Once again the problem is confined to transactions between business lenders and nonbusiness borrowers.
Interbusiness transactions are self-correcting. Transactions purely among households do not register in the tax
base.
The various options for dealing with demand deposits can be adapted to mortgage and other loans. One
difficulty that does not apply to demand deposits concerns measurement of outstanding liability, because the
contractual amounts may differ from market values. The cash-flow method could be applied and would
include financial services in the tax base. The bank would deduct the amount loaned and take into income all
repayments, which would have to cover service as well as financial costs. Note that it is not necessary to
contemplate special rules for selected institutions. The rules could be based on economic function (bundling
of a loan and financial services) and apply to all institutions.
Individual Cash-Flow Taxes
If individuals borrow on a qualified basis, the bank could set up a line of credit run through a qualified
account with interest charged to the debit balance. To use the line of credit, the borrower withdraws
Â
Page 362
from the qualified account. As with all other withdrawals, the amount is included in the borrower's income.
Repayments, whether called principal or interest, take the form of deposits to the qualified account and are
deductible. It should be apparent that the qualified account approach would permit taxpayers to deduct the
value of financial services associated with a loan.
Under the prepayment approach, neither the loan proceeds nor repayments have any tax consequence. So the
prepayment approach includes financial services in the taxpayer's base.
Obliging taxpayers to adopt the prepayment approach to demand deposits has few drawbacks. In the case of
mortgage borrowing there are policy reasons for allowing flexibility. With progressive rates, a large
withdrawal from a qualified account to buy a house might drive the taxpayer into a higher bracket. Allowing
borrowing on a tax-prepayment basis would eliminate this problem. Consequently, if it were deemed
worthwhile to seek equivalence in the treatment of financial services under the two approaches, one of the
293
294
other methods for imputing the value of services would have to be deployed.
Property-Casualty Insurance
A property-casualty insurance contract can be thought of as consisting of two elements: a wager with the
insurance company and a loan to the insurance company. The wager aspect is evident in what are called
short-tailed policies, in which any payoff occurs within a year or two. A long-tailed policy adds the element
of time. It involves a loan of the policy premium by the policyholder to the insurance company. The company
pays the loan back as compensation for the insured-against events and, perhaps, in the form of premium
rebates. The risk part is the same as for a short-tailed policy. The loan part is best understood by abstracting
from risk.
Current Law
The taxation of a property-casualty

policy, other than for business
purposes, is somewhat peculiar.
Casualty losses that exceed fairly
high limits are generally deductible
under the individual income tax.
Loss payments that exceed the
insured loss are included in income.
Since the policyholder cannot
deduct policy premiums, but loses
the
Â
Page 363
deduction for loss to the extent of the insurance payoff, there is an implicit tax on the insurance policy.
The tax treatment of the insurance policy ignores nondeductible loss. The policyholder pays the premium, net
of any payback, on an after-tax basis. For a short-tailed policy, the service fee embedded in the premium is
thereby included in taxable consumption.
On long-tailed policies companies may provide a financial return net of service costs. The tax rules that treat
premiums received by insurance companies as gross income and allow deduction of expenses, loss payments,
and the increase in the discounted value of expected future payouts (reserves) mean that the implicit yield on
policy premiums will be the before-tax interest rate. 32 The result is a small tax advantage, depending on the
marginal rate of the policyholder. If the premium is regarded as a pure loan to the company, the payback will
be net of service costs. So it is like the demand deposit. The fact that such long-tailed policies as malpractice
insurance are primarily issued to business clients presumably means the tax-free service value in
property-casualty policies is of no importance as a matter of policy. Any benefit to one side of the transaction
is offset by the disadvantage to the other.
Subtraction VAT
Under a VAT the problems for property-casualty insurance are the same as for bank loans.33 If the insurance
premium and the payoffs on the policy are regarded as financial, the company will have expenses but no
income. A VAT does not tax household interest receipts. So the long-tailed lines lose their interest-tax
advantage, especially for high-bracket taxpayers. For low-bracket policyholders and on all short-tailed
policies, the exclusion of service value would rise. Cash-flow treatment of the premium and payoffs by the
company would correct the income measure. Since insurance companies currently treat premiums as gross
294
295
income and deduct loss payments, this practice would depart less from current practice than would the same
treatment of bank deposits. The main change would be eliminating the reserve deduction. It should be
possible to implement such treatment on a transaction-specific rather than company-specific basis.
Individual Cash-Flow Taxes
Qualified account treatment of property-casualty insurance results in the exclusion of the insurance
company's services from the policy-
Â
Page 364
holder tax base. A simple

corrective would be to apply
prepayment rules, with no
deduction of premium and no
inclusion of proceeds. (This
would be independent of the
treatment of casualty losses.)
Life Insurance
Life insurance raises essentially the same issues, with the same implications, as property-casualty insurance.
34 In contrast with a property-casualty policy, where the insured-against event has a monetary value, there are
no natural limits on the amount of life insurance. This feature heightens the tax advantage of long-tailed
insurance that results from the failure to include accruing value in policyholder income. So long-tailed life
insurance is extensively held by individuals.
Current Law
A term life insurance policy has

no effect on income tax for the
policyholder. Premiums are not
deductible. Proceeds paid at
death are not taxed to either the
decedent or the beneficiary.
Since the payoff is net of the
insurance company's costs, the
costs must be reflected in the
price to the policyholder. In
other words, for a term policy
the financial service is taxed.
A long-term policy that accumulates cash value has an additional benefit for the policyholderthe accruing
value of the policy, the "inside build-up," is not taxed. The payoff at death is free of tax and the payoff by
redemption (or annuity) enjoys deferral of tax. In either case the service charge comes out of the payback to
the policyholder and is not subject to policyholder-level tax. Just like the bank account, the long-tailed life
policy provides a setting for service value to go untaxed.
VAT and Individual Cash-Flow Taxes
Under the VAT an insurance policy would be treated as a financial transaction. There would be costs of
servicing policies but neither receipts nor deductions associated with the premiums and loss payments. Like
295
296
the bank, the break-even insurance company will appear to be running losses. Insurance with cash-value
buildup will be relatively less attractive than it now is because all interest, not just the implicit interest in a
life insurance policy, will be free of tax. For low-bracket taxpayers and all term insurance, the net effect is to
increase
Â
Page 365
the extent of service value excluded from the tax base. Cash-flow treatment of the premium and payoffs by
the company would correct the income measure. Although these transactions would typically (perhaps only)
be entered into by identifiable insurance companies, the tax treatment could be specified functionally on a
transaction-specific rather than company-specific basis. For the individual cash-flow tax, the story is the same
as for the property-casualty insurance company.
Conclusion
There are broadly similar tax problems under the existing income tax system and potential alternative,
consumption-type tax systems for all services that may be charged for through implicit reductions in interest
paid to lenders or interest charged to borrowers. The problems show up at different placesat the company
level rather than at the individual client level. Where the client is a business and tax rates are the same, the
problem typically nets out between the two sides of the transactions. In many respects, producing consistent
tax consequences of financial contracts and instruments is easier under consumption-type than income-type
taxes.
Problems relate to services rendered to households. Where these problems seem to require solutionseven
where similar problems go unaddressed in the income taxcash-flow approaches promise relatively simple
solutions.
Notes
I thank Richard Goode, Louis Kaplow, Jerome Kurtz, Paul McDaniel, Nils Matson, Peter Merrill, and
conference participants. I also thank the John M. Olin Foundation and Princeton University's Woodrow
Wilson School for financial support of research on transition and implementation issues. None of these
individuals or organizations are responsible for the conclusions I express.
1. See Gibbons (1993, 1996).
2. For useful discussions of current proposals see U.S. Congress, Joint Committee on Taxation (1995) and
Arthur Andersen (1995). A detailed description of the USA tax, prepared by Alliance USA (1995), was
published as a special supplement, Ernest S. Christian and George J. Schutzer, "USA Tax System:
Description and Explanation of the Unlimited Savings Allowance Income Tax System," in Tax Notes, vol.
66, March 10, 1995. The actual legislative proposal was reproduced as a special supplement by the Bureau of
National Affairs, April 26, 1995.
Â
296
297
Page 366
3. People, not impersonal entities, ultimately bear taxes. Paid is thus not the same as borne. As the example
of the payroll tax paid by businesses illustrates, having companies pay taxes is simply a way to implement
collection. It does not in itself determine the ultimate burden of a tax.
4. Rules on which sales count as retail sales vary greatly in actual versions of this tax.
5. Subtraction method means that tax is based on total sales and purchases, less total purchases from other
businesses, rather than being built up from taxes on each transaction separately. Most of the world's
value-added taxes are based on the invoice-and-credit method, under which tax is levied on each sale and a
credit is given for documented taxes paid on purchases.
6. Hall and Rabushka (1995, p. 73). For an excellent overview of the problems of taxing financial institutions,
see Neubig (1984).
7. Tait (1988) covers the VAT

thoroughly; Tait (1991) gives more
compact treatment. See Messere
(1993, chap. 13) for a history of
value-added taxes as well as a
description of the principal alternative
systems. Good discussions of the
invoice-and-credit method used in
most countries to implement a
value-added tax can be found, for
example, in Aaron (1981), and
Cnossen, Galper, and McLure (1993).
For details of the treatment of
financial services see Chant (1989),
Henderson (1988), Hoffman (1988),
and Peter R. Merrill and Harold
Adrion, "Treatment of Financial
Services under Consumption-Based
Tax Systems," Tax Notes, vol. 68,
September 18, 1995, pp. 1496-1500.
8. Merrill and Adrion, "Treatment of Financial Services," p. 1497.
9. Merrill and Adrion, "Treatment of Financial Services," p. 1497. Emphasizing the great difference between
the usual base of a tax on financial services and the theoretically correct value-added tax base, Weichenrieder
(1994) reaches the same conclusion for the case of insurance in Germany.
10. Alliance USA (1995, pp. 201-43).
11. See Gibbons (1993, 1996).
12. Hall and Rabushka (1995, p. 74).
13. For a compact discussion, see U.S. Master Tax Guide, 1991 (1991, pp. 506-18).
14. Budget of the United States Government, Fiscal Year 1997: Analytical Perspectives, pp. 61-90.
15. Hoffman, Poddar, and Whalley (1987) provide a useful discussion of the treatment of banks under a
value-added tax.
297
298
16. As noted in the introduction, the problem vanishes to the extent that the bank's costs take the form of
employee expenses, because these costs are not deducted from a value-added tax base.
17. An adequate analysis of the efficiency consequences of various rules would take us into the realm of the
second best. The usual presumption is that there would be an efficiency gain when the price to the depositor
is brought closer to marginal cost. Theoretically, however, the subsidy to demand deposits could be offsetting
some other incentive effects of the tax system: so removing the subsidy could reduce efficiency. These
observations ignore general equilibrium effects of changed tax treatment on the
Â
Page 367
prices of financial services (and other prices, for that matter). If demand-deposit services are provided at
constant returns to scale and do not depend on highly specialized inputs, these further effects would be
small. Any reduction in the equilibrium levels of demand-deposit services would be accompanied by
redeployment of resources to other uses with similar value. Presumably, however, the short-run impact
would be hard on demand-deposit providers, who might be expected to oppose higher taxes on financial
services.
18. Imputed interest received by persons is taken from the Survey of Current Business (July 1994), table 8 17.
Personal consumption for 1993 is taken from the Economic Report of the President, February 1996, table
B-1; personal and corporate taxes are from table B-80.
19. For present purposes I can neglect the question that I have often stressed elsewhere of how one ought to
classify people by ability to pay See, for example, Bradford (1986).
20. Bradford (1986).
21. The Tax Reform Act of 1986 limited such miscellaneous deductions to the excess of a certain fraction of
income (in addition to the usual restrictions on itemized deductions). The motivation was not principled
income measurement but revenue and simplification.
22. One alternative to demand depositscashraises the same issue of untaxed consumption services. If there is
no way to reach this consumption, exempting the same consumption in the form of demand-deposit services
may raise efficiency issues. See note 19. The attempt to put quantitative flesh on this argument would
perhaps strain the analytical substance of the Schanz-Haig-Simons concept, which is not grounded in the
welfare economics that economists usually deploy. (Does it matter, for example, whether time saved by using
demand deposits instead of cash is devoted to extra paid work or to leisure?) The comparison to the expenses
of managing a portfolio simply underscores that the "correct" treatment of demand-deposit fees is not
clear-cut. Building on a model advanced by Foley (1970), Chia and Whalley (1989) (described also in
Whalley, n d.) argue that exempting financial services from tax may serve efficiency. See also the
development of the Chia and Whalley analysis by Davies (1991). Grubert and Mackie (1996) have recently
developed a similar argument.
23. Hall and Rabushka (1995).
24. There is a voluminous literature on the various forms of consumption and income taxes. The magisterial
Meade Committee report (Institute for Fiscal Studies, 1978) is particularly important. See also Bradford
(1986, 1987, 1996) and Bradford and the U.S. Treasury (1984).
25. Another example is the cash-flow tax proposal described in Bradford and the U.S. Treasury (1984).
26. An obligatory separation of the real and financial transactions might be effected by setting up a collateral
account at a separate financial institution to secure the interests of the bank providing demand-deposit
298
299
services. Checks written by the depositor would be made good by transfers from the independently
maintained collateral account. (This would correspond to obliging the car purchaser to obtain an installment
loan from an institution with no interest in the sale of the car.) It is harder to come up with an analogue to
obliging the car dealer to stand ready to sell for cash at the quoted price that is the nominal amount of the
installment loan.
Â
Page 368
27. Bradford (1996).
28. English and Poddar (1995). For a discussion of the problem in its general form, see Boadway and Bruce
(1984).
29. The approach described here is that proposed for the cash-flow tax in Bradford and the U.S. Treasury
(1984).
30. Bradford and the U.S. Treasury (1984).
31. Hall and Rabushka (1995, p. 75).
32. For a discussion of the

connection between the taxation of
property-casualty insurance
companies and competitive
premiums, see Bradford and Logue
(1995). See also Main (1983).
33. Barham, Poddar, and Whalley (1987) and Thomas S. Neubig and Harold L. Adrion, "Value-Added Taxes
and Other Consumption Taxes: Issues for Insurance Companies," Tax Notes, vol. 61, November 22, 1993, pp.
1001-11, provide useful discussions of insurance in the context of value-added taxation.
34. The tax treatment of both property-casualty and life insurance companies was substantially overhauled in
1984. For a discussion of some of the anomalies in pre-1984 tax rules for life insurance, see Aaron (1983).
References
Aaron, Henry J., ed. 1981. The Value-Added Tax: Lessons from Europe. Brookings.
. 1983. The Peculiar Problem of Taxing Life Insurance Companies: A Staff Paper. Brookings.
Alliance USA. 1995. "The USA Tax System." Washington.
Arthur Andersen. 1995. "Tax Reform 1995: Looking at Two Options." Office of Federal Tax Services (May).
Barham, Vicky, S. N. Poddar, and John Whalley. 1987. "The Tax Treatment of Insurance under a
Consumption Type, Destination Basis VAT." National Tax Journal 40 (June): 171-82.
Boadway, Robin, and Neil Bruce. 1984. "A General Proposition on the Design of a Neutral Business Tax."
Journal of Public Economics 24 (July): 231-39.
Bradford, David F. 1986. Untangling the Income Tax. Harvard University Press.
. "On the Incidence of Consumption Taxes." 1987. In The Consumption Tax: A Better Alternative, edited by
Charles E. Walker and Mark A. Bloomfield, 243-61. Ballinger. (Revised version published as "What Are
299
300
Consumption Taxes and Who Bears Them?" Tax Notes, vol. 39, April 18, 1988, pp. 383-91.)
, ed. 1995. Distributional Analysis of Tax Policy. Washington: AEI Press.
. 1996. "Consumption Taxes: Some Fundamental Transition Issues." In Frontiers of Tax Reform, edited by
Michael J. Boskin, 123-50. Stanford, Calif.: Hoover Institution Press.
Â
Page 369
Bradford, David F., and Kyle D. Logue. 1995. "The Effects of Tax Law Changes on Prices in the
Property-Casualty Insurance Industry." Princeton University.
Bradford, David F., and the U S. Treasury Tax Policy Staff. 1984. Blueprints for Basic Tax Reform, 2d ed.
Arlington, Va. Tax Analysts (original edition published as a U.S. Treasury study, 1977).
Chant, John F. 1989. "Financial Institutions and Tax Reform." In The Economic Impacts of Tax Reform,
edited by Jack Mintz and John Whalley, 223-54. Toronto. Canadian Tax Foundation.
Cnossen, Sijbren, Harvey Galper, and Charles E. McLure Jr. 1993. The Value Added Tax: Coming to
America? Washington: Tax Analysts.
Chia, Ngee Choon, and John

Whalley. 1989. "Should
Banks Be Taxed?" University
of Western Ontario.
Davies, James B. 1991. "The Treatment of Financial and Non-Financial Intermediation under a Value-Added
Tax." University of Western Ontario
English, Morley D., and Satya Poddar. 1995. "Taxation of Financial Services under a VAT: Applying the
Cash-Flow Approach."
Foley, Duncan K. 1970. "Economic Equilibrium with Costly Marketing," Journal of Economic Theory 2
(September): 276-91.
Gibbons, Sam M. 1993. "A Proposal for a New Revenue System for the Untied States Incorporating a
Value-Added Tax." Paper prepared for the Ways and Means Committee Annual Issues Seminar, March
12-14.
. 1996. "The Value-Added Tax: A Revenue System for America's Future." Washington.
Grubert, Harry, and James Mackie. 1996. "An Unnecessary Complication: Must Financial Services Be Taxed
under a Consumption Tax?" Office of Tax Analysis, U.S. Treasury Department.
Hall, Robert E., and Alvin Rabushka. 1995. The Flat Tax, 2d ed. Stanford, Calif.: Hoover Institution Press.
Henderson, Yolanda K. 1988. "Financial Intermediaries under Value-Added Taxation," New England
Economic Review (July-August): 37-50.
Hoffman, Lorey Arthur. 1988. "The Application of a Value-Added Tax to Financial Services" Canadian Tax
Journal/Revue Fiscale Canadienne 36 (September-October): 1204-24.
Hoffman, Lorey Arthur, S. N. Poddar, and John Whalley. 1987. "Taxation of Banking Services under a
Consumption-Type, Destination Basis VAT." National Tax Journal 40 (December). 547-54.
300
301
Institute for Fiscal Studies. 1978. The Structure and Reform of Direct Taxation: Report of a Committee
Chaired by Professor J. E. Meade. London: George Allen & Unwin.
Main, Brian G. M. 1983. "Corporate Insurance Purchases and Taxes." Journal of Risk and Insurance 50
(June): 197-223.
Messere, Kenneth C. 1993. Tax Policy in OECD Countries: Choices and Conflicts. Amsterdam: IBFD
Publications BV.
Â
Page 370
Neubig, Thomas S. 1984. ''The Taxation of Financial Institutions after Deregulation." National Tax Journal
37 (September): 351-59.
Tait, Alan A 1988. Value-Added Tax: International Practice and Problems. Washington: International
Monetary Fund.
, ed 1991. "Value-Added Tax: Administrative and Policy Issues." Occasional Paper 88. Washington:
International Monetary Fund.
U.S Congress, Joint Committee on Taxation. 1995. Discussion of Issues Relating to "Flat" Tax Rate
Proposals. JCS-7-95. Government Printing Office.
U S Master Tax Guide. 1991. Chicago: Commerce Clearing House.
Weichenrieder, Alfons J. 1994. "Mehrwertsteuer, Versicherungssteuer und Risikoallokation [VAT, Insurance

Tax, and Risk Allocation]." University of Munich.
Whalley, John. n.d. "Taxation and the Service Sector." Canadian Tax Paper 93. In Taxation to 2000 and
Beyond, edited by Richard M. Bird and Jack M. Mintz, 269-97. Toronto: Canadian Tax Foundation.
Â
Page 371
14
Fixing Realization Accounting: Symmetry, Consistency, and Correctness in the
Taxation of Financial Instruments
I
Introduction
This chapter was stimulated by reading a recent article by Alvin Warren, 1 dealing with certain fundamental
problems in income tax accounting for financial instruments. Subsequent browsing in the burgeoning
literature to which Professor Warren's article contributes2 has shown me there is much about the income tax
treatment of financial instruments that I do not know. Some of what is set forth in this chapter will be in the
nature of review for serious thinkers on this subject. I believe, however, there is something new in my
301
302
conclusions.
In his article, Professor Warren develops the difficulties created for the income tax system by its reliance on a
combination of different rules applicable to fixed and contingent returns to the holders of financial
instruments.3 The fixed returns are taxed according to a generalization of the longstanding treatment of
periodic interest. That is, there is inclusion in the taxable income of the holder and a deduction by the issuer
of the equivalent of periodic interest payments, using a yield to maturity calculated from the instrument as the
basis for specifying the applicable interest rate.4 The contingent returns generally are taxed upon a realization
event, with the difference between cost basis and realized amount included in the income of the holder
(deducted if it is a loss).5 Professor Warren calls the first the yield to maturity approach and the second the
wait and see approach.6 A third approach discussed by Professor Warren is the mark-to-market method.7 For
the case of an instrument that is held throughout a period during which it gives rise to no cash flows, the
holder includes in income the increase in the instrument's market value over the period. (If there is a cash
flow during the period, it is
Â
Page 372
added to the increase in market value. This principle would apply equally to assets and liabilities. For an
instrument with a party and counterparty, each inclusion by one party would correspond to a deduction by the
counterparty.)
The main challenge Professor Warren sees is in the taxation of contingent returns. 8 Postponement of income
until realization gives rise to familiar problems due to a differential between the treatment of accruing and
deferred income. Having convincingly demonstrated the limitations of the established approaches to dealing
with this differential, Professor Warren offers in conclusion the following provocative observation:
Serious consideration should therefore be given to reducing the differential by taxing at least some contingent returns in
accordance with a formula, such as the retrospective allocation of gain or the imputation of interest at a standard rate.
Although not without precedent, development of a formulaic approach would be a significant change in the concept of
realization. Such a change may be necessary, for the traditional concept no longer seems adequate to deal with innovative
financial contracts.9
In this chapter, I take yet another look at the taxation of financial instruments. To isolate certain income
measurement problems, I focus on a world with no transactions costs. This setting throws into particular
relief that the critical difficulty is dealing consistently with the intertemporal aspects of transactions. This is
well-known in a sense; timing is central to an income tax. But I shall argue that the difficulties are more
serious than is commonly realized, and the constraints on workable rules in the no-transactions-cost world are
extremely confining. Imputation of interest (at a specific, if not a "standard" rate) is the key to isolating the
timing problem. By contrast, there are surprising degrees of freedom in dealing consistently with
contingency.
Original issue discount (OID) bonds provide a useful illustration of what I mean by the purely intertemporal
problem, since taxing them does not obviously raise issues of contingency or collateral issues, such as
character. The yield to maturity method of assigning taxable income attempts to produce consistency between
the consequences of holding an OID bond and holding an otherwise similar instrument that generates
periodic interest payments. Assuming as the standard the mark-to-market method, which does not depend at
all on transactions, the yield to maturity method may accomplish a
Â
302
303
Page 373
reasonable approximation in practice. To understand or deal with more complex cases, however, it is helpful
to be clear that, strictly speaking, neither the yield to maturity method nor the other familiar rules that rely on
either actual or imputed cash flows between parties to a financial instrument precisely solve the intertemporal
consistency problem. This is partly because the yield to maturity is typically an amalgam of the shorter-term
interest rates that represent the "proper" measure. It is also because longer-term instruments almost inevitably
incorporate some degree of contingency.
As I argue, to eliminate differentials in taxation of the purely inter-temporal effects of financial instruments
in a transactions-based system, one needs to start with the taxation of one basic set of instruments that
accomplishes exchange of dollars at one time for dollars at another time. I take as what I regard as the natural
choice, the taxation of simple, one-period loans that have no risk of default. 10 The taxation of these
elementary transactions then restricts the possibilities for taxing other transactions without introducing
"complications." Chief among the complications are rules, such as the limit on the deductibility of realized
capital losses,11 which I describe as violations of linearity.12
Accordingly, my approach in this chapter is to examine the restrictions on the income measurement rules
applicable to financial instruments implied by the requirement that the rules be linear. More precisely, I focus
on the rules that must apply to a "bilateral financial instrument," by which I mean a bilateral contract that
consists entirely in the specification of cash payments to be made by a party to a counterparty. So, for present
purposes, an ordinary bond is a bilateral financial instrument, but a property lease is not, since the latter
provides that one party deliver certain services in return for cash payments by the other party.
The reason for restricting attention to bilateral financial instruments is not that I believe a special regime for
them would make sense. On the contrary, I take for granted that whatever income measurement rules exist
should apply as well to the restricted class of transactions in bilateral financial instruments. Bilateral financial
instruments are the focus here because they provide the opportunity for the tax arbitrage transactions that call
forth complicating fiscal defenses.
In an income tax based on transactions, the consequence of a particular sequence of transactions over time is
a sequence of taxable
Â
Page 374
income amounts. 13 The rules that translate the transactions into taxable incomes have the property of
linearity, in the sense used in this chapter if, when applied to the combination of two sequences of
transactions (adding them together), they produce the combination (the period-by-period sum) of the
corresponding taxable income sequences. Similarly, if every element of a sequence of transactions is
multiplied by some number (doubling everything, for example), the resulting sequence of taxable incomes is
multiplied by the same number. In particular, if everything is multiplied by negative one, the taxable income
sequence is reversed as well, rendering deductible losses taxable gains and taxable gains deductible losses.14
Linearity is a desideratum of a tidy tax system. Violations of linearity tend to produce inefficiencies and
anomalies, as when an investor is unable to deduct a loss that has no tax motivation. Nonlinearities also
typically imply a reward to careful tax planning, and add to tax complexity. The U.S. federal income tax rules
incorporate many violations of linearity, however. I use the limitation on the deductibility of capital losses as
representative. A taxpayer whose net capital losses in a given year exceed $3,000 is obliged to carry forward
the excess, to be netted against possible future gains.15 Doubling everything about the transactions of a
taxpayer who is right at the $3,000 capital loss limit will result in a net $6,000 loss. Since the extra loss
303
304
cannot be deducted currently, the consequence is generally that the taxable income will not be doubled, in
violation of linearity. Another example is the limitation on the deductibility of investment interest.16 Adding
dividends to a person's transactions will increase the tax due from a taxpayer not up against the limitation, but
will not increase the tax due from one who is. As will become clear, if it is not obvious, both of these
nonlinearities serve to bound tax arbitrage profit opportunities that otherwise would not be self-limiting.17
In the interest of sharpening understanding, I work in this chapter with some extreme assumptions. I have
mentioned the most significant extreme assumption I make: that there are no transactions costs. Arguably, it
is transactions costs that protect the income tax from much more extensive tax arbitrage than currently
occurs. Assuming them away removes the protection and exposes the income measurement problems. I also
assume (perhaps this is implied by the absence of transactions costs) a taxpayer can costlessly take any
zero-value position, borrowing to buy a security, for example.
Â
Page 375
There is a practical as well as an analytical reason to investigate the frictionless world. One of the most
striking developments in financial markets in recent years has been a steady decline in transactions costs,
reflected in the proliferation of new instruments. The chapter thus can be read as exploring problems that can
be expected to get worse, absent redesign of the rules.
A subtheme of the chapter is correctness, about which I have two general points. First, it is quite possible for
rules that fail to produce an accurate measure of a period's income (in the sense of change in accrued wealth)
nevertheless to produce a result that is economically equivalent to accurate measurement from the point of
view of the taxpayer. An example would be some sort of look-back calculation of interest accrued on a
discount bond. Second, it may be essential that income measurement rules involving different sorts of
instruments be related consistently to one another, even if the rules fail to measure income correctly. The
underlying program of most rule writing for financial instruments (for example, the treatment of OID) can be
described as seeking consistent treatment with that applied to plain vanilla interest. But consistency in this
sense in the treatment of different financial instruments is not the same as correctness of the treatment of any
of them. The tax treatment of plain vanilla interest may be a poor approximation of the real return on the
associated asset.
The difference is due to inflation.

It is conventionally accepted that
modest inflation2% to 3% per
yearsafely can be neglected in the
tax system. This chapter's
emphasis on the critical role of
time, and not risk, however,
invites the observation that the
real return to pure waiting (as
opposed to risk taking) in U.S.
financial markets over the past
few decades has been about 0.5%
per year. 18 If this is a reasonably
close approximation to the pure
intertemporal return, a very
modest income tax rate applied to
nominal interest of 2.5% to 3.5%
renders the real, after-tax return
negative. It is somewhat ironic
304
305
that great effort is applied to

assure that all transactions are
treated equally badly.
The most important instance in this chapter of rules that produce a correct result (in the sense of equivalence,
from the taxpayer's point of view, to taxing accruing wealth) arises in connection with taxation of risky
instruments. If the outcome of a risky situation is a gain of $1 by one party to a financial instrument and an
offsetting loss of $1 to the counterparty, an accurate measure of wealth change would add
Â
Page 376
$1 to the income of the winner and subtract $1 from the income of the loser. I think most of the rule writing
effort with regard to contingent returns seeks correctness in this sense. My conclusion is that the objective is
unimportant in the context of risk taking because of the flexibility people have to adjust their financial
positions. It is important that a taxpayer's gains and losses on a given asset be taxed at the same rate, but not
necessarily at the rate applied to other transactions. It is thus possible to identify rules that are economically
equivalent to taxing a taxpayer's change in wealth but make no pretense at producing a correct measure of it.
This chapter is divided into four Sections (preceded by this introduction and followed by a brief conclusion).
In Section II, I present a background discussion of the concept of tax arbitrage and review Professor Warren's
findings. In Section III, I focus narrowly on the problem presented by time, in the absence of risk. Much
recent writing emphasizes the central role of risk. 19 This emphasis is clearly justified. It may not be clear,
however, quite how time and risk interact to challenge the implementation of an income tax. The object of
Section III is to restate the problems of taxing purely intertemporal transactions. I found it desirable, and
probably necessary, to write down my own version of this material in order to address with precision the
problem of dealing with contingency. The main conclusion that I reach in Section III that may be somewhat
new or controversial is the one already mentioned. Strictly speaking, there is very little flexibility about the
"right" way to tax intertemporal transactions. The yield to maturity method, for example, is "wrong." The
practice of using the applicable federal rate for imputing interest to positions of different maturity20 is another
example of an approach that is, strictly speaking, wrong. Taxation based on realization requires either
imputing interest to basis at the going rates during the holding period, or what amounts to the same thing,
imposing a tax on realization that treats the proceeds as having been accumulated at the interest rates
prevailing during the holding period. I refer to the latter approach as the Auerbach method because Alan
Auerbach first spelled it out.21
In Section IV, I suggest that the

taxation of risk per se poses
surprisingly few problems. It
does upset, however, the
efficacy of the method of
imputing interest to basis as a
way of perfecting a realization
system. Instead, of other
approaches that have been
considered, only the Auerbach
method remains as a possibility.
Â
305
306
Page 377
In Section V, I describe an alternative approach to taxing contingent returns using realization accounting. It
eliminates any tax-induced timing advantages or disadvantages from realization. The key idea is that the rate
applicable to gains in excess of the current interest rate may be set freely, but it must be set in advance of any
information about the extent of any such gains. As it turns out, the Auerbach method is a special case of the
alternative approach, which offers considerably greater flexibility for the design of policy.
The title of the chapter suggests both reach, "fixing realization accounting," and narrowness of focus on
financial instruments. As I have indicated, financial instruments are central in the chapter because they raise
most sharply problems due to inconsistency in the taxation of different forms of return. But, with provisos
that I discuss in Section V, I believe that the method I have devised for taxing contingent returns on financial
instruments could be extended to contingent returns generally.
II
Background
A
Tax Arbitrage Profit in Market Equilibrium
The term arbitrage refers to the activity of buying and selling the same thing in different markets, and a profit
results if the price paid to buy is below that received on the sale (by enough to cover the costs of arranging
the pair of transactions). One could mean more than one thing by the term "tax arbitrage." 22 For example,
some would describe as tax arbitrage, the activity of borrowing with deductible interest to purchase a
business asset that offers accelerated depreciation.23 In this chapter, the focus is on tax arbitrage using
financial instruments. The instrument itself specifies cash flows between the parties. An opportunity for tax
arbitrage profit exists when there is a pair of instruments (or pair of packages of instruments) that are
identical in their cash flows but differ in the associated flows of taxable income. If the tax system is linear, by
entering into exactly offsetting positions in the two instruments, taxpayers may be able to reduce their taxes.
The opportunity for arbitrage profit of any kind is inconsistent with equilibrium. So the potential for tax
arbitrage must be eliminated by some combination of adjustments in asset prices, changes in effective
marginal tax rates (including changes resulting from such
Â
Page 378
rules as limits on deductions, so that the marginal rate on an extra dollar of the deduction in question is
effectively zero) and increases in transactions costs.
B
A Canonical
Problem:
Discount Bonds
The tax treatment of a zero coupon, discount bond provides a handy canonical example of tax arbitrage. In
the old days, a cash basis taxpayer holding a zero coupon, discount bond had no inclusion in taxable income
until the bond paid off at maturity, at which point the holder included the difference between the amount
received and basis, the amount paid for the bond. 24 By contrast, the holder of a bond that pays interest
currently was, and still is, obliged to bring into taxable income the successive interest payments.25 Consider
the problem posed by this historical treatment of the two sorts of instruments. For purposes of this discussion,
assume all the taxpayers in question are on a cash basis. Also, since the concern here is with timing, not
306
307
character, assume no difference between ordinary and capital gains rates.
In the absence of taxes, neglecting transaction costs (the cost of arranging the arbitrage) and ignoring possible
unpredictable variations in the short-term interest rates (to which I return below), an investor who undertakes
a sequence of investments in short-term bonds paying ordinary interest and reinvests the principal and
interest, can achieve exactly the same net period-by-period cash flows as the holder of the OID bond. There
are two important points to emphasize about the equivalence between the two positions, the OID bond and
the program of investment in ordinary one-period loans with reinvestment. First, with a little cleverness, two
parties could enter into an agreement to exchange one for the other in such a way that no cash flow ever
occurred between them. In all but superficial details, this would be a pure arbitrage transactionbuying and
selling the exact same thing. Second, in the capital markets, this equivalence would be recognized in that the
two positions would have the same value. They could be exchanged at arm's length, without any cleverness
between two parties.
With the tax rules as assumed, the exact equivalence of the two positions is upset when taxes are taken into
account. The OID bond delivers a better flow of taxable income to the holder because the gain is not taxed
until maturity. Relative to the taxation of the
Â
Page 379
duplicating sequence of short-term bond transactions, the OID bond is tax-favored to the holder.
If the timing of the inclusions of the

holder were identical to the timing of
the deductions of the issuer, the tax
advantage to the holder would be
matched by a disadvantage to the
issuer. A simple solution to the OID
problem then might seem to be to let
the market absorb the tax aspects of
transactions. The problem with this
solution is that the extent of the tax
advantage to the holder of the OID
bond depends on the holder's
marginal tax rate, and the extent of
its disadvantage to the issuer depends
on the issuer's marginal tax rate. (A
tax-exempt entity is functionally the
same as a taxable entity with a zero
marginal tax rate.) If the marginal tax
rates of holder and issuer were the
same, the advantage of the
tax-favored instrument to the holder
would be balanced by its
disadvantage to the issuer. In that
case, the market-determined terms of
the two types of instruments would
be expected to correct for their
different tax attributes. Through
adjustment in the prices of debt
contracts, the implied after-tax yield
307
308
would be equated, regardless of the

cash flow pattern involved, and
regardless of the relationship of the
taxation of any one of them, taken in
isolation, to any particular income
concept.
The possibility of such an adjustment illustrates the point that correctness of the treatment of a transaction is
not required to achieve consistency between the effective taxation of two types of instruments. In the
example, both currently taxed and OID bonds (taxed by the old rules) could coexist in financial market
equilibrium with no real consequences. If the result for the currently taxable bond were correct, then taxing
OID bonds by the old rules also would give the correct result, even though the treatment is not correct,
viewed in isolation. Notice that this is not simply another manifestation of the point that inequities tend to be
erased through market reactions, at a cost in the form of inefficiency. 26 In this case, the market erases
incorrectness at no cost. The opportunities faced by the saver or dissaver would be the same (namely, the
after-tax rate of currently taxed interest), regardless of the form of the chosen instrument. So if the current
taxation of ordinary interest were correct, the saver or dissaver would be taxed correctly, regardless of the
form of the chosen instrument.
The further point also is illustrated by this case, since, with inflation, the basic rule of taxing periodic interest
does not produce a measure of accruing real wealth: Correctness of the taxation of neither
Â
Page 380
type of instrument is required to eliminate the opportunities for tax arbitrage profit. (Note, however, that,
after the market has done its work, with a uniform rate of tax, the incorrect rule of taxing nominal interest
will produce the economic effect of taxing real income. This is because the market will build the taxation of
the ''inflation premium" in interest rates into the interest rates themselves.) 27
By extension of the same line of reasoning, debt involving any arbitrary cash flow, with arbitrary rules about
the timing of inclusion of interest received and deduction of interest paid, could coexist with debt paying
periodic interest and taxed according to the usual rules, provided the same marginal rate of tax applied to all
payers and recipients, and inclusion and deduction were simultaneous. I develop this point at greater length
below where I look, in particular, at the possibility that such market adjustment also would deal with the
cases of contingent return that are the subject of Professor Warren's analysis.28
If it is taken for granted that a single rate system is not a realistic possibility, this analysis serves simply to
highlight the much more serious challenge of designing satisfactory rules for a multiple rate world. The
existence of taxpayers in different marginal rate brackets virtually eliminates the potential to use market
adjustment as a substitute for consistent rules to measure returns over time. Something like the yield to
maturity approach will work for risk-free instruments, but, as Professor Warren's analysis makes clear, that
approach will not generalize to cover all the problem cases.
C
Professor Warren's Analysis
In his essay, Professor Warren developed two general propositions:
(1) The current income tax is based on transactions and accords very different treatment to two classes of
transactions: those that are held to depend for their resolution on contingencies to be determined in the future
(wait and see transactions), and those that are held to involve fixed and determinate terms at the outset (taxed
on some sort of yield to maturity basis).29
308
309
(2) This distinction is not tenable in theory nor, increasingly, in practice.30
The put-call parity theorem provides the key to understanding these propositions. That theorem builds on the
fact that buying and
Â
Page 381
holding a share of stock (with market value S) plus a put on that share with a particular strike price and date
(valued in the market at P), while writing a call on that share with the same strike price and date (valued at
C), produces exactly the same cash flow pattern as holding a zero coupon bond over the period until the strike
date. That being the case, the two positions must have the same value in the market. Letting T ("Treasury")
stand for the discount bond's value, the theorem says:
The theorem derives from the fact that the cash flow consequences of holding the portfolio reflected in the
left side are identical under every contingency to holding the Treasury bond reflected in the right side. (I
ignore here possible consequences of different rights in the governance of the corporation.) The problem is
that the tax treatment of the right side is yield to maturity, and of every element of the left side is wait and
see. In general, the two ways of determining taxable income lead to different results. Specifically, under the
current standard rules, the left side package, the components of which are assumed acquired at the outset of
the option period and disposed of or settled for cash on the strike date, is taxed the way OID bonds used to be
taxed: capital gain on the difference between purchase price and maturity value. 31 The right side is taxed
(roughly) on accrual.32
It seems almost inevitable that the taxation of financial instruments will involve a blend of yield to maturity
and wait and see elements, which I take to be more or less synonymous with "accrual" and "realization"
accounting. Does this inevitably imply the need for nonlinear, anti-abuse rules that render the tax system
complex and introduce unwanted incentive effects (for example, by disallowing true economic loss
deductions)? This is the question to which I now turn.
III
Time and Marginal Tax Rates
A
The World of a Single Marginal Tax Rate
To develop the analysis, I proceed in steps, first looking at the world of certainty, where only time matters. In
that world, I take up first the case where all participants in the market confront the same rate of tax. This is
the context in which it is most likely that adjustments
Â
309
310
Page 382
in the market prices of financial instruments can neutralize potential tax arbitrage profit opportunities.
1
Symmetry is Sufficient
For purposes of the rest of this chapter, I attach particular meanings to the terms "symmetry" and
"consistency" as applied to the tax system's treatment of financial instruments. By symmetry, I mean the sort
of equal and opposite treatment of the party and counterparty to a financial instrument that obtains for debt.
Reed Shuldiner uses this term to describe the tax consequences of a transaction if there is "equivalent"
treatment of the two sides. 33 I mean here by "equivalent" that, whenever a transaction has as a consequence a
deduction from taxable income for one of the parties to a financial instrument, it also has as a consequence an
equal and simultaneous inclusion in the taxable income of the counterparty. Symmetry involves a party and
counterparty. As I use the term henceforth, the concept of consistency supplements symmetry with additional
requirements on the tax results from different sets of transactions of a single party. I give more details on this
notion of consistency as needed below.
When there is just a single marginal tax rate applicable to all transactors, and the tax treatment applicable to
the instrument in question is symmetrical in the sense just defined, tax effects can be capitalized into market
prices of financial instruments. To develop this point, it is sufficient to consider in detail a two-period world,
identified by three time points, 0, 1 and 2, one year apart.
The basic building blocks are simple one-period ("unit") interest-bearing bonds. A Time 0 unit bond sells for
1 at Time 0 and pays off 1 + r01 at Time 1. A Time 1 unit bond sells for 1 at Time 1 and pays off 1 + r12 at
Time 2.
Table 14.1 lays out in general form the cash flows associated with the two possible unit bonds as well as with
an arbitrary alternative
Table 14.1
Cash flows to lender using various instruments
Time point 0 1 2
Time 0 unit bond -1 1 + r01
Alternative instrument -z c1 c2
Discount bond -z02 1
Â
Page 383
instrument; c1 and c2 represent

the payoffs from investing z in
the alternative instrument at
Time 0. For reference, Table
14.1 also shows the cash flows
associated with borrowing in
the form of an OID bond that
will pay 1 at Time 2 and
310
311
nothing at Time 1. Its price at

Time 0 is denoted z02,
reflecting the fact that it is
issued at Time 0 and will pay
off at Time 2, two periods
later.
The two unit bonds fully determine the relative prices of dollars in the three periods. In the absence of taxes,
the price of any arbitrary cash flow sequence is determined by eliminating the opportunity for arbitrage
profit. This is because any sequence can be reproduced exactly by an appropriate package of purchases and
sales of unit bonds. To match the future payoff from the alternative instrument in Table 14.1, invest
in the Time 0 unit bond at Time 0. At Time 1, the payoff will be
Extract c1 and purchase
units of the Time 1 unit bond. At Time 2, the payoff will be c2.
From this description, it seems apparent that the price of the alternative instrument must be given by (1). At
the risk of being tedious, I would emphasize that the underlying mechanism assuring this result is arbitrage.
That is, if the going price, z, of the alternative instrument differs from the sum needed to be invested at Time
0 to reproduce its future cash flows, there will be arbitrage profit to be made. The profit would take the form
of a positive cash flow in some period with no offsetting negative cash flow in any period. As the description
suggests, such a pure surplus could be had by an appropriate sequence of borrowing (selling) and lending
(buying) transactions in the unit bonds, coupled with either buying or selling (depending on the direction of
the inequality) the alternative instrument.
Â
Page 384
Table 14.2
Taxable income flows from borrower to lender with symmetry
Time point 0 1 2
Time 0 unit bond r01
Time 1 unit bond r12
Alternative instrument t0 t1 t2
Discount bond 1 - z02
311
312
Table 14.3
Demand (lender) and supply (borrower) prices for a financial instrument
Price at which lender would be willing to buy
the instrument
Price at which borrower would be willing to

sell the instrument
Introducing taxes on unit bonds, the price of the alternative instrument depends upon the taxable income
flows associated with holding it. Table 14.2 lays out schematically the taxable income associated with the
unit bonds as well as an arbitrary sequence of taxable income flows (denoted t1) that, hypothetically, might be
attached to the alternative instrument specified in Table 14.1. Also shown is the taxable income flow for the
OID bond under the old rules: no tax until realization at Time 2, at which point taxable income is the
difference between basis and payoff.
Readers will be very familiar with the idea that after-tax cash flows are discounted at the after-tax discount
rate by taxable investors. This is the essence of the story spelled out in Table 14.3.
In the expressions in the table, the symbol ml stands for the marginal tax rate of the lender, and mb for the
marginal tax rate of the borrower. The first row of Table 14.3 shows the condition on the price z1 of the
alternative financial instrument specified in Table 14.1 at which a lender (the demander of the instrument)
would find it an interesting proposition when the applicable tax consequences are as
Â
Page 385
shown in Table 14.2. The second row of Table 14.3 shows the condition on the price zb of the alternative
financial instrument at which a borrower (the supplier of the instrument) would find it an interesting
proposition.
A low price for the instrument means a high yield. The equality case of the condition in both rows shows the
point of indifference between putting money into the instrument and, instead, putting money into some
combination of unit bond purchases and sales. For the lender who is buying the instrument, any price is
attractive that is below a critical break-even level that depends on the lender's marginal tax rate. For the
borrower, who is selling the instrument, the inequality runs in the other direction. An attractive price is one
that is higher than a break-even level that depends on the borrower's marginal tax rate. Again, the conditions
in Table 14.3 are based on arbitrage considerations. Thus, for example, if the price is strictly greater than the
right side value in the second row, there is an arbitrage profit to be made by a taxpayer with marginal tax rate
mb by borrowing in the form of selling the instrument for a relatively high price, and using the proceeds to
lend in the unit bond market, to finance the future payoff on the alternative instrument (taking into account
the associated tax consequences).
If m is the single applicable marginal tax rate, there will be exactly one value of the instrument satisfying
both conditions specified in Table 14.3. The opportunity for arbitrage profit for both borrowers and lenders is
eliminated simultaneously by the price given in expression (2), which corresponds to (1) with taxes taken into
account.
The fact that there is a single tax rate implies that the discount factors (the factors multiplying the cash flows)
on the right side of (2) are the same for everyone, and the fact of symmetry in the taxable income flows
312
313
means that the amount added to or subtracted from the after-tax cash flows as a result of the tax is the same
for everyone. Any positive amount of tax that has to be paid by the lender will be reflected in a lower price
that the lender would be willing to pay for the instrument, that is, a lower amount that can be realized by the
borrower, who is exactly compensated by the corresponding deductions.
Â
Page 386
2
What if Demand and Supply Prices Differ?
With different marginal tax rates, the critical values of the demand and supply prices, the prices that would
render demanders and suppliers indifferent between the alternative instrument and the package of unit bonds
that reproduces the future cash flows, generally will differ. There are two possibilities: First, the maximum
price the lender would be willing to pay for the instrument could be below the minimum price the borrower
would be willing to accept. In this case, the transaction in question would not be observed. The same
effective transaction would be carried out using unit bonds.
The second possibility is that the maximum price the lender would be willing to pay for the instrument
exceeds the minimum price the borrower would accept. Here capital market equilibrium does not exist in the
model world. Translated into application, it means that there would be no natural limit on the tax arbitrage
profit to be made at a price somewhere between the two limits. The lender would sell Time 0 unit bonds to
buy the alternative instrument and the borrower would buy Time 0 unit bonds with the proceeds of selling the
alternative instrument.
These two possibilities are not independent. If there is a transaction that would not take place between a
borrower in one tax bracket and a lender in the other (because it is unattractive to both sides), the reverse
transaction will provide an opportunity for tax arbitrage profit.
Price adjustments cannot eliminate this profit potential. Either the marginal tax rates must be equated, or
linearity of the tax rules must be sacrificed (for example, incorporating limits on the allowable loss that can
be claimed by a taxpayer). 34
As the algebra suggests, if there is any potential for arbitrage profit, the absolute amount of profit is
proportional to the scale of the transactions. Since the pure arbitrage transaction involves no actual cash
changing hands (except with the tax collector), there is no natural limit. As a practical matter, however, the
transactions are not costless, so the quantitative importance of the potential for arbitrage profit depends
sensitively upon the degree of difference in marginal tax rates and on the interest rate.
3
The Case of a Zero Coupon Bond
The taxation of cash basis issuers and holders of a zero coupon bond under the old rules provides a
convenient illustration of these ideas.
Â
313
314
Page 387
Under this rule for taxable income, the market clearing price of the OID bond implied by (2) is (3).
Some algebra translates

condition (3) into (4), which
should have a familiar look.
The discount factor on the right
side is modified from the no tax
case by terms that incorporate
the advantage to the lender of
the deferral of tax, from Time 1
to Time 2, on the implicit yield
between Time 0 and Time 1. If
the marginal tax rate is zero,
the right side reduces, as
expected, to the no-tax
expression. A positive marginal
tax rate raises the equilibrium
price of the instrument, the
amount the lender would be
willing to give and the
borrower required to be paid
for the future cash and taxable
income flows. (It is not clear
from the derivation, but is the
case, that the tax rate in
question in this example is the
one that applies at Time 1.)
4
Under General Conditions, Symmetry is Necessary
As has been demonstrated, with a single tax rate, symmetry of the rule determining the income of the party
and counterparty of a financial instrument is sufficient for any disparity from the treatment of the unit bonds
to be absorbed completely in the price of the instrument. There is no need to provide special rules to limit tax
arbitrage profit. (Note that this statement requires that there really be just one rate. This condition would be
violated by, for example, the presence of tax-exempt participants in a market where taxable persons or
entities all have some positive marginal rate.)
A natural question is whether symmetry is also necessary. That is, would a rule that violated symmetry
necessarily result in an unlimited opportunity for tax arbitrage profit? The answer is yes if attention is
confined to the case in which the sum of the deductions from taxable income by one party over the life of the
instrument equals the sum of inclusions in taxable income by the counterparty. (Two additional conditions
are taken for granted: interest rates and the single marginal tax rate are positive.)
314
315

Page 388
Here is the argument. Suppose the tax rules do not have the symmetry property with respect to some
particular position. For example, let the offending case involve a deduction for the "borrower" side of the
position (I use the term just to identify one of the counterparties), that occurs before the inclusion in the
income of the lender. This would describe the former treatment of the OID bond issued by an accrual basis
taxpayer to a cash basis taxpayer in the same rate bracket. To be concrete, suppose that the borrower in a unit
position of this financial instrument gets to deduct $100 at some point, but that the lender does not have to
include the $100 until two years later. Now suppose a person takes both sides of this financial instrument,
being both borrower and lender. There is no net cash flow involved apart from taxes. But by virtue of being
the borrower, the taxpayer gets a deduction of $100 at some point, which is offset by an inclusion of the same
amount two years later. If there is a positive after-tax interest rate, this change in the timing of tax liabilities is
valuable. By further transactions in the unit bonds, the taxpayer could arrange for a cash flow sequence that is
positive at some time point and never negative.
There is no way to eliminate this tax advantage through adjustment in the price at which the position is
exchanged, since it results from holding simultaneously opposite sides of the same instrument.
5
Current Realization Approach to Capital Gains Taxation Fails the Symmetry Test
The discussion of the zero coupon bond, showing that capital markets could incorporate taxation at maturity
on a capital gains basis, assumed that both issuer and purchaser of the instrument held the instrument to
maturity. Barring limits on capital losses, however, it generally is advantageous for one side or the other of
this transaction to accelerate the realization of a loss at Time 1. The tax treatment of the position held to
maturity, while formally correct, is therefore of limited relevance. The question is, whether the tax that results
from optimized realization behavior can be incorporated into the market price (always given a single
applicable marginal tax rate).
Examination of the sequence of

transactions in Table 14.4
establishes that the taxable income
flows that result when taxpayers
advantageously time their
realizations fail the test of
symmetry. It is possible to
establish zero-market value
positions that shift tax liability
toward the future. Table 14.4 lays
out the taxable income flow to
Â
315
316
Page 389
Table 14.4
Cash and taxable income flows to borrower (taxpayer option to realize)
Time point 0 1 2
-1
Discount bond if paid off at maturity Cash z02
Taxable income - (1 -
z02)
Discount bond if transferred to another Cash (aggregate to borrower's side) z02 -1
person at time 1
Taxable income (aggregate to -(Z12 - -(1 - z12)
borrower's side) z02)
the borrower under two scenarios. In the first, the original borrower pays off the loan at maturity. In the
second, the original borrower pays someone else an amount z12, the price of the discount bond as of Time 1,
to take over the obligation. There is no net cash flow in the aggregate to the borrower's side at Time 1, since
the original borrower's outflow is matched by the inflow of the person taking over the obligation. There is,
however, an increase in the original borrower's liabilities for tax purposes amounting to z12 - z02, which can be
claimed as a deduction. The remaining unrealized loss to the borrower side of the financial instrument
appears as the deduction, 1 - z12, at Time 2. If the lender waits to realize gain until maturity, the taxable
income flows are asymmetric, and the net effect, in the aggregate, is to accelerate the deduction side, relative
to the inclusion side. Tables 14.5 and 14.6 spell out the details.
Table 14.5 describes an arbitrage sequence in that the transactions are organized to produce a net cash flow of
zero at every time point except Time 1. The "bottom line" in the table shows just such a sequence. There is an
arbitrage profit if the one nonzero result, shown in the lower right cell of the table, is positive. To bring about
the required pattern, the issuer of l/z02 units of the discount bond uses the $1 proceeds to buy a Time 0 unit
bond. The third column of Table 14.5 shows the cash flow at Time 1, when the discount bond issuer unwinds
the pair of positions. The Time 0 bond pays off (1 + r01). The discount bond is equivalent to a quantity of
Time 1 unit bonds, since there is no tax advantage to the holder looking ahead from that point. So the l/z02
units must be worth the payoff amount at Time 2, l/z02, discounted at the Time 1 unit bond rate r12. The issuer
of the
Â
Page 390
Table 14.5
Details of discount bond issuer's tax arbitrage transactions
Time point 0 1
Transaction: Issue l/z02 discount bonds Receive principal plus interest on the unit bond.
for $1.
Buy unit bond. Pay someone to take over the discount bonds, realizing a
loss.
Pay any taxes due.
Details of
transactions:
Time 0 unit bond -1 1 - r01
316
317
Discount bond 1
Tax 0
Net 0
discount bond thus will have to pay someone that amount to take over the instrument, giving rise to the cash
outflow
There are, in addition, cash flows owing to the tax consequences: -mr01 (an outflow) due to the interest
received, and
(an inflow) due to the fact that the amount paid to buy out of the discount bond obligation exceeds the
amount received on its issue. The net proceeds are shown in the lower right cell of Table 14.5; the no profit
condition for the issuer, expressed algebraically by (5), is that this amount be less than or equal to zero.
Condition (5) simplifies to (6),
Â
Page 391
Table 14.6
Discount bond lender's tax arbitrage transactions (taxpayer option to realize)
Time point 0 1 2
Transaction: Buy l/z02 discount bonds for $1. Pay principal plus interest Collect 1/z02 on the dis
Sell Time 0 unit bond (borrow). on the unit bond. Pay principal plus inte
Collect tax savings from Time 1 unit bonds.
interest deducted. Pay any taxes due.
Sell Time 1 unit bonds to
cover the difference.
Details of cash flows:
Time 0 unit bond 1 -(1 + r01 )
Time 1 unit bond (1 + (1 - m)r01) -(1 + (1 - m)r01)(1 + r1
Discount bond -1 1/z02
Tax 0 mr01
317
318
Net 0 0
Â
Page 392
Table 14.6 depicts the other side of the market, showing the result of financing the holding to maturity of $1
worth of the discount bond by a sequence of borrowings in the form of unit bonds. Note that there is a fourth
column, showing the cash flows at Time 2. The transactions are designed to produce a net cash flow of 0
except at Time 2. The lower right cell of Table 14.6 contains the net result at Time 2, from which it is seen
that the no profit condition for the person who lends in the form of a discount bond (the buyer of the
instrument) is given by (7),
which simplifies to (8).
Taking (6) and (8) together, if the marginal tax rate and the first period interest rate are positive, that is, m > 0
and r01 > 0, the critical price b for the lender is greater than the critical price a for the borrower. A price for
the discount bond that simultaneously eliminates opportunities for arbitrage profit by borrower and lender
thus must satisfy the condition that it is less than itself, z02 < z02. This is impossible; no such price exists.
6
Possible Fixes for Capital Gains
The stark conclusion is that, even if there is just one marginal tax rate, some nonlinearity (such as the ceiling
on the deductibility of losses) is required to limit tax arbitrage profit when income is based on the current
realization conventions. Note that this particular conclusion has nothing to do with cherrypicking or other
risk-related phenomena, but rather turns on timing alone: The party and counterparty can choose different
realization paths, to their mutual advantage. Some other approach, satisfying the symmetry requirement,
would be required, of which the following four are examples.
a
Mark to Market
Under a mark-to-market rule, the holder of an instrument would recognize gain or loss each period by
including in
Â
318
319
Page 393
income the difference in the instrument's market value and its basis at the end of the period. Basis would be
adjusted for recognized gain and loss. Cash payments received during the period would be subtracted from
basis. If the issuer of such an instrument were treated in the same way, except that cash payments paid during
the year were added to basis, where holding one side of the position resulted in a loss, holding the other side
would result in an equal gain, thereby satisfying the symmetry property.
b
Ignore Intermediate Transactions
Another approach would be to provide that a sale of an instrument, such as the illustrative zero coupon bond,
would have no tax consequences. Basis would be transferred with the instrument. The purchaser of the
instrument at Time 1 thus would obtain with the future cash flow, the future taxable income flow that
otherwise would have gone to the initial holder. In this way, symmetry of tax treatment would be assured.
As a result, the market could incorporate any apparent mismeasurement of income into the prices of financial
instruments, provided everyone in the market faced the same tax rate. The future tax consequences for the
ultimate holder of the instrument at maturity would be discounted into the price obtained at intermediate
points. In the example, it readily is shown that the market-clearing price at issue in this case would be given
by (4) and the price at Time 1 by (9).
c
Impute Interest Currently
Another possibility would be to impute interest to the holder of a financial instrument, with a corresponding
addition to basis. 35 (The issuer would obtain an imputed deduction and corresponding basis adjustment.)
Cash payment from borrower to lender, whether representing ''interest" or "principal," would give rise to a
deduction from the basis of the lender and a corresponding adjustment for the borrower. It can be shown that
under these rules, the price a taxpayer would be willing to pay for the discount bond would be the same as in
the no-tax world. This implies that the price would be independent of the applicable tax rate (by contrast with
the previous example).
This is very close to the current treatment of a discount bond, which involves calculating a yield to maturity
and imputing annual
Â
Page 394
interest income to the lender

(and allowing a deduction to
the borrower). 36 The main
difference from the suggested
alternative is that the
adjustment required to
eliminate the potential for
arbitrage profit from selective
timing of capital gain and loss
319
320
on the instrument is the current

period's market rate, not a
yield to maturity.
The yield to maturity of a financial instrument is that uniform rate of period-by-period interest that equates
the market value of the instrument to the discounted value of its future cash flows. If the actual, known,
future period-by-period interest rates vary in the risk-free context assumed in this Section, then the yield to
maturity of an instrument with any arbitrary cash flow profile will be a kind of average of them. If, on the
other hand, the actual, known, future period-by-period interest rates are the same, the yield to maturity of an
instrument will equal that common value. It is only in this case that the present yield to maturity method of
assigning interest income and deductions forecloses profit from tax arbitrage using unit bonds. When there is
variation in the one-period interest rate, the yield to maturity approach to imputation results in a difference
between an instrument's basis and its market value, opening up the possibility for tax arbitrage profit.37
(There is an additional element involved in the term structure of interest rates in practice: Future unit bond
rates are not known, so that the price of a future dollar incorporates a bet on future interest rates, as well as
purely intertemporal value. I develop this point below.)38
In the world of perfect certainty, imputation of interest should result in exact equality between the market
value of an instrument and its adjusted basis. This precise equality would fail, in general, under the yield to
maturity approach. As a practical matter, however, by narrowing the difference between basis and market
value, the yield to maturity imputation presumably greatly narrows the potential for profit, that is, increases
the likelihood that transactions costs would swamp the gain, even if a limitation on the deductibility of capital
losses did not address the problem in another way.
d
Impute Interest on Realization
Another possibility would be to impute interest on realization. That is, at the time the instrument is sold, the
taxable income would be based, in effect, on the assumption that the proceeds of the sale were the result of a
sequence of investments in unit bonds (at each intermediate point, reinvesting the
Â
Page 395
proceeds from the previous point's investment) from the time of acquisition to the sale date. The
period-by-period interest payments thus imputed to the instrument at the time of sale would be allocated to
the intervening tax years. Interest would be charged on any taxes that would have been due, with everything
settled up at the time of sale. 39 (The tax on an instrument held to maturity would be the same as that on one
sold at that time for the payoff amount. Cash payments received during the holding period would be treated
as sales proceeds on a position held since the acquisition date of the underlying instrument.)
The only observed facts about

the transaction used in this
approach would be the sales
price of the instrument and its
acquisition date. The amount of
money that changed hands at
the time of acquisition would
not enter the calculation; the
concept of basis would not be
invoked.
320
321
The symmetrical treatment would apply to the issuer of the instrument, who would obtain, in effect, a
deferred interest deduction. Intermediate cash flows would be treated as realizations, dated from the time of
acquisition.
Table 14.7 lays out the details of this approach in the two-period context.
Note that the net-of-tax payoff on the discount bond at Time 2 (the lower right cell of Table 14.7 simplifies to
(10).
The market equilibrium price of the instrument is determined by discounting the after-tax cash flows at the
after-tax interest rates. As shown in (11), as in the case of imputing interest currently to basis (and adjusting
basis accordingly), under the suggested tax rules, the price is independent of the investor's tax rate, being
simply the discounted (at the before-tax market interest rates) value of before-tax flows.
Â
Page 396
Table 14.7
Imputing interest at maturity to discount bond lender
Time point 0 1 2
Transaction: Buy 1 discount bond Collect 1 on the discount
for z02. bond.
Pay any taxes due.
Valuation implicit in tax calculation at
time 2
Implicit interest in tax calculation at

time 2
Implicit tax on implicit interest in tax
calculation at
time 2
Tax on implicit interest
Interest on deferred implicit tax
After-tax cash flow -z02
Although I have not spelled it out in detail, it is the case that this rule results in incorporating tax effects in
asset prices even though it does not satisfy the property of symmetry (since either party to the instrument has
321
322
the option to sell at Time 1). It might be thought that this fact contradicts the conclusion that symmetry is
necessary, as well as sufficient, for tax effects to be fully incorporated in market prices of instruments.
Necessity, however, was shown to obtain for rules that result in equality between the sum of inclusions in,
and deductions from, income for the party and counterparty to the instrument. The
interest-imputation-on-realization rule may not satisfy this condition, and would not do so in the case in
which one party chose to realize at Time 1, whereas the counterparty realized at Time 2. This rule satisfies,
instead, the weaker symmetry condition, that the discounted sum of taxable inclusions equal the discounted
sum of deductions from the tax base, which is sufficient as well as necessary for the result.
Â
Page 397
7
Correctness in the Treatment of Time
With symmetry, or with its present value equivalent, and a single rate of tax, tax arbitrage profit will be
eliminated in financial market equilibrium, where the taxation of the unit bondsthe taxation of ordinary,
one-period interestis taken as the fixed element of the system. As I have already stressed, this fixed element
does not give the correct measure if by "correct" is meant real (inflation-corrected) income. 40
Getting to a correct measure would require adjusting the nominal interest receipts and payments to extract the
inflation premium.41 A relatively simple procedure for doing so would be to generalize the imputation of
interest to basis for all assets, including the ordinary one-period bonds that are the foundation stones of the
income measurement system. Income for the year would be the product of adjusted basis and the
inflation-adjusted interest rate. The annual adjustment to basis would include, in addition to the usual
increase for imputed interest and decrease for any cash received, a percentage increase equal to the change in
the price level during the period. (Basis adjustment for cash flows between the beginning and end points of
the year would require some appropriate approximation for price level change during the year.)
Given that interest imputation is required for all but the fundamental one-period instruments in any case, it
thus would be conceptually fairly simple to implement an inflation-corrected tax. If the analysis were
extended to take into account, for example, real investment, inflation correction would be an important
consideration (since tax arbitrage forces come into play in that connection as well).
In the world of a single marginal rate, taxing nominal interest involves an incorrect income measure, but need
not produce an incorrect result. That is because the market could take into account both the real and nominal
elements of the transaction. For example, the nominal interest rate should vary by 1/(1 - tax rate) per
percentage point of anticipated, steady inflation. With a 33% marginal rate, an interest rate of 3% with no
inflation, and interest rate of 18% with inflation of 10% per year would offer exactly the same real after-tax
borrowing and lending terms (18% - (33% Ã 18% in tax) = 12% after-tax nominal return. Subtracting the
10% inflation leaves a 2% real after-tax return, the same as in the no-inflation case.). With multiple tax rates,
however, it is impossible for the financial markets to produce an adjustment that works for all taxpayers.
Then incorrect income measurement produces "incorrect" real results, as well.42
Â
322
323
Page 398
B
Multiple Tax Rates and the Consistency Requirement
Leaving behind the world of the single tax rate, the constraints on the rules necessary to permit the
elimination of tax arbitrage profit in a linear income measurement system become more stringent. Symmetry
is no longer sufficient. Rather consistency of income measurement is both sufficient and, under conditions,
necessary.
Professor Shuldiner defines consistency as the "equivalent treatment by a single taxpayer of two or more
individual transactions making up parts of a larger overall transaction." 43 What I have in mind is similar. I
specify consistency as requiring that a given sequence of net cash flows give rise to the same taxable income
flows for all taxpayers, regardless of how that cash flow is composed out of one or more financial
instruments. Because it is more specific about the meaning of "equivalent treatment" and refers to all
taxpayers, it is more restrictive than Professor Shuldiner's concept. As I use the term here, it is also more
restrictive in adding symmetry, which relates to the tax consequences to the two parties to a financial
instrument. Consistency, as I use the term here, implies symmetry under my definition, but not vice versa.
Assuming that Professor Strand means to imply the treatment is universally applicable to all taxpayers, it
appears to be close to his usage, at least for the case of certainty: "A tax system is consistent if and only if
every cash flow pattern has a unique tax treatment.''44
The "old" treatment of the cash basis issuer and holder of an OID bond is an example of measurement that is
symmetrical, but not consistent with the treatment of the equivalent sequence of one-period loans. Yield to
maturity treatment of an OID bond gives an example of an effort to achieve consistency. The ideas are: (1)
that the taxpayer who has a cash flow from the date of issue of the OID bond to its maturity should have the
same set of inclusions in taxable income, regardless of whether the cash flow is the result of holding the OID
bond or a series of one-period instruments, and (2) whenever one party has an inclusion in income, the
counterparty has a deduction (subject to any limitations that would apply to a borrower paying ordinary
interest).
The sufficiency of consistency, so defined to eliminate the potential for relevant tax arbitrage profit using
financial instruments, follows almost by definition. Tax arbitrage is effected by taking a position in
Â
Page 399
two or more financial instruments so as to produce a zero net cash flow each period on a before-tax basis, but
a valuable flow of taxable income. (In the usual case, a flow of taxable income is valuable if it has a positive
net present value at after-tax discount rates; in other words, when it gives rise to an acceleration of deductions
or deferral of taxable income.) If income measurement is consistent, a combination of instruments that gives
rise to a zero net cash flow each period must bring with it the same sequence of taxable income flows as any
other such combination of instruments. But it cannot be the case that holding no financial instrument (and
therefore a zero net cash flow each period) results in a favorable flow of taxable income.
The more interesting proposition is that consistency is necessary under reasonable conditions. These
conditions are the same ones as applied in the single rate case: For rules that imply the sum of inclusions and
the sum of deductions on the two sides of a financial instrument are the same, consistency is necessary if
there is always more than one tax rate in the population of taxpayers and if interest rates are positive.
The argument is very close to the one given above for the necessity of symmetry. Start with the taxable
income flows that are associated with the unit bonds. Refer to as "the" after-tax discount rates the implied
323
324
net-of-tax rates of return from unit (one-period) bonds. If the tax rules do not have the consistency property,
there must be some instrument such that the sequence of one-period bond transactions that duplicates its cash
flow produces a different sequence of taxable income flows. Suppose the price of this other instrument
adjusts, along the lines discussed earlier, so as to render the two positions indifferent for taxpayers with a
particular tax rate. (If the price does not so adjust, there is an opportunity for tax arbitrage profit for those
taxpayers.) Then the discounted sum of inclusions in the tax base of the "lender" side of the other instrument,
using the after-tax discount rate, must be the same as the discounted sum of deductions from the tax base of
the "borrower,'' using the same after-tax discount rates. At a different tax rate, the after-tax discount rates will
differ. But two different sequences of cash flows that have the same discounted value at one set of discount
rates will have different discounted values at another set of discount rates. It is impossible to eliminate the
potential for arbitrage profit without giving up the linearity of the tax (or setting all marginal tax rates equal).
Â
Page 400
1
OID Rules Illustrate the (Strictly Speaking, Unsuccessful) Quest for Consistency
As discussed above in connection with the analysis of the single tax rate system, the current rules for
inclusion of interest on an OID bond (impute interest at the implied yield to maturity) fail the symmetry test
when combined with the option to realize gain or loss. Restricting attention to OID bonds held to maturity by
both counterparties, there is symmetry but still, in general, a failure of consistency. Except when the
one-period interest rates are all the same, the sequence of transactions in unit bonds that reproduces the OID
cash flows will generate a different path of taxable income flows. Whereas this difference in tax results can
be built into the price of the instrument when there is a single tax rate, when there are multiple tax rates, there
always will be an opening for tax arbitrage profit through transactions between a pair of taxpayers with
different marginal rates. 45 (To be sure, the opening may be very small. Transactions costs typically will
swamp it.)
2
Current Realization Approach to Capital Gains Taxation Fails the Consistency Test
Since the current realization regime for capital gains fails the symmetry test, a fortiori it fails the consistency
test. Some of the techniques, discussed in the single rate context, that produce symmetric treatment require
modification to produce consistency.
a
Mark to Market
Since mark-to-market taxation produces exactly the same path of taxable income as results from the
duplicating sequence of unit bond transactions, it satisfies consistency.
b
Ignoring intermediate transactions in an instrument between issue and maturity (with basis carried over),
which guaranteed symmetry, generally would not produce consistency. This method assures that the path of
deductions by the "borrower" side of an instrument is equal and opposite to the path of inclusions by the
"lender" side, regardless of intervening sales by either side. The problem, however, is that this symmetric
result is not the same as the path of taxable income associated with the duplicating sequence of unit bond
transactions.
324
325
c
The technique of imputing interest currently would produce consistency. It is important that the interest
Â
Page 401
imputed be identical to that obtaining on the unit bonds, as this is how the potential for tax arbitrage is
eliminated. This could be effected by taxing the unit bonds themselves on an imputed interest basis, in which
case, the applicable rate could be chosen as a matter of policy (for example, to equal the inflation-corrected
rate).
d
As noted, the technique of imputing interest on realization fails the strict symmetry test, but satisfies an
appropriate generalization in the case of a single rate system. Roughly the same conclusion applies in the
multiple rate system to a modified specification of this approach. To achieve the effect of consistency, the
look back that treats the seller of an instrument as having accrued interest at the market rate would have to
apply the seller's marginal tax rates (both for the implied periodic taxes and for the effect of any acceleration
or deferral of tax payments). Then the discounted value of the taxable income flows (at the after-tax discount
rates applicable for the taxpayer in question) will be the same as that of the duplicating unit bond
transactions. This equality is the generalization of the consistency notion that is necessary to eliminate the
potential for relevant arbitrage profit with multiple marginal tax rates.
Using, instead, a single

marginal tax rate, rather than
the seller's actual marginal tax
rates through time, to infer the
factor to apply to the proceeds
of sale would fail the
consistency test.
IV
The Taxation of Risky Instruments
A
Pure Risk Effects: Timeless Gambles
Perhaps because uncertainty is inextricably bound up with sequence (before and after its resolution), one
tends to think of risk and inter-temporal return together. Thus, for example, if I enter into a forward contract
to purchase some asset at a specified price on a specified date, in the typical situation, the price is so chosen
that the value of my position at the outset is zero. Immediately thereafter, however, as information about the
future begins to unfold, my position acquires value. Whether the value is positive or negative depends upon
whether the new information implies the future price is more likely to be higher or lower than anticipated
earlier. If the former, my position acquires a positive value; if the latter, my position acquires
Â
325
326
Page 402
a negative value. The value of the position of the counterparty to my contract moves in exactly the same
amount but in the opposite direction. These effects occur with the passage of time, but time is not of the
essence in the changing values. On the contrary, if all that happens is the passage of time, and no new
information develops, the value of the position will not change at all.
Taxing pure bets between a party and a counterparty, involving sequence (before and after the resolution of
uncertainty) but not the passage of time, is less problematical than the taxation of transactions involving time.
The striking fact is that, provided the same tax rate applies to the gains and losses from a risky position, the
tax rate itself is of no fundamental importance. The reason is simple: The tax has the effect of reducing the
scale of the bet, but not the relative payoffs under the various relevant contingencies. Since transactors have
the option of varying the scale of their bets without recourse to the tax system, taxation does not affect the set
of possible bilateral bets at all.
To illustrate, by analogy with the

choice between a series of
one-period loans and an OID
bond, suppose there were two
kinds of bets, taxable and not,
with the choice left up to the
transactors. For example, suppose
gains and losses from flipping
pennies are taxable if the bet is
governed by a contract written in
red ink, but not if it is governed
by a contract written in green ink.
The green ink deals get the wrong
result from the point of view of
proper income accounting, but
one cannot produce tax arbitrage
profit consequences.
The analogy to the discount case would involve something like my writing a red contract betting on heads
and a green contract betting on tails for the same coin flip (thereby eliminating any actual risk), and the
counterparty doing the opposite. It is true that this would convert an apparently riskless situation for me into a
genuine gamble, because I will either pay tax or collect a refund, depending on the coin flip. My counterparty
on the red contract will experience the opposite taxable income outcomes. But there is no opportunity for
arbitrage profit at the expense of the tax system, so long as the marginal rates of the transactors are taken as
given. All that is required is that the applicable tax rates be specified in advance of entering into the bet.
Allowing the parties to specify the fraction of gains (and losses) to be included in (deducted from) taxable
income would not enable
Â
326
327
Page 403
Table 14.8
Illustrative risky instruments: cash flows to lender
2
Time point contingency: 0 1 Heads Tails
Time 1 unit bond -1 1 + r12 1 4 + r12
Risky instrument H 1 0
Risky instrument T 0 1
them to obtain an arbitrage profit at the expense of the tax system. This result holds even if the inclusion
fraction is allowed to differ for the party and counterparty. The essential requirement is that the treatment of
gains and losses for a given party be the same and specified before the uncertainty is resolved. (Clearly, if I
can decide, after the fact, whether my position is governed by a red or green contract, I can profit at the
expense of the tax system.)
The tax arbitrage at the heart of the analysis in this chapter is thus a matter of time, and not risk. The question
is what this insight implies for the taxation of complex financial instruments.
B
Putting Time and Risk Together
Most financial instruments have both intertemporal and risk aspects. An option is a good example. So too,
however, is a long-term bond. Even if there is no risk of default, there is a risk of changes in the short-term
rates of interest. Thus, the value of a long-term bond between issue date and maturity is risky. 46
The two-period framework can be used to consider the possibilities. Table 14.8 illustrates a pair of typical
risky instruments, absent taxes. One of the instruments pays the holder 1 at Times 2 if a coin comes up heads.
The other instrument pays the holder 1 at Times 2 if the coin comes up tails. The prices of the instruments,
determined in the market, are denoted and , respectively, where the subscript indicates the time point at
which the value is determined.
It is instructive to consider the effect of variations in the time point at which the uncertainty is resolved. One
possibility is that the uncertainty is resolved (the coin is flipped) immediately after the instrument is created
and sold, at Time 0. In that case, the positive
Â
Page 404
payoffs of the two primitive bets offered in the financial market are, in effect, OID bonds at Time 0. (The
instrument is worthless if the coin comes up on the wrong side.) The price of an instrument, say the heads
bet, will jump immediately from to either zero or 1/(1 + r )(1 + r ), depending on the outcome. In either
case, the price of the instrument subsequently will appreciate01at the risk-free
12
unit bond interest rates until
Time 2.
A second possibility is that the uncertainty is resolved at Time 1. Just before that moment, each instrument
will be identical to what it was at Time 0, except that the payoff moment will have come closer. Its value at
that moment simply will be its initial value x (1 + Time 0 unit bond interest rate). Just after that moment, it
327
328
will be known which payoff will occur. It is converted into either zero or a one-period unit discount bond.
The third possibility is that the uncertainty is resolved at Time 2. Just before that point, each instrument will
be indistinguishable from its issue date except that the time of payoff will have arrived. The value of the
heads claim, for example, will be . Upon resolution of the uncertainty, its value will jump
to 1 or 0.
1
Tax Rules to Prevent Unlimited Tax Arbitrage Profit
The possible time paths of the value of the alternative instrument under different scenarios about the timing
of the resolution of uncertainty suggest what is required to preclude tax arbitrage profit. To test for the
possibility of arbitrage profit, the required step is to hedge away the risk (as in the case of the put-call parity
theorem) and determine whether a proposed treatment will eliminate the tax arbitrage potential from the
resulting pure intertemporal transaction. Here risk is eliminated by buying one each of the risky instruments.
One overarching proposition follows immediately from the analysis of the risk-flee case: Barring appropriate
adjustments to basis, if the rules are linear, taxpayer option to choose the time to recognize gain or loss will
imply the potential for unlimited tax arbitrage profit. Risk is not needed to establish this conclusion, although
the presence of risk renders difficult fixes based on the market value of the instrument.
The various approaches to deal

with the problem of gain
recognition may be divided, as
in the risk-free analysis, into
the single and multiple tax rate
regimes.
Â
Page 405
a
One Marginal Tax Rate for All Transactors
Mark-To-Market
Mark-to-market income measurement precludes tax arbitrage profit.
Carrying basis along until maturity of the original instrument should work as well in the presence of risk as in
the certainty case. The price of the instrument will reflect the tax advantage of deferral to the lender and its
disadvantage to the borrower.
To illustrate, suppose that in the case displayed in Table 14.8, a lender purchases a unit of risky instrument H
and that the contingency is resolved, with "tails" as the outcome revealed, immediately after purchase, but
still at Time 0. The lender's basis is . After the outcome is revealed, the instrument is worth 0. If the lender
holds the instrument to maturity, the payoff after tax will be, where m is the marginal tax rate,
. By the same argument used in the risk-free analysis, this payoff will be worth
in the market at Time 0, after the resolution of the uncertainty. If, alternatively, the lender sells the asset at
Time 0, this will be the amount realized, because the basis in the asset will be conveyed to the purchaser and
328
329
deducted at the maturity of the instrument. Since the net present value (at after-tax discount rates) of the two
outcomes is the same, there is no payoff (or penalty) to early realization of losses.
In applying the technique of imputing interest to the holder of an OID bond in the world of certainty, the
payment by the borrower to the lender at maturity gives rise to a reduction in both the borrower's and lender's
basis to zero, thereby bringing the imputation to an end. These payments are not reflected otherwise in
taxable income. There is no current deduction of the payment by the borrower, nor inclusion of amounts
received by the lender.
In the world of risk, there is, in addition, settlement of a bet involved in the payments from borrower to
lender. Under the usual treatment of bets, the payment from loser to winner is currently deducted by the loser
and included by the winner. The question is how one can separate the two (other than by marking to market).
Â
Page 406
A possible answer is to wait until maturity, along the way imputing, and taxing, interest on basis (with basis
adjustment). At settlement time, any amount paid by the borrower in excess of adjusted basis would be
deducted by the borrower and included by the lender. Any shortfall of the settlement amount from basis
would be deducted by the lender (as a loss) and included by the borrower.
This relatively straightforward approach would work out well in a world in which the instrument is held to
maturity. As in the risk-free analysis, the effect is to eliminate the tax rate from the determination of the value
before maturity. There are two problems presented by sale of the instrument at an intermediate point. One is
the potential for shifting gains to low marginal rate holders and losses to high marginal rate holders. This
does not arise in the one rate case. The second problem is the potential for cherrypicking, which arises
because the information about winners and losers generally emerges before the payoff date.
To illustrate, consider the first scenario described above, in which the coin flip takes place at Time 0. For
specificity, focus on instrument H. The outcome of the coin flip is to convert it into one of two risk-free
discount bonds, one paying off 1 at Time 2, and one paying off 0. Immediately following the issue of the
instrument, the coin is flipped and there is a jump in its market value, up or down, from the issue price to
whatever the value of those discount bonds may be when account is taken of the tax consequences to the
seller and buyer.
In what might be regarded as the natural procedure, the OID bonds would be taxed to an acquiror like any
other bond with the same terms, and the seller of the instrument would take into income the proceeds of the
sale, deducting basis (and thereby reducing to zero the amount on which interest is imputed to the original
lender). Since the bond-equivalent will be treated by the new "lender" exactly like a newly issued bond, its
market value will have to be either 1/(1 + r01)(1 + r12) or 0, depending upon the resolution of uncertainty at
Time 0. Interest imputation based on the issue price would continue to be deducted by the borrower (with
adjustments to basis). Interest imputed to the new lender would be higher or lower, depending on the payoff.
At maturity, the payments of 1 or 0, as appropriate, would be made. For the new lender, the payoff will just
equal adjusted basis, so there will be no tax consequences at payoff time (other than elimination of basis and
its associated interest
Â
329
330
Page 407
imputation). For the borrower, the payoff amount will exceed basis in the high-payoff case, resulting in a
deduction. In the low-payoff case the payoff amount will be 0, below basis, resulting in a taxable gain.
The question is, would selling the instrument after the uncertainty is resolved produce a tax advantage?
Would it pay, for example, to sell a losing position, thereby accelerating the deduction of basis? The issue
can be posed in general terms by supposing that the outcome after resolution of uncertainty at Time 0 is some
arbitrary amount, a(1 + r01)(1 + r12), payable with certainty at Time 2. Immediate sale for a by the original
lender would give the seller a net payoff of , after taking into account deduction of basis.
Holding to maturity would imply a larger basis deduction at that time, worth , but at the
price of paying tax on imputed interest in the interim, amounting to at Time 1 and at
Time 2. The extra basis deduction exactly compensates for the extra tax along the way, but the gross sale
proceeds grow to a(1 + r01)(1 + r12), that is, on before-tax terms, resulting in a payoff to deferral whenever a is
positive.
An alternative treatment would be to require carryover of basis until maturity, much as under the "ignore
intermediate transactions" approach, but with the addition of interest imputation to basis. The purchaser
would be obliged to carry the basis, with its imputed interest, and would take the consequences at maturity of
the gain or loss in settlement with the issuer. In other words, the purchaser would step into the shoes of the
original lender. The further new twist is that any excess of the amount paid by an acquiror over the basis
carried over would be treated as settlement of a bet, deducted by the acquiror and included by the seller.
Because the tax rate is the same for all participants, the result is that the net flow of taxes associated with the
instruments in the market would be independent of intermediate transactions in those instruments. The
outcome would be the same as under obligatory holding to maturity.
Unlike the case of imputation of interest under certainty, which resulted in the market prices of all
instruments being independent of the tax rate, under this generalization of the approach to the case of risk,
"the" tax rate would manifest itself in the market prices of assets for which gain or loss has been determined.
Imputing interest on realization also should work as well in the risky as in the risk-free setting. This
Â
Page 408
is the essence of the Auerbach analysis. 47 Recall that basis is dispensed with in this approach. Instead, the
actual payoff is treated as having been accumulated at the risk-free rate from the acquisition date to the
realization point. Typically, the result would be a lower tax than under a mark-to-market regime on the holder
of the winning risky instrument (who is not taxed on the gain, but simply on the implied past interest), offset
by higher taxes on the loser. This becomes clear when it is noted that, for an instrument with any positive
value, no matter how far below basis, the loser would owe tax on realization.
b
The Multiple Tax Rate World
In the multiple tax rate world without risk, a stronger condition than symmetry, namely consistency,
generally was required to preclude tax arbitrage profit. The general conclusion is that those techniques for
dealing with the realization problem for capital gains that give rise to the independence of the market value of
an instrument of the single rate of tax in the one rate world continue to work out in the multiple tax rate
world, even with risk.
330
331
Of the techniques considered thus far, only mark-to-market accounting and the Auerbach method of
retrospective allocation of gain pass the test. Ignoring intermediate transactions and imputing interest
currently work only with a uniform tax rate, because under these methods, the tax rate affects equilibrium
asset prices. The essential reason the mark-to-market and Auerbach methods work is that they eliminate any
timing advantage in the risk-free case and they imply that the same effective rate of tax will apply to gains
and losses due to the resolution of contingency.
V
A General Procedure for Taxing Financial Instruments
The conclusion of the discussion above is that, with zero transactions cost and multiple marginal rates of tax
(and positive interest rates), the only approaches thus far suggested that would permit a linear tax are the
mark-to-market and Auerbach look-back methods. Of these, only the Auerbach look-back method is based on
realization. If it is true, as is generally accepted, that the mark-to-market approach is practically infeasible,
the implication would be that the Auerbach look-back system is the only practically feasible approach thus
far discovered that will "work." The question is: Is it the only such approach, period? Is it necessary as well
as sufficient?
Â
Page 409
The answer is no. As argued above, risk presents ''no problem," provided the tax rate that applies is specified
in advance of its resolution. The difficulty with cherrypicking is that it allows the parties to a financial
instrument, in effect, to decide after the fact what effective tax rate will apply. They chose a high tax rate to
apply to losers (deduction) and a low tax rate to apply to winners (deferral).
A system that will work must (1) result in consistent taxation of pure timing and (2) establish in advance the
effective tax rate that will apply to any taxpayer's gain or loss due to the resolution of contingency. If the
parties are to have a choice of when to realize, the trick is to find a way to make the parties commit to an
effective tax rate in advance.
A
The Case of a Risk-Free Term Structure
To develop rules sufficient to address both problems, it will be convenient to start with the assumption that
the future short-term interest rates are known with certainty in advance. Other returns may be risky, but the
short-term interest rate at each date is predictable with certainty. In that case, the term structure of interest
rates (the implicit rates of return on zero-coupon bands with different maturities) would fully inform about
the path of future short-term rates.
1
The Proposed Approach When There is No Cash Flow Before Realization
The first step under the proposed approach is to specify at the time a taxpayer takes a position in a financial
instrument (at the writing of the contract or on acquiring the instrument by purchase), a gain reference date
("GRD") And a gain tax rate ("GTR"). (Although I call these the "gain" rate and date, it is intended that they
would apply to losses as well; "gain" might be negative.) The specification of GTR and GRD is arbitrary, and
they would not need to be the same for the party and counterparty to an instrument. The GRD and GTR could
be specified by the tax authorities. They even could be chosen freely by the taxpayer. In practice, one might
want to render these foreign ideas more digestible by restricting the specification in some way. For example,
the GTR might be the investor's marginal rate in the preceding year; the GRD might be specified as the date
of acquisition of the instrument. The key requirement is that the date and rate be out of the taxpayer's control
331
332
before new information about the instrument comes in. The GRD and the GTR would be used only for
Â
Page 410
determining that party's tax consequences of gains and losses on the instrument.
Up to the GRD, a taxpayer who acquired an instrument would be taxed on interest imputed to basis (with a
basis adjustment for the imputed amount), exactly as discussed above in the case of the world without risk. A
taxpayer with a net liability (the "borrower") would obtain a deduction for imputed interest and a
corresponding basis adjustment. The new element is that there would be, at the GRD, a special one-time
adjustment to the basis for imputing interest. At that point, the basis would jump to the value that would have
had to have prevailed to grow at the unit bond rates to the actual realization price at the realization date. Since
this amount would not be known until realization, this part of the interest imputation would have to be
worked out after the fact. (In the event of realization before the GRD, another procedure would have to be
used to make sure the interest imputation results would be the same as they would be if the asset had been
held to the GRD.)
The gain tax would apply as of the GRD to the jump in basis as of the GRD just described. The sale of the
instrument, regardless of when it occurred, would give rise to a tax on the seller of any gains (with a tax
rebate in the event of loss) at the GTR, as though all the gains relative to basis, adjusted for interest imputed
without taking into account the special, one-time adjustment, had occurred at the GRD and had been realized
then. Under this approach, the extent of gain would be determined by the taxpayer decision to realize, the
revealed amount of gain or loss being projected forward or backward to the GRD at the going interest rates.
The idea of this system is, I think, fairly simple, even if the details seem somewhat complex. Apparent
complexity arises from the steps needed to arrange the gain tax so that it were as if all the gain occurred at the
GRD and to implement the imputation of interest to basis. Here are rules that will do this:
(1) In all cases, interest would be imputed to basis and subjected to tax until the point of realization. Basis
would be adjusted upward by the imputed interest. (For a liability, the imputation would result in a
deduction.)
(2) If the realization were at the GRD, the excess of the realized proceeds over adjusted basis would be
subject to tax at the GTR. In the event of a loss, there would be a credit of the GTR times the loss.
Â
Page 411
(3) If the realization were after the GRD, there would be a look-back calculation. The excess of realized
proceeds over adjusted basis would be projected back to the GRD at the before-tax rate of interest. The gains
tax that would have been due at that point would be calculated by applying the GTR to the back-cast amount.
The implied tax liability as of the GRD would be brought forward with (deductible) interest and paid at the
realization point. In the event of a loss, there would be a credit instead of a positive tax.
Also, the basis to which the interest imputation would apply would be adjusted retroactively upward by the
amount of the gain as of the GRD. (It would be as though the basis had been marked to market at that point,
except that the value used would be the back-cast amount, rather than the actual market value, which is not
observed.) Thus, interest on the gain, with yearly compounding would be imputed to the years between the
332
333
GRD and the realization year. Implied changes in tax for past years would be brought forward with
(deductible) interest and paid at the realization point. (In the event of a loss, the extra interest income would
be negative; that is, there would be imputed to the years between the GRD and the realization year, a
deduction of interest on the loss, with yearly compounding.)
(4) If realization were before the GRD, the gain or loss relative to adjusted basis would be projected to the
GRD at the going rates of interest. At the GRD, the gain thus projected would be taxed at the GTR (for a loss
there would be a credit). 48 Between the realization date and the GRD, the seller is supposed to continue to
get the benefit of deferral (a negative benefit in the case of a loss). So between the realization date and the
GRD, the taxpayer would be allowed a deduction of interest imputed to the amount of the gain (adjusted each
year by the amount of interest deduction). (The idea is to offset, until the GRD, the part of the imputation of
interest to basis in a new position taken with the sales proceeds that is the result of the realization.) In the case
of a loss, there would be an annual imputed interest inclusion.
The choice of the GRD and GTR together would determine the effective rate of tax on gains and losses. For
any given date of realization, the variation in the GRD would alter the taxation of gains and losses because
the rules would adjust the basis to which interest would be imputed as of the GRD. Using continuous
discounting, and assuming a constant marginal tax rate on interest and a constant interest rate on the
continuous equivalent of the unit bonds, the tax
Â
Page 412
payable at realization date s on an extra dollar's worth of sales proceeds would be 1 - (1 - GTR)e-mr(s-GRD).
Given the realization date s, an increase in the GRD would make this effective tax rate on gains smaller. 49
(For further details on the continuous discounting version of the rules, see the Appendix.)
2
An Illustration of the Proposed Approach
To take an illustration, suppose the tax rate is 30%, the going interest rate is 10%, and T invested $1,000 in a
share of stock this morning. T established a GRD of one year from now, and a GTR of 50%. By this
afternoon, T's stock has jumped in value to $1,200. T knows that from here on, the stock will yield with
certainty the going rate of return on risk-free bonds, with no cash dividends.50 Consider the choice of when to
sell.
(1) If T waits one year, his asset will be worth $1,320. His basis would be adjusted up to $1,100, at a cost to T
of $30 in tax on the imputed interest. The gain calculated at that time would be $220, on which T would own
$110. T would be left with $1,320 - $30 - $110 = $1,180.
(2) If T waits an extra year to Time 2, T would pay tax at the one-year point of $30 on imputed interest. In
order to get everything to one point in time, assume T borrows the $30. His position at the end of the second
year would be an asset worth $1,452 and a liability of $33 (bringing with it a $3 interest deduction). The basis
of the asset would be $1,210($1,000 + $100 + $110) and T would owe tax of 30% of the second year's
imputed interest of $110.
T's gain of $1,452 less $1,210, or $242, would be projected back at the unit bond rate of 10% one year to the
GRD. So T would be treated as having a gain of $220 at the GRD. As a consequence of the back-projected
gain of $220 at the GRD, T would have a $110 gain tax liability as of that point. The $110 gain tax would be
treated as a loan from the government. T would pay it at the two-year point, plus deductible interest. His gain
tax payment at the two-year point thus would be $121, of which $11 would be deductible interest.
333
334
The extra $220 T would be treated as having realized at the GRD also would imply a higher basis, by $220,
for interest imputation from that time forward (one year in this case). So at the two-year point, T would have
total imputed interest received of 10% of $1,100 plus $220, or $132, total interest to pay of $3 (on the money
T bor-
Â
Page 413
rows to pay the tax at Time 1) plus $11 (on the money he would be treated as having borrowed to pay the tax
on gain at the GRD); since the interest paid would be deductible, T would have a net income tax liability of
30% of $132 - $14 = $118, or $35.40.
T's position at the two-year point would be $1,452 (from selling the asset) - $121 (tax on gain plus interest for
the deferral of one year since the GRD) - $33 (repay, with interest, the money borrowed to pay the income
tax at Time 1) - $35.40 (income tax at Time 2) = $1,262.60. This is the same position T would be in as under
the other scenarios: $1,180 Ã 1.07% = $1,262.60.
(3) If T realizes now, he would realize $1,200, for a gain of $200 over his acquisition cost, which is also his
basis. For tax purposes, he would be treated as though the gain (inflated by the interest rate) actually occurred
as a jump a year from now. In the meantime, T would qualify for an interest deduction on $200, and would
earn interest on the $1,200 realized proceeds. So, at year end, T would have $1,320, including $120 in interest
earned on the $1,200. There would be an imputed interest deduction of $20, so, for tax purposes, T would
have net interest income of $100, on which he would owe $30 in income tax. In addition, his gain would be
calculated as $200 times 110%, resulting in $110 in gains tax. The net proceeds would be the same as if he
waited to the GRD to realize: $1,320 - $30 - $110 = $1,180.
If, in this illustrative case, T had chosen the same GTR but a GRD of two years from now, the results would
be different. As in the example, he would be indifferent whether to realize immediately or at some time
before or after the GRD. But the amount at the two-year point that would be common to his projections of the
consequences of realizing at different dates would be $1,452 (from selling the asset) - $121 (tax on the gain
of $242) - $33 (repay, with interest, the money borrowed to pay the income tax at Time 1) - $32.10 (income
tax at Time 2 on $110 in interest imputed to the adjusted basis at Time 1 of $1,100 less $3 in interest paid on
borrowing) = $1,265.90 instead of $1,262.60. The reason for the difference is that T would save on extra
basis of $110 that would have come into the interest imputation calculation at the one-year point if that had
been his chosen GRD.
The choice of GRD and GTR thus does "matter" ex post. In the event of a gain, the outcome with a two-year
GRD would be better than with a one-year GRD. Similarly, in the event of a gain, a lower
Â
Page 414
GTR would be better than a higher GTR. But these pluses are balanced exactly by the minuses in the event of
a loss. Then a later GRD is worse and a lower GTR is worse.
To spell out how a loss would be handled, take the original fact situation (income tax rate of 30%, interest
rate of 10%, $1,000 invested in a share of stock this morning, GRD one year from now, and a GTR of 50%),
except the share falls to $800 in the course of the day. Suppose T sold immediately. Under the rules, interest
still would be imputed for one year on $1,000, but T would get a deduction for interest imputed on $800, for
334
335
net imputed interest of $20. The $800, in the meantime, earned $80 in interest, so T would owe tax of $30 on
the sum of imputed and actual interest income. Basis and sales proceeds would be projected to $1,100 and
$880, respectively, so T's loss would be treated as $220 at Time 1, saving him $110 in tax. His net position
would be $880 (proceeds of sale of stock plus interest) -$30 (tax on interest) +$110 (tax saving due to capital
loss, assumed refundable or otherwise usable to offset other taxes) = $960.
If T waits a year to realize, he could anticipate the asset being worth $880. T would owe $30 in interest
imputed to his basis of $1,000, which would be adjusted to $1,100. T would have a loss of $220, qualifying
him for $110 in tax savings. His net position would be $880 (proceeds of sale of stock) -$30 (tax on imputed
interest) +$110 (tax saving due to capital loss, assumed refundable or otherwise usable to offset other taxes) =
$960.
If T instead had chosen the date of acquisition of the asset as the GRD, T would have a tax saving at the point
of $100 in the case of immediate realization of the loss. T would have $900. A year from now T would have
$990 less tax of $27 on $90, or $963 (that is, more than with the later GRD). In the case of waiting a year, T
could anticipate the asset being worth $880. The loss would be back cast to $200 at the acquisition date; the
tax saving of $100 would be brought forward with (taxable) interest to Year 1. T also would have an imputed
interest deduction on $200. His net position would be $880 (sales proceeds) +$110 (capital gains tax saving
due to loss, plus interest) -$30 (tax on imputed interest on original basis) -$3 (tax on interest paid on capital
gains tax saving) +$6 (tax saving due to back-cast adjustment in basis for imputation) = $963. Similarly,
choosing a higher GTR would turn out, after the fact of a loss, to work to T's advantage.
Â
Page 415
3
Comments on the Proposed Approach
Both new elements of the proposed approach, the GTR and the GRD, serve the same function: to specify in
advance of the resolution of uncertainty, the effective rate of tax that would apply to the "bet" aspect of an
instrument with contingent returns. (The treatment of the intertemporal aspect would be dealt with by the
imputation of interest to basis and interest charges for any postponement of tax payments.) The GTR
obviously would serve this function. But the GRD also would play a role in determining the effective taxation
on gains and losses. The farther the GRD is in the future, the greater the extent of deferral advantage
conferred on winnings (and disadvantage on losses).
Given the option to set the GTR, the option to set the GRD is essentially cosmetic. Considerations of
simplicity and the risk of non-arm's length circumstances might argue for setting the GRD at zero (relative to
acquisition). This implicitly treats all gain as occurring at the moment after acquisition of the asset. A GRD
of zero eliminates any problem of projecting gain or loss from a pre-GRD realization point.
The policy interest in the GTR and GRD is an open question that invites further exploration. There is no
obvious case for allowing the taxpayer free choice. I have suggested free choice as an option simply to
emphasize the arbitrary role played by these parameters. If there is the possibility of non-arm's length terms
of financial instruments, it might be desirable to impose a requirement of the same GTR and GRD on both
sides of a bilateral transaction. (A simple way to do this would be to specify a uniform GTR and GRD at the
acquisition date.)
A similar observation applies to the possible extension of the proposed approach to positions other than in
bilateral financial instruments. The owner of a small business or piece of real estate, for example, perhaps is
best understood as obtaining a form of "labor income" in gains realized from the enterprise or other property.
In such cases, policy might imply setting the GTR in some appropriate relationship (for example, equality)
with the rate of tax on compensation.
335
336
I have argued that the proposed approach is sufficient to satisfy the requirements of a linear income
measurement system. An open question is whether, among realization methods, the proposed
Â
Page 416
approach is necessary, that is, encompasses all the possibilities. Is there another realization method that works
and cannot be described in terms of the proposed approach's GRD and GTR? I believe the answer is no, but I
have not proved it.
An interesting instance of this question is the connection of the proposed approach to the Auerbach method,
which works. The Auerbach method treats the sales proceeds on realization as all due to accumulation of gain
at the going interest rate since the acquisition date. Only that gain is subject to tax (with appropriate
look-back interest charges). Although, at first sight, it may appear different (since there is no current
imputation of interest to basis), the Auerbach method is effectively a special case of the proposed approach,
where the GRD is zero and the GTR is also zero: Gains and losses would be taxed at a zero rate as though
occurring a moment after acquisition; interest would be imputed to the implied basis and taxed in retrospect.
51
4
Treatment of Intermediate Cash Flows
Any cash inflow from the "borrower" to the "lender" side of an instrument (an interest payment received, for
example) between the acquisition date and the time of realization would result in a reduction to basis. The
cash outflow (payment of interest, for example) would reduce the liability basis of the "borrower."
Sale of a fractional interest in an instrument could be handled by apportioning the basis between the interest
sold and the interest retained. Treatment of the sale of any more complex claim to the instrument (selling off
the rights to the dividends, for example) would require more thought. Probably such a transaction would be
treated best as the issue of a new instrument.
5
The Proposed Approach and Correctness
Since it relies on imputation of interest, the proposed approach suffers from the general shortcoming of
interest as a measure of income: failure to adjust for inflation. If an inflation adjustment were implemented,
or for the case of no inflation, the interest imputation would produce a correct measure of income for riskless
instruments. In fact, for such instruments, the path of taxable income would be the same as under a
mark-to-market regime.
For risky instruments, however, there would be no connection between the path of taxable income and the
path of market value of the
Â
336
337
Page 417
instrument. For one thing, there would be a one-time extra liability (or rebate) at the moment of realization
(as in the case of realization with conventional income tax rules). Furthermore, there would be no necessary
relationship between the discounted value of tax liability under the proposed system and the path of liability
under a mark-to-market regime. Nor could there be, since there is no observation of an instrument's market
value except at the moments of its acquisition and sale.
There would be, however, equivalence of the proposed method's results and those obtaining under
mark-to-market accounting. Offered a choice at the time of acquiring the asset, the taxpayer should be
indifferent between any particular specification of the proposed method (GTR and GRD) and mark-to-market
accounting. (Strictly speaking, to establish indifference with full generality requires a sufficiently full set of
contingent instruments to permit the taxpayer to take any desired position in the risk aspects of an
instrument.) The correctness thus resulting from the proposed method is akin to that produced by taxation of
nominal interest income in a single tax rate world. The outcome for the taxpayers would be the same as that
with income correctly measured for tax purposes, even though the actual flow of tax liabilities would be
different. 52
B
Extending the Analysis to Uncertain Future Short-Term Rates
As I have pointed out above, the imputation of the yield to maturity to holders of a long-term instrument, in
general, will not yield the right tax result, where "right" refers to a result that precludes arbitrage profit
opportunities. The exception is the case of short-term rates that are known to be constant. When the term
structure of interest rates results simply from the variation over time in the short-term rate, but not from
uncertainty about that variation over time, the imputation of the current short-term interest to basis (rather
than yield to maturity) produces the right result. The holder of a long-term zero-coupon bond will be fully
taxed by the imputation of interest to basis at the going rate. The adjusted basis on a discount bond will
always equal its current market value, so the effect is the same as a mark-to-market rule.
When, however, the short-term rates are themselves uncertain (even if they are expected to be constant), this,
in general, will not be the case.53 Instead, the machinery described in the previous Section
Â
Page 418
comes into play and gives rise to taxation of gains and losses on the instrument that are contingent on the
movement of short-term rates. 54
Although it is beyond the reach of this chapter to address the issues in depth, I describe in this Section the
way the proposed method of taxing financial instruments would apply to a long-term discount bond when
future short-term rates are not known. In doing so, I would reemphasize that the issue is not correctness of the
measure in some absolute sense. Rather, the objective is a linear income measurement system. The theme of
the design of rules is consistency of the taxation of all financial instruments with the taxation of the
duplicating sequence of unit bond transactions.
U.S. income taxation already makes use of standardized interest rates (the applicable federal rates) for
various purposes. These rates are identified for various terms of the instruments in question.55 As has been
shown, strictly speaking, these rates are not quite what are called for. The rates observed for contracts of any
length blend different underlying short-term rates that correspond to the simple model's unit bond's interest.
Furthermore, the future values of those short-term rates are not known with certainty and the longer-term
rates reflect that uncertainty. So they represent a blend of pure intertemporal returns and risk premia. It is for
337
338
this reason that I was a bit vague in referring, in the introductory discussion of tax arbitrage, to the sequence
of transactions in short-term bonds required to duplicate a discount bond for tax arbitrage purposes.
The story is illustrated in Table 14.9.
Here the uncertainty refers to the unknown interest rate on the Time 1 unit bond. The events, "heads" and
"tails," referred to in the columns for Times 1 and 2, are the same ones. That is, there is a
Table 14.9
Term structure illustrated
1 2
Time point 0 Heads Tails Heads Tails

contingency:
Time 0 unit bond -1 1 + r01 1 + r01
Time 1 unit bond -1 -1
Discount bond -z02 1 1

Risky instrument H 1
Risky instrument T 1
Â
Page 419
single coin flip, at Time 1, that determines the short-term interest rate that will prevail between Time 1 and
Time 2. The two explicitly risky instruments shown in the table pay off at Time 1 according to the same coin
flip that determines the interest rate on the Time 1 unit bond. (They can be thought of as options at Time 0 on
the Time 1 unit bond interest rate.) Also shown is a discount bond that pays 1 at Time 2, regardless of the
outcome of the coin flip. Its payoff is not risky, but its value at Time 1 is, because that value depends on the
coin flip that determines the short-term interest rate then. At Time 0, the newspapers would report r01 as the
interest on a bond with one-year maturity, and the yield to maturity of the discount bond, based on its price
z02, as the interest on a bond with two-year maturity. Notice that the actual unit bond interest at Time 1, r12, is
not known at Time 0.
It is not possible to duplicate the cash flows from the discount bond by a planned sequence of unit bond
transactions. If I buy z02 Time 1 unit bonds and reinvest the payoff from them in Time 2 bonds at Time 1 the
amount I will have at Time 2 will not be a certain amount, but rather an amount that will depend on the
outcome of the coin flip. To obtain a certain return at Time 2, other than by buying the discount bond, I must
purchase at Time units of risky instrument H and units of risky instrument T. By
reinvesting the payoff at Time 1 in the Time 1 unit bond, a certain outcome of 1 at Time 2 is assured. Thus, if
the coin comes up heads at Time 1, the payoff from the position in the package of risky instruments is
(instrument H pays off 1 per unit, and instrument T pays off zero). Reinvesting the proceeds in the
Time 1 unit bond generates an outcome at Time 2.
For consistency (that is, to eliminate the potential for arbitrage profit), the tax liability of the person who
takes the direct route to a certain $1 at Time 2 by purchasing the discount bond must be the same as that of
one who buys the appropriate duplicating package of risky instruments and unit bonds through time. The
precise meaning of ''the same" merits closer examination, but the condition is clearly met if the cash tax
liability is the same at every point in time and under every contingency.
338
339
The fixed point of this situation is the tax treatment of the unit bonds, period by period. A sequence of cash
flows identical to a Time 0 unit bond can be obtained by purchasing 1 + r01 units of each of the two risky
instruments. An arbitrage profit opportunity will be
Â
Page 420
available unless the package is taxed the same as the Time 0 unit bond (thus integrating the taxpayer's
positions in the two instruments). Applying to the contingent returns, the system discussed in the previous
Section, using the Time 0 unit bond rate to adjust for timing, will produce the desired tax outcome.
To illustrate the system, start with the condition for financial market equilibrium in the absence of taxes.
Since the outcome is a certain payoff of $1 at Time 1, the position consisting of one unit of each of the risky
instruments can be financed by borrowing the purchase price of the position, , in the form of the Time
0 unit bond. The repayment due at Time 1 will be
.
Now suppose there is a tax and the person in our illustration elects a GRD of Time 1 and a GTR of 50% to
apply to the package of risky instruments acquired at Time 0. In the case examined here, the uncertainty is
resolved at Time 1, when the coin is flipped. Until then, the instruments increase in value at the Time 0 bond
rate. If the prices are those determined without consideration of taxes, at Time 1, the investor will have no
profit apart from possible tax considerations. There will be interest imputed to the basis of but this
will be matched by the deduction of interest paid on the money borrowed to buy the risky instruments. For
tax purposes, the gain at Time 1, which is the chosen GRD, will be the market value of the position, 1, less
the adjusted basis, . This difference will be zero if the assumed prices prevail. So they will
be equilibrium prices, and there will be no tax due on the position.
What about the tax results of buying the equivalent to the Time 2 discount bond in the form of the appropriate
position in risky instruments, combined with investment of the payoff at Time 1? Suppose again that the tax
scheme works and that the pricing in the absence of taxes prevails in the presence of taxes. Then (12) holds.
Consider the case that the coin comes up heads. Applying the same rules, there is a taxable gain at Time 1 on
the position in instrument H equal to
Â
Page 421
and a gain (negative) on the position in instrument T equal to
for a net taxable gain overall of
339
340
In general, the net of the two gains will not be zero and a tax payment or refund, at the GTR, would be due.
Applying the same method to the Time 2 discount bond would have the same tax consequences, given the
same choice of GRD and GTR, if the bond were sold at Time 1, with the proceeds reinvested in the Time 1
unit bond. It may be less obvious that the result will be the same if, instead, the bond is held until maturity.
Under the rules, and still assuming the coin comes up heads, the interest rate used for imputing income to the
bondholder would be r in the first period and in the second. At Time 2, the bondholder would be treated
01
as having gain . The gain would be "discounted" to Time 1, treated as
at that point. Substituting the candidate equilibrium relationship (12) into the expression for gain yields
which will be seen to be identical to (15), the gain on the position in the risky instruments used in the
duplicating sequence of transactions. The results thus would be the same, after all taxes, of holding the Time
2 discount bond and the equivalent sequence of a package of risky instruments at Time 0, followed by
reinvestment of the proceeds at Time 1 to obtain a certain 1 at Time 2.
VI
Conclusion
The approach to realization in the existing income taxation of financial instruments renders inevitable the
complexities I describe as
Â
Page 422
nonlinearity. In a general sense, this is well known. I have suggested that the fundamental difficulty is dealing
with the intertemporal aspects of income accounting (the reward to waiting) rather than the challenges of
uncertainty (the reward to risk-bearing). The present analysis implies that the intertemporal problems may be
more serious than may have been thought (as reflected in shortcomings of the rules for instruments with fixed
and determinate returns), whereas the treatment of risk (the rules for contingent returns) is in a sense simple.
It is, however, the element of uncertainty that gives rise to realization accounting. You cannot record what
you do not yet know. In this chapter, I conclude that the only approaches that would make realization
accounting "work" involve virtually universal application of retrospective allocation of gain and, especially,
imputation of interest at a standard rate.
In the world of zero transactions cost examined in this chapter, there is no substitute for nonlinearity of the
income measurement system (such as the limit on the deduction of capital losses) barring near-universal
application of interest at a standard rate to basis, retrospectively if need be. The only instruments for which
what is labeled "interest" in the contract is the amount subject to tax are the unit bonds. These are the default
risk-free instruments against which tax arbitrage is constructed. (I do not think these "standard" instruments
340
341
need to be free of risk, default or otherwise, but it would greatly complicate the story to spell out why.) For
all the other instruments, including most of those currently regarded as unproblematic (fixed-return assets),
either mark-to-market taxation or imputation of interest to basis is required. Furthermore, since essentially all
other instruments are risky, relative to the unit bonds, for all other instruments, either mark-to-market
taxation or the realization rules described in this chapter must be applied.
One implication is that, if it is desired to have a linear income measurement system, considerations of
transactions cost must enter the design of rules. Another, more radical, thought is suggested. Assuming the
taxation of interest income is regarded as worth the candle, perhaps consideration should be given to truly
universal application of a standardized rate of return to basis (with the sort of realization rules described in
this chapter). There would be no taxation of interest as observed in the market (no unit bonds). That
Â
Page 423
would make it a simple matter to adjust basis for inflation, and probably come very much closer to a true
income tax than is achieved by the current hybrid rules.
Appendix: A Continuous-Discounting Version of the Proposed Rules
Some readers may find helpful a demonstration, using continuous discounting, that the proposed rules render
the taxpayer indifferent to the timing of realization when it is assumed that the asset in question will increase
in value at the unit bond yield. Consider the taxpayer's decision to postpone realization from some time s to
some future time s'. (To simplify matters, assume both times are after the GRD for the instrument in question.
The mathematical expressions developed in this Appendix apply as well for a realization before the GRD.)
As noted previously, 56 in the case of realization at a time s, an extra dollar realized implies an extra gain tax
liability at the realization date of 1 - (1 - GTR)e-mr(s-GRD), where there is continuous compounding at a constant
interest rate r and where a constant marginal tax rate m is assumed.
If an asset has value V(s) at time s, the path of its value at future time s', accruing at the unit bond rate, is
described by V(s)er(s'-s) (assuming throughout no cash flow from the asset). If the asset follows this path, the
amount of capital gain as of the GRD would be unchanged by postponing realization. The extra tax due at a
future realization date would rise, since there would be more postponement of liability to make up. Suppose,
for example, that the current adjusted basis in the asset is B(s), so that the gain, reckoned at the GRD, is
which also would be the extra basis subject to interest imputation upon realization. Then the gain tax as of the
GRD would be GTR X. This translates into extra tax due at the hypothetical later realization date of GTR
Xe(1-m)r(s'-GRD), where the liability cumulates at the after-tax interest rate (since interest paid on postponed tax
payments would be deductible).
The extra interest imputation follows the path rXer(t-GRD) starting at t = GRD and running up to the realization
date s'. This, in turn, implies an extra flow of tax liability at time t between s and s' of mrXer(t-GRD), which must
be brought forward and paid at s', resulting in extra liability at s' of
341
342
Â
Page 424
Putting the two pieces together, the total tax "pulse" at s' would be
.
The extra after-tax cash flow at t due to waiting until s' to realize would be the tax on interest imputed to
basis,
for t running from s to s'. The net gain from waiting from s to s' would be the discounted value (to s, at the
after-tax rate) of this flow of tax payments and the pulse of realized proceeds of sale at s' less the pulse of
extra tax payments due at s',
less the forgone cash flow (net of taxes) that would have been obtained from realization at s,
.
If this total is positive, it pays to postpone realization. If it is negative, it pays to realize immediately.
By laborious algebra, making use of the fact that, by definition of X, this expression for
the net current cash flow equivalent at s of waiting to realize from s to s' reduces to
which equals, on collection of terms,
Notes
Comments of participants at the Tax Law Review Colloquium on Financial Instruments held at New York
University School of Law, May 22, 1995, subsequent helpful comments from Alan Auerbach, Joseph
Bankman, Mark Gergen, Louis Kaplow, Diane Ring, Robert Scarborough, Deborah Schenk, Daniel Shaviro
and Alvin Warren are gratefully acknowledged. In addition, I would like to thank the John M. Olin
Foundation for support of the research underlying this chapter. Views expressed are my own and should not
be taken as representing any institution.
1. Alvin C. Warren, Jr., "Financial Contract Innovation and Income Tax Policy," 107 Harv. L. Rev. 460
(1993) [hereinafter "Innovation"].
Â
342
343
Page 425
2. See, e.g., Frank V. Battle, Jr., "Bifurcation of Financial Instruments," 69 Taxes 821 (1991); David P.
Hariton, "The Taxation of Complex Financial Instruments," 43 Tax L. Rev 731 (1988), Randall K. C. Kau,
"Carving Up Assets and LiabilitiesIntegration or Bifurcation of Financial Products," 68 Taxes 1003 (1990);
Robert H. Scarborough, "Different Rules for Different Players and Products: The Patchwork Taxation of
Derivatives," 72 Taxes 1031 (1994); Reed Shuldiner, ''A General Approach to the Taxation of Financial
Instruments," 71 Tex. L. Rev. 243 (1992).
3. Warren, "Innovation," note 1, at 461-65.
4. IRC Â§ 1272(a)(1) (inclusion of interest in income of holder), Â§ 163(e)(1) (deduction for holder), Â§
1271(a)(3) (providing for yield to maturity calculation).
5. IRC Â§ 1001(a).
7. See id. at 474.
8. See id at 475.
9. Id. at 492.
10. Such instruments may be subject to inflation risk, however. As I emphasize, consistency, not correctness,
is the objective.
11. IRC Â§ 1211 (limiting individual's capital losses for taxable year to capital gains plus $3,000).
12. See text accompanying notes 13-14.
13. A transaction itself may be more or less complex. So, for example, acquisition of an OID bond gives rise
under the rules to a sequence of increments to future taxable income. If the bond is sold, the sequence is
revised in a way specified by the rules.
14. This definition is very close

to that of Jeff Strnad, "Taxing
New Financial Products: A
Conceptual Framework," 46
Stan. L. Rev. 569, 576 (1994).
15. IRC Â§ 1212.
16. IRC Â§ 163(d).
17. For a comprehensive analysis of the role and effect of loss limitations to obstruct tax arbitrage, see Robert
H. Scarborough, "Risk, Diversification and the Design of Loss Limitations Under a Realization-Based
Income Tax," 48 Tax L. Rev. 677 (1993).
18. This is the average real rate of return on U.S. Treasury bills. See R. G. Ibbotsen Assocs., Stocks, Bonds,
Bills and Inflation: 1995 Yearbook 31 (1995).
19. See, e.g., Louis Kaplow, "Taxation and Risk Taking: A General Equilibrium Perspective," 47 Nat'l Tax J.
789 (1994); Deborah H. Schenk, "Taxation of Equity Derivatives: A Partial Integration Proposal," 50 Tax L.
Rev. 571 (1995); Daniel Shaviro, "Risk-Based Rules and the Taxation of Capital Income," 50 Tax L. Rev. 643
(1995); Warren, note 1.
20. See, e.g., IRC Â§ 1274(b)(2)(B).
343
344
21. Alan J. Auerbach, "Retrospective Capital Gains Taxation," 81 Am. Econ. Rev. 167 (1991).
Â
Page 426
22. For definitions of tax

arbitrage, see C. Eugene
Steuerle, Taxes, Loans, and
Inflation 57-80 (1985)
(describing tax arbitrage to be
influenced by different taxation
of various assets); David F.
Bradford, "The Economics of
Tax Policy Toward Savings," in
2 The Government and Capital
Formation, 38-50, 62 (George
M. von Furstenberg ed., 1980)
[hereinafter "Tax Policy"]
(suggesting that applicability of
both income and consumption
tax rules to similar assets may
give rise to tax arbitrage); Alvin
C. Warren. Jr., ''Accelerated
Capital Recovery, Debt, and Tax
Arbitrage," 38 Tax Law. 549,
564-74 (1985) (arguing that
market adjustment may not
always eliminate advantages of
tax arbitrage).
23. See IRC Â§ 163(a) (providing deduction for interest), Â§ 168(a) (providing cost recovery deductions).
24. See generally Bradford L. Ferguson, "The Rationales for the Rules: How to Think About Derivatives in
the Tax World," 72 Taxes 995, 1003 (1994).
25. IRC Â§ 61(a)(4) (providing for inclusion of interest in gross income).
26. See Boris I. Bittker, "Equity, Efficiency, and Income Tax Theory: Do Misallocations Drive Out
Inequities?," in The Economics of Taxation 19 (Henry J. Aaron & Michael J. Boskin eds., 1980).
27. See, e.g., David F. Bradford, Untangling the Income Tax 229-30 (1986) [hereinafter Untangling].
28. See Section III A.
30. Id. at 465-73.
31. See note 24.
32. IRC Â§Â§ 1272-1275.
344
345
33. Reed H. Shuldiner, "Consistency and the Taxation of Financial Products," 70 Taxes 781, 782 (1992)
[hereinafter "Consistency"].
34. In a graduated rate system, the marginal tax rate can adjust endogenously. So, for example, if high
bracket taxpayers could deduct interest on borrowing to hold tax-exempts, and low bracket taxpayers could
go short tax-exempts costlessly, tax arbitrage would lead everyone to end up in the same marginal rate
bracket. Someone whose marginal tax rate was below this common level would lend on a taxable basis and
borrow tax-exempt by going short. The result would be an arbitrage gain, a higher taxable income and a
higher marginal tax rate. The opportunity for arbitrage profit would persist until marginal tax rates were
equal. Note that a limit on allowable loss deductions also can be viewed as making the marginal tax rate
endogenous, since a taxpayer up against the limit faces a marginal tax rate of zero on further losses of the
type subject to the limit. For further development of the role of endogenous adjustment in marginal rates, see
Bradford, "Tax policy," note 22, at 47-49.
35. This is the approach suggested by NoÃ«l B. Cunningham & Deborah H. Schenk, "Taxation Without
Realization: A 'Revolutionary' Approach to Ownership," 47 Tax L Rev. 725 (1992).
36. See note 4.
Â
Page 427
37. See Joseph Bankman & William A. Klein, "Accurate Taxation of Long-Term Debt: Taking into Account
the Term Structure of Interest," 44 Tax L. Rev. 335 (1989); Theodore S. Sims, "Long-Term Debt, the Term
Structure of Interest and the Case for Accrual Taxation," 47 Tax L. Rev. 313 (1992).
38. See Section V.B.
39. See Auerbach, note 22.
40. I have described elsewhere taxation of nominal interest as giving rise to a "nominal economic income"
tax. See David F. Bradford & Kyle D. Logue, The Effects of Tax-Law Changes on Prices in the
Property-Casualty Insurance Industry (NBER Working Paper No. 5652, 1996).
41. For discussion and references to the literature, see Bradford, Untangling, note 27, at 40-45, 229-35.
42. The comprehensive income tax proposal described in Treasury Dep't, Blueprints for Basic Tax Reform 75
(1977) erred in relying on financial markets to produce an automatic adjustment of interest income for
inflation. I commented on this error in the second edition. David F. Bradford, Preface to David F. Bradford &
Treasury Tax Policy Dep't, Blueprints for Basic Tax Reform xvii-xviii (2d. ed. 1984).
43. Shuldiner, "Consistency," note 33, at 782.
44. Strnad, note 14, at 573.
45. See Bankman & Klein, note 38; Sims, note 38.
46. The modern theory of the term structure of interest rates treats all but the instantaneous current interest
rate as stochastic, with a risk premium typically built into the expected return on longer-term instruments. See
John C. Cox, Jonathan E. Ingersoll, Jr. & Stephen A. Ross, "A Theory of the Term Structure of Interest
Rates," 53 Econometrics 385 (1985).
47. See Auerbach, note 22.
345
346
48. A peculiarity of this procedure is that the tax consequences of realization would not be fully known until
the GRD. This is a consequence of not knowing the taxpayer's marginal rates (and perhaps the current interest
rates) for the intervening period. If the marginal tax rates and current interest rates, period by period, were
known in advance, the tax on the gain could be brought to the realization date, with appropriate discounting.
In that case, the basis for the imputed interest deduction would be reduced by the amount of the gain tax paid
at realization. (The idea of the interest credit is to give the holder the same amount of "tax-free" accumulation
as would obtain if the asset had been held to the GRD.)
49. In the case of realization at time s, an extra dollar realized implies an extra gain tax liability (reckoned as
of the GRD) of GTRe-r(s-GRD) where continuous compounding at a constant interest rate r has been used to
simplify the calculations. This amount of tax regarded as payable at the GRD translates into an extra tax at
the realization date of GTRe-r(s-GRD)e(1-m)r(s-GRD) = GTRe-mr(s-GRD), where a constant marginal tax rate, m, is assumed.
There is an additional effect due to the revision in the basis for imputation, starting at the GRD. (The
description applies to a realization after the GRD, but the mathematical expressions apply as well for a
realization before the GRD.) The extra basis is e-r(s-GRD), implying a flow of extra interest imputed of
re-r(s-GRD)er(t-GRD) =
Â
Page 428
rer(t-s) starting at t = GRD and running up to the realization date s. This, in turn, implies an extra flow of
tax liability, mrer(t-s), which must be brought forward and paid at s. An extra dollar realized at s thus
implies an extra amount of tax to be paid at s, due to extra interest imputation, of
Putting the two pieces together, the extra tax due at the realization date due to an extra dollar realized
then is GTRe-mr(s-GRD) + 1 - e-mr(s-GRD) = (1 - GTR)e-mr(s-GRD).
50. These assumptions are made

to allow us to focus on the
essential issue of cherrypicking.
Alternatively, one could specify a
hedging transaction that would
eliminate any risk of gain or loss
on the position, as in the put-call
parity theorem, but it would
greatly complicate the
description.
51. Under the Auerbach approach, all the imputation of interest would take place retrospectively at the time
of realization, working back to a kind of reconstructed basis from the amount realized. It would not matter
whether the asset is sold at a gain or loss. In the proposed approach, there would be interest imputed from the
moment of acquisition, but the basis for imputing interest would be adjusted retrospectively, from the GRD
forward, at the time of realization. So if the GRD is the acquisition date, the effective imputation of interest
would occur as under the Auerbach approach. Since, under the Auerbach approach, the amount of gain or
loss is irrelevant, the GTR is effectively zero. This may be seen in the continuous compounding example in
note 50: If GTR = GRD = 0, the only extra tax at the time of realization would be the "correction" to the
imputation of interest since acquisition. This is, effectively, the Auerbach result.
346
347
52. As in that case, to complete the establishment of equivalence from the point of view of everyone,
including the government (representing the rest of the taxpayers), requires the government to take a position
appropriately offsetting that taken by the taxpayer. To illustrate, in the example of the bets using red and
green contracts, if one side of the bet chooses red and the other green (and both have the same tax rate), the
government would need to write a taxable contract on the same event to avoid any net cash flow due to the
bet.
53. Profit from buying and selling as the prices of longer-term instruments vary with changes in short-term
rates is the main focus of Jeff Strnad, "The Taxation of Bonds: The Tax Trading Dimension," 81 Va. L. Rev.
47 (1995).
54. Taking into account that the future unit bond rates are not known adds (to the unknown future tax rates)
another reason to wait until the GRD to complete the taxation of the sale of an instrument before the GRD. It
is likely, however, that the equivalent at the time of realization could be derived from options prices. I have
not tried to work out the details.
55. See, e.g., IRC Â§ 1274(d)(1) (describing federal short-term, mid-term and long-term rates).
56. See note 50.
Â
Page 429
IV
ECONOMICALLY MEANINGFUL MEASURES OF NATIONAL SAVING
Â
Page 431
15
Market Value Versus Financial Accounting Measures of National Saving
1
Introduction
This chapter is a venture into well-trodden terrain: the definition of saving. Because so many others have
thought about the same issues, probably nothing I say here has not been said before by someone else. J. R.
Hicks (1946) mapped the territory in a particularly well-known theoretical treatment. More recently,
Auerbach (1985), Boskin (1986, 1988), Eisner (1980, 1988), Goldsmith (1982), Peek (1986), Ruggles and
Ruggles (1981), and Shoven (1984) have discussed many of the points raised here in connection with
empirical explorations of saving and wealth. In his presidential address to the American Economic
Association, Eisner (1989) included the main theses argued here in a broadside indictment of the divergence
between measurement and theory to be found in economics. This paper differs, perhaps, in degree of
emphasis of two propositions. The minor theme is that saving should be defined by reference to the
347
348
underlying concept of wealth to which the saving is an increment. The major theme is that the most useful
wealth concept is the market value of assets, not the cost-based measure of capital implied by the use of
national income and product account (NIPA) saving. Whereas NIPA investment measures tell us something
about the margin of productive additions to the stock of wealth in a particular form, the (definition-ally equal)
saving measures are neither those that the microeconomic theory of consumption explains nor those
appropriate to assess national economic performance.
Inspection of a sample of the extensive literature commenting on and analyzing national saving has surprised
me by the diversity of positions, often implicit, on these issues. It appears that the macro-economists are truer
to microeconomic principles than are many of
Â
Page 432
those who approach the subject from a public finance perspective. The fact that so much research is carried
out making use of statistical measures of saving that seem to me to bear so little relationship to economic
theory suggests there is a place for a review of fundamentals and display of some basic data related to them.
1.1
Income, Saving, and Wealth
Beginning students are taught that saving is a residual, what is left from personal income after deducting
consumption and taxes or after deducting from aggregate income consumption by households and
governments. But saving is also conceived of as an addition to wealth, and it is not always recognized that the
three ideasconsumption, income, and wealthare not independent. Defining any two determines the definition
of the third. The Schanz-Haig-Simons (SHS) conception of income familiar to public finance takes the ideas
of consumption and wealth as fundamental and defines income as the sum of consumption and the change in
wealth during an accounting period. The basic notion of wealth, in turn, is the market value of a household's
(or household aggregate's) stock of claims on goods and services in the future. 1 This is the approach to
saving taken by the microeconomic theory of household behavior.
Most commentary on and analysis of national saving, by contrast, start with a NIPA definition of income. To
make life confusing, the term "income" in the national income account context is attached to factor payments
and makes distinctions between taxes regarded as falling on factor payments and those that do not (indirect
business taxes). It is doubtful that there is an economically meaningful distinction between taxes that bear on
factor payments and those that do not. We can cut through the problem if, for the concept of income in the
SHS sense, we read "product" in the national accounting sense.
Which of the three notionsproduct, consumption, wealthare fundamental in the case of national income
accounting is not immediately obvious. As is well known, national income accounts involve two conceptions
of product, gross and net. Gross national product, "the market value of the goods and services produced by
labor and property supplied by residents of the United States" (U.S. Department of Commerce, Bureau of
Economic Analysis 1986), and consumption, personal and governmental, can reasonably be described as the
fun-
Â
348
349
Page 433
damental ideas. Together (by subtraction) they define gross investment and saving. To reach net product, net
investment, and net saving, it is necessary to subtract an allowance for the "using up" of the reproducible
capital stock, a wealth notion. Here, then, it is the wealth and consumption ideas that are fundamental: we can
think of net product (income) as definitionally equal to the sum of consumption (personal and governmental)
and the change in the reproducible capital stock owned by U.S. residents.
1.2
NIPA Saving and Financial Accounting
In its treatment of business

investment and its yield, the
NIPA net income concept can be
loosely characterized as a
consolidation of the account
books of business firms. This is
not to suggest that the NIPA
accountants actually aggregate
the income statements and
balance sheets of firms. It is
rather to emphasize that
investment (and therefore
saving) in the national income
and product accounts consists of
acquisitions of tangible property
and is, furthermore, cost-based,
constructed from historical data
on expenditures for machines,
structures, and inventories.
Increments in the value of
intangible property and (what
may be the same thing)
revaluations of tangible property
arising from its location within
going businesses are excluded
from the NIPA income and
saving concepts. Net saving in
the national income and product
accounts constitutes the change
in the stock of reproducible
business capital. 2 The NIPA
capital data can be thought of as
the figures financial accountants
would present if they used the
NIPA depreciation conventions
and adjusted their historical
cost-based entries on tangible
assets (including inventories)
annually to what they would be
had historical prices been instead
at current levels.
The main difference between the two conceptions of wealth corresponds roughly to the difference between
349
350
financial accounting for the net worth of business firms, on the one hand, and the market valuation of those
firms on the other ("roughly" because financial accounts include intangible assets acquired by purchase from
another firm). The difference is sometimes summed up as that between recognition or not of "capital gains,"
but this description hides as much as it reveals. The market value of the equity of a firm may differ from the
"book" value of its tangible property for many reasons, including changes in the supply price of the capital
items in
Â
Page 434
question (for which national income accounting makes a correction), changes in discount rates, and changes
in the beliefs about the future upon which market valuation of assets dependsall of these give rise to capital
gains in the popular sense of the term. But the two values also may differ because of the genuinely stochastic
character of the returns on investment and the conservative quality of business accounts, which result in little
or no tracking of the accumulation of intangible capital and of such assets as proven oil reserves.
1.3
Empirical Relevance: A First Look
Available data suggest that the difference in definition corresponds to a significant difference in aggregate
wealth measures. Table 15.1 shows estimates of the net worth of nonfinancial corporate business in the
United States (including corporate farms) and of the market value of the equity claims on those firms. The
figures are derived from the Balance Sheets for the U.S. Economy (hereafter, National Balance Sheets)
prepared by the Board of Governors of the Federal Reserve System (1988). 3 Net worth consists of the
difference between assets and liabilities on the account books after various adjustments. Assets in this case
include reproducible assets at replacement cost (i.e., after adjusting valuation based on historical cost for
changes in the acquisition prices of the same assets), land at market value, and direct investment abroad by
U.S. firms. Liabilities include all the usual sorts of debt (at book value), profit taxes payable, and foreign
direct investment in the United States. I would emphasize that in its treatment of fixed investment the net
worth in table 15.1 is essentially the concept implicit in NIPA accounting for saving. The market value of
equity is essentially that appropriate for the SHS saving concept, which, in turn, is in the concept "explained"
by micro-economic theories of saving behavior.
It is evident from table 15.1 that the market value of equity and the net worth on firms' books are very
different. The column titled "Market Value/Net Worth Ratio" shows the ratio of the market value of the
equity claims to the consolidated nonfinancial corporate sector to the consolidated financial accounting
measure of net worth, that is, the sum of tangible and financial assets (including direct investment abroad)
less the sum of debt claims (at book value), profit taxes payable, and foreign direct investment in the United
States. Since 1948 this ratio has varied over a remarkable range, with a high
Â
Page 435
350
351
Table 15.1
"Book" net worth and market values of U.S. nonfinancial corporate business, year end, 1948-87
Net worth of U.S. nonfinancial corporate Market value Market value/ Net worth
business of corporate equities net worth ratio less market
Year ($ millions) ($ millions) (%) to GNP (%)
1948 209,615 83,862 40.0 48.1
1949 219,672 92,205 42.0 49.0
1950 244,190 116,647 47.8 44.2
1951 269,211 138,250 51.4 39.3
1952 285,071 149,941 52.6 38.4
1953 300,142 144,776 48.2 41.8
1954 315,117 216,033 68.6 26.6
1955 342,531 269,173 78.6 18.1
1956 378,078 289,169 76.5 20.8
1957 403,297 242,470 60.1 35.7
1958 419,289 342,082 81.6 16.9
1959 439,972 361,299 82.1 15.9
1960 448,422 354,114 79.0 18.3
1961 461,733 428,294 92.8 6.3
1962 475,580 389,171 81.8 15.0
1963 489,970 456,076 93.1 5.6
1964 513,321 509,516 99.3 .6
1965 543,746 553,720 101.8 -1.4
1966 583,906 504,223 86.4 10.3
1967 621,655 651,678 104.8 -3.7
1968 668,880 736,506 110.1 -7.6
1969 729,963 646,230 88.5 8.7
1970 784,634 648,492 82.6 13.4
1971 856,111 758,897 88.6 8.8
1972 934,346 855,233 91.5 6.5
1973 1,048,013 678,436 64.7 27.2
1974 1,337,118 499,098 37.3 56.9
1975 1,491,060 684,337 45.9 50.5
1976 1,647,452 787,807 47.8 48.2
1977 1,817,268 748,002 41.2 53.7
1978 2,107,859 773,143 36.7 59.3
1979 2,419,386 933,373 38.6 59.2
1980 2,780,531 1,293,116 46.5 54.4
1981 3,109,641 1,214,845 39.1 62.1
1982 3,230,025 1,382,773 42.8 58.3
1983 3,327,399 1,638,730 49.2 49.6
351
352
Â
Page 436
Table 15.1
(continued)
Net worth less
Market value/net market
Net worth of U.S. nonfinancial Market value of corporate worth ratio to GNP
Year corporate business ($ millions) equities ($ millions) (%) (%)
1984 3,447,798 1,617,733 46.9 48.5
1985 3,503,026 2,022,648 57.7 36.9
1986 3,560,138 2,332,629 65.5 28.9
1987 3,657,167 2,331,322 63.7 29.3
Source: See text. Based on Board of Governors of the Federal Reserve System (1984).
of 110.1 percent at the end of 1968 and a low of 36.7 percent at the end of 1978.
To put the divergence between accounting and market values of corporate equity in perspective, the column
of table 15.1 headed "Net Worth Less Market to GNP" shows the ratio of the difference to the GNP. The
difference ranges between an excess of over 7 percent and a shortfall of over 62 percent, with a substantial
decrease on average. Figures 15.1 and 15.2 make the points graphically.
It seems clear that the basic objective of the National Balance Sheets, to measure wealth at market value, is
the one appropriate for discussions of saving. Nevertheless, economists widely accept and use for this
purpose the NIPA saving data. Distinguished examples (and I make no claim to a systematic review of the
literature) include Blades and Sturm (1982), Boskin and Lau (1988), Campbell (1987), Lipsey and Kravis
(1987), most of the contributors to Lipsey and Tice (1989), Poterba (1987), and Summers (1985).
In at least some of these instances, lack of market-value wealth data is taken to justify resort to NIPA
concepts, and some analysts (e.g., Auerbach 1985; Boskin 1986, 1988; Poterba and Summers 1987) have
noted the potential role for the market-value data provided in the National Balance Sheets. Summers and
Carroll (1987) explicitly analyze aggregate saving in the National Balance Sheets sense (although they do not
regard it as preferable to the NIPA measure). Noting that "national income account (NIA) data provide
notoriously poor proxies for the economic concepts of saving and investment," Obstfeld (1986, 82) explores
some of the biases that may result from the use of NIPA data in comparing saving and investment behavior of
countries. Some macroeconomistsfor example, Hall (1978, 1988) and Campbell and Deaton (1988)go out of
their
Â
352
353
Page 437
FigureÂ 15.1
MarketÂ valueÂ ofÂ corporateÂ equity/corporateÂ netÂ worth,Â 1948-87
FigureÂ 15.2
''Book"Â lessÂ marketÂ valueÂ ofÂ corporateÂ equity,Â 1948-87
Â
Page 438
way to avoid measuring saving.

Hall, in particular, has argued that
income aggregates are misplaced in
macroeconomics; focus should
instead be on aggregate
consumption and labor earnings.
Granting some such exceptions in
the literature, I think it is fair to say
that there is wide acceptance of
353
354
NIPA saving measures.
In this paper I argue that wealth

and consumption are both
important variables in economic
models and important measures
of economic performance, that
income should be viewed as a
derivative concept in this
connection, and that the
appropriate concept of wealth is
measured at asset market value.
We should use NIPA saving
measures only to the extent that
they serve as reasonable proxies
for the market-value measures.
(This is not to suggest that the
corresponding investment
concepts are not useful in the
analysis of production.) Although
it is ultimately a statistical
question whether the NIPA
saving measures are reasonable
proxies, the evidence from the
National Balance Sheets leads me
to doubt it.
In the next part of the paper I review the relationship between the two notions of wealth (and therefore of
saving): market value of assets and financial accounting net worth. I then take up objections to the use of
market-value wealth. The fourth section presents time-series data on the behavior of national saving in the
U.S. economy, and the fifth raises, without solving, some significant problems with the National Balance
Sheets data as measures of market value.
Much attention has been paid in recent years to the saving performance of U.S. residents, which has been
generally judged disappointing. My contention, that the NIPA saving aggregates and ratios of NIPA saving to
NIPA income measures are poor indicators upon which to base conclusions, is neither inherently in favor of
this assessment nor opposed to it. One may still be dissatisfied with the U.S. saving record when it is looked
at in the framework suggested by microeconomic theory. The sixth section presents some observations on
this issue.
2
Concepts of Wealth
2.1
Market Value of Assets
The SHS notion of income underlying the base of an income tax (or at least generally accepted by academic
commentators as the proper base of an income tax) is the sum of the change in the wealth and the
Â
354
355
Page 439
consumption of the taxpaying

unit, be it an individual or a
family. Consumption and wealth
are the primitive concepts, which
need to be given operational
substance to produce a tax system.
Although the general ideas seem
obvious enough, both pose
difficult problems of definition at
the margin. Within limits, the
standard to which the operational
definitions refer in a tax policy
context is essentially
normativeone starts with a notion
of ability to pay and designs the
income measure to implement it.
(The limits relate to the
substitutability of different forms
of wealth in taxpayer portfolios.)
In Untangling the Income Tax (Bradford 1986) I suggested that the usual arguments justifying the SHS
income concept as a tax base imply a definition of a person's wealth as "the maximum amount of present
consumption he could finance currently by selling or otherwise committing all of his assets" (22). If this
definition is accepted, the operational focus shifts to the identification of "assets" and quantifying the
opportunities of ''selling or otherwise committing" them. Examples of significant but hard-to-quantify assets
are human capital (the present value of a person's future earning power) and the discounted value of
inheritances. It is interesting that these two are also examples of assets that are difficult to sell or "otherwise
commit." Proponents of SHS income taxation normally exclude both human capital and the value of great
expectations from the wealth component of the definition of income.
Experience with tax administration gives us numerous examples of the fact that it is the market value of
wealth, rather than its accounting value, that figures in individual behavior. If tax on accruing market value
(capital gains) is deferred, taxpayers will concentrate their portfolios in assets that generate accruing value
rather than cash income. If accounting measures of depreciation are different from actually accruing changes
in value of assets, taxpayers respond in well-known ways.
A simple two-period model of a person's intertemporal budget constraint will help clarify the role and nature
of wealth in the analysis of behavior, in this case the explanation of consumption levels. For the purpose, we
can imagine a world in which there is just one consumption good and in which labor is supplied inelastically,
with no welfare significance. We conceive of people as born into this world with inherited resources (to be
specified), working one unit of time during the first period to earn the wage w1 (measured in consumption
units), consuming an amount C1, and applying any excess
Â
355
356
Page 440
of the wage over consumption to purchase assets. In the second period, the person also works one unit of time
to earn the wage w2 and consumes that amount plus the results of liquidating the assets. The problem is to
choose an amount of first-period consumption and a portfolio of assets.
In the most basic model, there is

no uncertainty (so there is no
information problem). The
second-period wage is known
and there is a single asset
available, which we may think of
as a discount bond paying one
unit of consumption in period 2.
The going price for the asset is
P2. The person is born holding B1
units of the bond and, in the
course of period 1, chooses the
number of units of the asset to
buy (or sell) so as to carry B2
units into period 2. Two
equations(1) and (2)define the
lifetime budget constraint.
The intermediate asset position, B2, can be eliminated between (1) and (2) to yield a single lifetime budget
constraint, (3).
The right-hand side of equation (3), w1 + w2P2 + B1P2, is the market value of "opening wealth" (including
human capital). We see from (3) that in this simple world we can specify the person's opportunity set
completely with two numbers, opening wealth and P2, the price of claims on period-2 consumption (or the
interest rate). To specify the opportunity set without capitalizing labor services, we need four numbers, B1P2,
w1, w2, and P2: opening nonhuman wealth, wages in the two periods, and the interest rate.
This simple formulation reminds us that if we are looking forward from a point in time and want to explain
consumption levels, wealth is a needed piece of information. It also demonstrates that it is not the only piece
of information we need to explain consumption or, a related problem, to assess a person's welfare, even under
the simple, perfect market conditions of the model. In general, information about prices is neededhere, wages
and the interest rate and in a multiperiod setting, wages, relative prices of goods, and a term structure of
interest rates. By inspection of condition (3) we see that in the simple model the welfare of the individual is
increasing in opening wealth including human capital and decreasing in the price
Â
356
357
Page 441
of future consumption (i.e., increasing in the interest rate). But even in this case, when human capital is
excluded, although welfare is still increasing in opening wealth, the effect of an increase in the interest rate
on the assessment is indeterminate and hinges on the taste for consumption in period 2. Simply put, a high
interest rate is bad for someone who wants to borrow against tomorrow's earnings to consume more today. 4
Initial nonhuman wealth is a

given, a parameter, in the model
described above; wealth along
the way (initial wealth
augmented by saving) is chosen,
endogenous. A complete model
would explain initial wealth, too,
so it would drop out of the
analysis. Wealth would return as
an explanatory variable, though,
with the introduction of
uncertainty. Then the wealth
along the way is the result of the
individual person's choice and
luck, so second-period
consumption would depend upon
the market performance of the
portfolio. The same would be
true for the aggregate of
individuals.
The model reminds us that to

predict the level of consumption
we need to take into account the
market value of nonhuman
wealth, the interest rate, and
current and future wages. In a
stochastic setting the distribution
of future wages could be
correlated with the value of
nonhuman wealth, marketed and
unmarketed. In particular, one
might expect workers observing
prosperity (high market value of
wealth) to raise their forecasts of
future wages. If we take into
account that lifetime labor
supply is chosen along with
consumption levels, it is far
from clear what sort of
consumption behavior one ought
to expect to see associated with
movements in the market value
of wealth.5
With enough simplifying assumptions, though, one can derive from the general approach outlined above the
conclusion that a person's current consumption will be a function of his forecasted labor earnings and current
357
358
wealth, for example,
where a and b are constants, Et is the expectation conditional on information at time t, and Wt is the
(stochastic) market value of nonhuman wealth.6 Such a model will generate a time path of consumption and
wealth, and hence of saving, defined as the change in wealth. The point to emphasize here is that such
regularity as the models do lead us to look for is in the relationship among consumption, labor earnings, and
wealth at market value.
Â
Page 442
2.2
Net Worth as an Accounting Idea
We can capture in a crude way the role for financial accounting in the simple model of behavior by adding an
explicit, real asset, say a certain number of machines, M1, as another element of endowment. In the typical
financial accounting context, there is no readily observable market for fixed capital. Assume, therefore, that
the machines are inalienable (i.e., they cannot be sold). The number of machines is tracked by the financial
accounts. A machine generates output in period 2 ( would be stochastic in a realistic model). Then the
budget constraint is expressed by equations (4) and (5); the single-constraint version that eliminates the
financial assets carried over is expressed by (6).
It is evident from (6) that, in a

world of certainty, with
unlimited borrowing and
lending of the financial asset,
the only use of the financial
accounting information is to
provide a basis for estimating
what the market value of the
machines would be . If
one knows the market value of
the machines, the accounting
information is superfluous.
Complicating the model by introducing an explicit treatment of uncertainty and asymmetries of information
does not suggest a further role for financial accounting information. With complete Arrow-Debreu contingent
claim markets, the market value of wealth continues to define the position of the budget constraint. Owing to
the increased number of prices, ambiguities about the signs of derivatives multiply in welfare comparisons or
positive predictions of the effect of changes in parameters on consumption or labor supply. Missing markets,
asymmetries, liquidity constraints, and the like render budget sets nonlinear and reduce the information
contained in any single parameter, such as initial wealth, of the individual's problem. Nevertheless, there does
not appear to be a general role for accounting information except as the basis for estimating implicit market
values.
358
359
Â
Page 443
The function of financial accounting for a business firm is not to duplicate market valuation. A clear
statement of this point is presented in an official pronouncement of the Financial Accounting Standards
Board: "Financial accounting is not designed to measure directly the value of a business enterprise, but the
information it provides may be helpful to those who wish to estimate its value" (Financial Accounting
Standards Board, 1978, as excerpted in Gibson and Frishkoff, 1986, 19). Financial accounting for asset value
and market value converge where there is an actual transaction that renders the market value objectively
measurable. Between transactions, accounting rules prescribe transformations (depreciation, amortization,
etc.) of the original market-value data to describe the stock of assets involved.
It is tempting, and I think even usual among economists, to attribute to the accounting measure of net worth
(appropriately corrected to some sort of replacement-cost basis) the status of a kind of "permanent income"
measure, a stationary point in the noisy world of asset revaluations. I am not aware, however, of any
empirical evidence in support of this characterization of accounting net worth in relation to the valuation of
firms (nor of the related characterization of accounting depreciation). 7
There are really two reasons we

should expect accounting values
to differ from market values of
firms. First, accounting practices
clearly lay no claim to tracking
the market values of those assets
that are carried on the books.
Thus, for example, the
depreciated accounting value of
fixed investment neither is, nor
claims to be, a stand-in for
market value for the assets
involved.8 Intangible assets
acquired by purchase are
generally amortized according to
formula.9 Depreciation or
amortization deductions for
retirements from the stock of
assets, based on the amounts
paid for the assets, are needed to
account for the fact that some
systematic effect can be
expected with the passage of
time. These allowances are, to
be sure, based on experience
with the physical or otherwise
determined useful lives of
similar assets in the past, but to
serve their purpose they must be
formally prescribed in
accounting rules. They do not
refer to assessments of current
359
360
market value in the context of

the firm, which may deviate up
or down from the path implied
by accounting rules of thumb.
Second, important intangible assets created by the activities of a firm (i.e., not bought from another firm) are
typically not carried on
Â
Page 444
the balance sheet at all. As is well known, research and development and advertising outlays are expensed
currently. Successful efforts do not generate assets on the books unless there is an actual transaction, such as
a sale of patent rights. The value of a firm that discovered the laser or the transistor and could appropriate the
resulting value would surely jump in market value. Its accounting net worth, however, would not change. The
same is true for an economy under NIPA capital accounting practices. Since the inventor of a new idea may
have difficulty capturing the rents, there is a better basis for excluding the value increase from company
books than for excluding it from a national aggregation. Technological and market surprises of many kinds
(oil price shocks, technological breakthroughs, discovery of a new oil field) are excluded from company
books and from NIPA income and capital accounts. Observation of the histories of firms such as computer,
automobile, and pharmaceutical companies makes clear that large movements in value are associated with the
success or failure of ideas (including marketing) and organizational innovations. Such value changes are
clearly of great quantitative significance, quite stochastic, and weakly, if at all, related to investment in fixed
capital.
In short, the accounting net worth of the firm is a measure of some of its past inputs. It represents the solution
to an intractable statistical problem: how to aggregate information about financial commitments through time
embodied in property of one sort or another. It is not a shortcoming of accounting net worth that it does not
perfectly match the valuation of the firm by those making use of accounting information. Accounting data are
designed to inform, rather than duplicate, market evaluation. 10
2.3
NIPA Saving and Investment
Gross investment in the national

income and product accounts is
the sum of net exports of goods
and services (as emphasized by
Eisner and Pieper 1989, a
measure of the accumulation of
claims on foreigners, not a
measure of the change in market
value of net claims on
foreigners), business
expenditures on fixed investment
(structures, including residential
structures, and producers'
durable equipment), and the
change in business inventories. If
we think of gross national
product as a flow of physical
goods and current services, we
360
361
can think of gross investment as

the portion of that flow devoted
to
Â
Page 445
adding to the stock of wealth. This may be an interesting measure; it is arguably the appropriate horizontal
axis on a marginal efficiency of investment schedule. (This is not the place to develop the point, but it may be
that a market value aggregate belongs in a production function for SHS income. When a firm purchases a
piece of real estate for a "revalued" price, presumably it expects to obtain as much extra value of output as it
does when it constructs a new building for the same amount.)
It is a further issue whether there is a useful aggregate, called the capital stock, that can be sensibly employed
in a production function. 11 The idea that there is such an aggregate that generates a flow of productive
services underlies the capital stock figures compiled by the Bureau of Economic Analysis. Although value
data provide the starting point, like GNP itself, the capital stock is conceived of as a physical quantity. The
depreciation estimates ("capital consumption allowances with capital consumption adjustment") in the NIPA
are intended to capture the loss over time in the current productive service flow potential embodied in the
accumulation of fixed investment. Other things equal, we might expect the profitable investment
opportunities to increase with increases in depreciation allowances, which would signal the need for
"replacement" investment. If this model captures the essence of the flow of investment opportunities, it is net
investment, not gross, that belongs on the horizontal axis of a marginal efficiency of investment schedule.
NIPA depreciation allowances

are not intended to represent the
decline in market value of the
assets in question and would not
do so even if there were no
measurement problems except
under very special assumptions
about the time path of discount
rates and about the way
productive capacity of the assets
declines over time. (Basically,
what it required is constancy of
discount rates and exponential
decay of productivity.12 The
actual rules used in constructing
the depreciation allowances are
rooted in studies of retirement
and other measures of physical
life.13
"Economic depreciation" is defined to be the decline in market value of a piece of equipment or a structure
between the beginning and end of the accounting period. As it happens, Hulten and Wykoff (1981) have
concluded that the U.S. Department of Commerce capital consumption estimates are reasonably similar to the
average historically experienced economic depreciation for a subset of assets for which there is an active
second-hand market. It is difficult to know,
361
362

Page 446
however, how relevant such ex post data on a subset of assets are to the forward-looking market valuation of
the bolted-down assets of business firms. A striking implication of the data in table 15.1 and figures 15.1 and
15.2, taken at face value, is that the NIPA capital consumption allowances for the nonfinancial corporate
sector differed sharply and unsystematically from economic depreciation over the 1948-87 period. 14 (I take
up below some of the reasons one might not take the figures at face value.)
3
Objections to the Use of Market-Value Measures of Saving
Various objections are sometimes raised to the use of asset market-value data, rather than NIPA measures, in
analyzing saving.
1. Asset markets are too volatile. They register paper gains and losses, not the steady accumulation of real
things.
To a degree that seems often unappreciated, the determinants of wealth are psychological. We need only be
reminded of Ponzi schemes and tulip manias, not to mention stock market crashes, to bring home how
dependent asset values are upon beliefs about the future. The modern literature on the rationality of
expectations and the efficiency of pricing in asset markets has emphasized in a refined way the unpleasant
difficulty of rooting asset values in "fundamentals."
Asset valuation is also inherently dependent upon the structure of information. I like to illustrate this
dependence with the case of a building that is destined to be destroyed by a meteor on a certain date. As long
as no one knows when and where the meteor will strike, the building has the same value as others like it. At
the moment the astronomers make public a prediction, the building loses value (to a degree dependent on the
distance into the future of the catastrophic event and on the confidence the public places in astronomers'
forecasts). It is clear that the owner of the structure suffers a fall in wealth at the point the information is
revealed, and presumably we would say that "society" suffers the same fall in wealth, even though, in a sense,
nothing is changed by the knowledge that causes the loss in value. The meteor was going to crash into the
building in any case.15
An interesting intermediate case arises if the information about the future is revealed only to the owner of the
building. (The analogous
Â
Page 447
situation is not unusualit gives rise to the "lemons" problem.) If he keeps the matter a secret and sells the
structure, he suffers no loss, nor is there any observable private or social loss until the meteor strikes.
As the examples suggest, the market value of assets has a kind of ephemeral quality that may, for example,
lead to doubts about the efficacy of capital markets as institutions of resource allocation. 16 Unfortunately, the
ephemeral quality of market assessments of value does not alter the role implied for them in economic theory.
Real risk and uncertainty about the future are apparent facts of life that cannot be avoided by focusing on
inputs that can be measured with relative precision. The purpose of asset measures produced by financial
accountants is to assist in the estimation of market values. The usual argument applies that the market price
will incorporate whatever information the accounting data contain. There is, presumptively, no money to be
362
363
made by betting on accounting net worth against the market.17
2. Asset market value changes incorporate price effects. What we need are real saving and wealth stock
concepts that are independent of discount rates and other relative asset value changes.
Various examples suggest the importance of taking into account price effects, especially in using wealth
measures to draw conclusions about welfare. One of the most important is the effect of changes in the
discount rate. At any moment the stock of claims to future goods and services is heterogeneous with respect
to the time and contingencies under which the claims pay off. When the prices of future consumption claims
change, so does the value of an unchanged stock of assets. In his discussion of the concept of income, Hicks
(1946) favored a wealth measure that would be unchanged if the steady-state level of consumption did not
change.
The increasing site value of land that we might expect to accompany population growth provides another
example. When the value of all houses (including mine) increases, I may be no better off, in spite of my
higher wealth, because I have to live somewhere. A third example was suggested to me by John Shoven:
discovery of a new technology that made computers of enormous power virtually cost-less and
instantaneously producible would render the existing stock of computers valueless (while we are at it, assume
that all software transfers costlessly to the new machines).
Â
Page 448
These are index number problems of the classic sort. 18 A financial accounting measure of saving appears
attractive in the particular instances because they seem to call for no change in the real-wealth measure in the
face of actual changes in market value. (I have not actually tried to sort out whether a real-wealth measure
would not change in the examples.) But this is surely fortuitous. Dealing with the index number problem
requires transforming market-value data, and it is only by chance that financial accounts may sometimes give
the right answer.
The discount-rate change

problem is a particularly
important one. When we assess
performance, it would make
sense to look at both wealth and
discount rate data. There is no
basis, however, for presuming
that financial accounting
measures of wealth perform
adequately as indices of real
wealth.
3. There are no reliable data on market value of wealth, therefore, we have to use the NIPA saving measures.
There may be problems with existing data on market values, although very extensive and accurate data are
available on assets such as corporate equities. The National Balance Sheets data seem to me an
underexploited resource. Furthermore, as in other contexts, an objection such as this one should be grounds
for devoting efforts to improving the data and to establishing the adequacy of the proxies we use if direct
measurements are not at hand.
4
Time-Series Data on Wealth at Market Value
363
364
Figures derived from the National Balance Sheets cast doubt on the adequacy of NIPA saving measures as a
proxy for changes in the market value of assets. Table 15.2 shows the time series of various wealth
aggregates. The nominal dollar figures have been reduced to common units using the implicit GNP deflator
(taking the average of fourth- and first-quarter values to approximate the year-end figure corresponding to the
balance sheet observations). The aggregate net worth of households includes the market valuation of
corporate shares and of land. The National Balance Sheets value fixed investment owned directly (in
unincorporated businesses and in the form of owner-occupied housing and consumer durables) at
replacement cost (using the NIPA data).19
The column titled "Government Net Worth" in table 15.2 is simply the aggregate debt of local, state, and
federal governments held by
Â
Page 449
the public (of course, it is a negative number). Government debt is, directly or indirectly, included on the
asset side of household balance sheets: to avoid double counting, the column headed ''Aggregate Wealth at
Market" sums the household and government net worth to produce an aggregate wealth measure. Notice that
no attempt at all has been made to evaluate the real asset position of governments. 20
The difference in aggregate wealth from one year to the next gives us "Aggregate Saving" in table 15.2.
Given what we know about the volatility of the stock and real property markets, we should expect significant
volatility in the wealth and saving measures, and we find it. Figure 15.3 displays the wealth time series
graphically, and figure 15.4 shows the saving series, normalized by dividing by GNP. For comparison, as
described numerically in table 15.3, figure 15.4 also displays the ratio of net national saving to GNP, a figure
derived from the national income and product accounts. As we might expect, the market-value measure is
much more variable than the NIPA measure. The measure based on the National Balance Sheets oscillates
over a range from a low of almost-15 percent to a high of almost 25 percent of GNP. The NIPA measure
drifts from a high of 10 percent in 1949 to a low of 2 percent in 1987. The two series are very different.
Fluctuations in market value are not all that accounts for the difference between the two measures. In
particular, the National Balance Sheets concept includes the stock of consumer durables in wealth. The
National Balance Sheets include estimates of the "consolidated net assets" of the United States, consisting of
the sum of reproducible assets (including consumer durables), land at market value, U.S. gold and special
drawing rights (SDRs), and certain claims on foreigners.21 Subtracting government debt and excluding land
from this total and taking the difference from year to year gives us a saving figure purged of market
revaluations. It consists mostly of reproducible assets: residential structures, nonresidential plant and
equipment, inventories, and consumer durables. It thus differs from NIPA net national saving mainly in
inclusion of consumer durables, and, in avoiding the inclusion of market revaluations, it is conceptually
directly comparable to NIPA saving. Indeed, the figures are taken from the Bureau of Economic Analysis
tangible-wealth tabulations. To emphasize that this hybrid series is derived from financial accounting data
(although it is far from the historical-cost book values on firms' balance sheets), I refer to it as "'Book'
Â
364
365
Page 450
Table 15.2
Household net worth and aggregate wealth, 1948-87
Government Aggregate wealth Aggregate
Net worth of U.S. households net worth at market saving
Year (millions $1982) (millions $1982) (millions $1982) (millions $1982)
1948 3,487,654 -857,494 2,630,160
1949 3,671,501 -889,339 2,782,162 152,002
1950 3,883,883 -819,665 3,064,218 282,056
1951 4,189,833 -790,067 3,399,766 335,548
1952 4,287,324 -795,849 3,491,475 91,709
1953 4,377,281 -825,250 3,552,031 60,556
1954 4,687,899 -840,255 3,847,644 295,613
1955 4,938,926 -814,882 4,124,044 276,400
1956 5,075,169 -780,335 4,294,834 170,791
1957 4,984,653 -770,061 4,214,592 -80,242
1958 5,427,604 -806,349 4,621,255 406,663
1959 5,571,610 -809,815 4,761,795 140,541
1960 5,680,642 -808,103 4,872,539 110,744
1961 6,086,197 -824,849 5,261,347 388,808
1962 5,928,471 -832,285 5,096,186 -165,161
1963 6,274,049 -833,201 5,440,848 344,662
1964 6,576,652 -837,694 5,738,958 298,110
1965 6,871,566 -823,076 6,048,490 309,532
1966 6,833,612 -806,728 6,026,885 -21,605
1967 7,370,297 -831,246 6,539,050 512,166
1968 7,827,453 -825,822 7,001,631 462,581
Â
Page 451
Government Aggregate wealth Aggregate

Net worth of U.S. households net worth at market saving A
Year (millions $1982) (millions $1982) (millions $1982) (millions $1982) t
1969 7,493,648 -790,676 6,702,972 -298,659 -
1970 7,432,952 -790,358 6,642,595 -60,377 -
1971 7,752,823 -817,570 6,935,254 292,659 1
365
366
1972 8,190,783 -815,700 7,375,083 439,829 16.9

1973 7,889,046 -758,254 7,130,791 -244,291 -8.9
1974 7,457,661 -712,663 6,744,998 -385,793 -14.1
1975 7,830,318 -794,411 7,035,907 290,909 10.8
1976 8,348,919 -834,194 7,514,726 478,819 16.9
1977 8,642,746 -837,706 7,805,040 290,315 9.8
1978 9,111,741 -816,502 8,295,239 490,198 15.7
1979 9,631,709 -784,075 8,847,635 552,396 17.3
1980 10,046,585 -790,784 9,255,800 408,166 12.8
1981 10,064,616 -812,292 9,252,323 -3,477 -0.1
1982 10,061,786 -925,358 9,136,427 -115,896 -3.7
1983 10,544,681 -1,067,194 9,477,487 341,060 10.4
1984 10,731,277 -1,188,518 9,542,759 65,272 1.9
1985 11,372,752 -1,328,645 10,044,108 501,349 13.9
1986 11,907,562 -1,473,725 10,433,837 389,729 10.5
1987 12,257,233 -1,596,916 10,660,317 226,480 5.9
Source: See text. Based on Board of Governors of the Federal Reserve System (1984); U.S. Department of Commerce (1986, 1987).
Â
Page 452
FigureÂ 15.3
HouseholdÂ andÂ aggregateÂ wealth,Â 1948-87,Â ratioÂ toÂ GNP
less Land" in figure 15.5 (fig. 15.5 simply adds the new series to fig. 15.4).
Although the resulting series is smoother than that of aggregate wealth, significant differences from NIPA
national saving remain. Exploration of the reasons for the remaining differences would be a side excursion
366
367
from my principal line of argument. The evidence from the National Balance Sheets data clearly supports the
conclusion that financial accounting saving misses significant amounts of the value change that is revealed in
asset markets.
5
Caveats on the National Balance Sheets Wealth Figures
Several problems with the National Balance Sheets data should be recognized.
1. The market value of equity incorporates the capitalized value of certain variations in tax liabilities that are
not balanced by offsetting measured asset values. An instance is the "trapped-equity" problem. 22 Corporate
payouts in the form of dividends are subject to tax
Â
Page 453
FigureÂ 15.4
MarketÂ valueÂ andÂ NIPAÂ saving,Â 1949-87,Â ratioÂ toÂ GNP
at the shareholder level, and shareholders ought to discount this tax in bidding for shares. A considerable (and
inconclusive) literature now exists developing the technical ins and outs of the tax and securities law and
practice in relation to the trapped-equity argument. To the extent that dividend taxes are discounted in the
price of equity, the value of a corporation's shares will be below the market value of the assets owned by the
firm.
Another instance is the value of tax liabilities accrued by corporations via such tax rules as accelerated
depreciation. An increase in such accruals ought to lower the value of corporate equities. 23
A possible third instance is the tax consequence of changing corporate financial structure. The tax system has
set up incentives, which have varied through time, bearing on the choice between debt and equity. One view
of the current intense leveraged-buyout activity in the United States is that it is strongly motivated by such
tax considerations, and the gradual realization of the private profit (at the expense of public revenue) to be
made by financial restructuring accounts for some of the bidding up of equity prices.
367
368

Page 454
Table 15.3
Net national saving in the United States, NIPA basis, 1948-87, ratio to GNP (in percentages)
Year Gross saving Capital consumption allowances Net saving
1948 19.4 7.8 11.6
1949 14.0 8.4 5.6
1950 18.2 8.2 10.0
1951 17.6 8.2 9.4
1952 14.9 8.3 6.6
1953 13.7 8.3 5.4
1954 13.9 8.7 5.1
1955 16.9 8.5 8.4
1956 18.1 8.9 9.2
1957 17.1 9.1 8.0
1958 14.1 9.4 4.8
1959 16.2 9.0 7.2
1960 16.3 9.0 7.3
1961 15.5 9.0 6.5
1962 15.9 8.6 7.3
1963 16.3 8.5 7.8
1964 16.7 8.3 8.4
1965 17.5 8.1 9.4
1966 16.9 8.0 8.8
1967 15.9 8.3 7.6
1968 15.6 8.3 7.4
1969 16.5 8.4 8.0
1970 15.2 8.7 6.5
1971 15.6 8.8 6.7
1972 16.5 8.9 7.7
1973 18.5 8.7 9.8
1974 16.8 9.3 7.5
1975 14.9 10.1 4.8
1976 15.9 10.1 5.8
1977 16.9 10.1 6.7
1978 18.2 10.2 7.9
1979 18.3 10.6 7.7
1980 16.3 11.1 5.2
368
369
1981 17.1 11.4 5.7

1982 14.1 12.1 2.0
1983 13.6 11.6 2.0
1984 15.1 11.0 4.1
1985 13.3 10.9 2.4
Â
Page 455
Table 15.3
(continued)
Year Gross saving Capital consumption allowances Net saving
1986 12.7 10.8 1.9
1987 12.4 10.6 1.8
Sources: 1948-84: Economic Report of the President, February 1988; 1985-87; Survey of Current Business, July
1988.
FigureÂ 15.5
ComparisonÂ ofÂ savingÂ measures,Â 1948-87,Â ratioÂ toÂ GNP
There is, in all of these instances, a balancing asset "owned" by the public through the public's "ownership"
of the government, which we might describe as accrued tax liability. Unfortunately, however, we cannot
observe the value of this asset in the market, and so the empirical problem does not go away with aggregation
across sectors.
2. Anticipated tax claims are also important in assessing pension reserve assets, which are viewed as
belonging to households. Presumably, the great bulk of these claims is subject to income taxation upon
distribution. When household and government financial claims are netted in reaching a national wealth
369
370
figure, this problem goes away.
Â
Page 456
3. As Auerbach (1985) has emphasized, unfunded pension liabilities of corporations represent unmeasured
assets of the households that are presumably offset by an effect on measured corporate equity value in the
market. This component of wealth is missed in the National Balance Sheets.
4. Debt is carried on the National Balance Sheets at book value. Corporate debt liabilities are thus incorrectly
valued. Correcting for inflation, of course, is relatively easy. But there is also a divergence between book and
market value in current dollars that varies through time. Tax incentives plus simple changes in the nominal
discount rates result in such divergences. Furthermore, the leveraged buyout wave may be responsible for a
systematic divergence between book and market valuation of debt. The large premiums paid for equity claims
in corporate takeovers are sometimes explained by the implied expropriation of the interests of bondholders.
The value of the bonds of RJR Nabisco is said to have fallen by 20 percent as a consequence of the successful
takeover of the firm in a leveraged buyout in December 1988.
It might be thought that the misstatement of the value of bonds as liabilities on the books of corporations
would be balanced by their misstatement as assets in the hands of the public in an aggregation across sectors.
This would be so if the aggregation were in terms of financial accounting concepts. But aggregation to
national saving will sum the market values of equity with the book value of debt. To correct for this problem
will require gathering data on the market value of bonds. (Brainard, Shoven, and Weiss, 1980 have developed
such estimates for the debt of a large population of U.S. corporations.)
5. I have mentioned above the likelihood that some of the recent increase in equity value has come at the
expense of bondholders and of the government (through lost tax revenues otherwise expected). Shleifer and
Summers (1987) have suggested that other "stakeholders" in corporations have also lost wealth in the wave of
corporate acquisitions. We would probably describe the wealth effects on noncorporate, nonbondholder
stakeholders as impacts on human capital; the effects are in any case presumably not reflected in asset market
data.
6. The National Balance Sheets present no estimates of the market value of businesses owned directly by
households. The data in table
Â
Page 457
15.1 show a large and variable divergence between book and market values of property owned by
corporations. There is no obvious reason there should not be a similar degree of divergence in the valuation
of noncorporate firms.
7. The Flow of Funds staff of the Federal Reserve Bank expresses reservations about the adequacy of the
estimated market value of land, which is built up using ratios of assessed to market values from real estate tax
administration reported in censuses of governments taken once every five years. I have no independent basis
for evaluating these reservations. (Corporate holdings are presumably captured in equity values, but
corporations own a small fraction of U.S. land.)
370
371
6
The Saving Performance of the United States
It is usual to assess aggregate saving behavior by reference to saving "rates," ratios of saving to aggregate
income. Although dividing the aggregate saving by a national income measure is a natural method of
normalizing for the size of the economy, one should be cautious in drawing conclusions about economic
performance from trends in, or comparisons across countries of, such ratios. Saving rates thus defined do not
obviously relate to the objective of assessing the level of aggregate consumption against a standard either of
consistency with past behavior or of prudence with respect to future welfare. For these purposes, measures of
wealth per capita are called for or, more generally, measures of the wealth of various subgroups in the
population. 25
Table 15.4 displays wealth per capita data for the United States, where wealth is interpreted in the National
Balance Sheets sense of household net worth (at market value) minus government debt. Saving per capita is
simply the first difference of wealth per capita, and thus incorporates population growth. Figure 15.6 displays
the saving series expressed as the year-to-year growth of wealth per capita (labeled "Growth in Wealth per
Capita" in table 15.4). Because wealth is a stochastic variable, a particular year's experience conveys limited
information.
It is not clear what one should regard as either a normal or a "good" rate of increase in wealth per capita. If
productivity were stationary we would probably expect wealth per capita to be constant, and welfare
considerations would also presumably prescribe
Â
Page 458
Table 15.4
Per capita wealth and saving at market value, 1948-87
Wealth per Saving per Growth in wealth
Year capita ($1982) capita ($1982) per capita (%)
1948 17,937
1949 18,649 711 4.0
1950 20,123 1,475 7.9
1951 21,951 1,828 9.1
1952 22,161 209 1.0
1953 22,175 14 .1
1954 23,601 1,427 6.4
1955 24,854 1,253 5.3
1956 25,428 574 2.3
1957 24,506 -922 -3.6
1958 26,425 1,919 7.8
1959 26,777 352 1.3
1960 26,969 192 .7
1961 28,642 1,673 6.2
1962 27,320 -1,323 -4.6
371
372
1963 28,751 1,431 5.2

1964 29,908 1,157 4.0
1965 31,129 1,221 4.1
1966 30,662 -467 -1.5
1967 32,907 2,245 7.3
1968 34,885 1,978 6.0
1969 33,072 -1,813 -5.2
1970 32,395 -678 -2.0
1971 33,397 1,002 3.1
1972 35,137 1,740 5.2
1973 33,650 -1,487 -4.2
1974 31,540 -2,110 -6.3
1975 32,578 1,038 3.3
1976 34,466 1,888 5.8
1977 35,439 973 2.8
1978 37,268 1,829 5.2
1979 39,313 2,045 5.5
1980 40,639 1,326 3.4
1981 40,203 -436 -1.1
1982 39,293 -910 -2.3
1983 40,364 1,071 2.7
1984 40,265 -100 -.2
1985 41,977 1,712 4.3
Â
Page 459
Table 15.4
(continued)
Wealth per Saving per Growth in wealth
Year capita ($1982) capita ($1982) per capita (%)
1986 43,184 1,208 2.9
1987 43,705 521 1.2
Sources: See text. Based on Board of Governors of the Federal Reserve System (1984); U.S. Department of
Commerce (1986, 1987).
372
373
FigureÂ 15.6
GrowthÂ inÂ netÂ worthÂ lessÂ governmentÂ debtÂ perÂ capita,Â 1949-87
constancy. In general, both predicted and optimal accumulation would be related to technological progress
and demographic structure. As shown in figure 15.6, there appears to be a long-term declining trend to the
rate of growth of real wealth per capita. Interestingly, the performance of the most recent three years is on or
slightly above trend.
For those looking for good news (bearing in mind the caveats mentioned above about the use of wealth as a
measure of welfare), figure 15.7 displays the trend in real wealth per capita. The picture shows that, on
average, since 1948 U.S. residents have been adding
Â
Page 460
FigureÂ 15.7
HouseholdÂ netÂ worthÂ lessÂ governmentÂ debt,Â 1982Â dollarsÂ perÂ capita
373
374
to the stock of wealth per

capita about $700 (1987 price
level) per year. According to
figure 15.7, the current level
of wealth per capita is just a
bit above its long-term trend.
7
Conclusion
Although the NIPA saving measures, and especially NIPA saving rates, are widely used in both scholarly and
journalistic treatments, their shortcomings as representations of the saving concepts derived from economic
analysis should not be controversial among economists. Saving is the change in a stock of wealth. NIPA
saving describes the change in a cost-based measure of some past resource commitments. Households,
individually and in the aggregate, measure their situations instead by reference to a forward-looking
assessment of the success or failure of those and other resource commitments. These assessments find
expression in the capital market's valuation of enterprises, broadly conceived. The annual change in that
value is the measure of saving.
Â
Page 461
Whatever their usefulness as measures of a certain class of inputs, the NIPA saving and wealth measures are
not good proxies for the market-expressed assessments of results. The National Balance Sheets present the
conceptually appropriate measures of national wealth and saving. It is clear, though, that much needs to be
done to improve the quality of the statistics and to refine their interpretation.
Notes
This is a revision of a paper prepared for an NBER conference on saving held January 6-7, 1989. The author
would like to thank Alan Auerbach, William Beaver, Michael Boskin, John Campbell, Angus Deaton,
Bronwyn Hall, Robert Hall, Alan Huber, Robert Lipsey, James Poterba, Robert Shiller, John Shoven, Scott
Smart, Frederick Yohn, Jr., and conference participants for helpful discussions of various aspects of this
research and Kathleen Much for editorial advice. This paper was completed while the author was a fellow at
the Center for Advanced Study in the Behavioral Sciences, Stanford, California. The author is grateful for
financial support provided by Princeton University; the John M. Olin Program for the Study of Economic
Organization and Public Policy, Princeton University; National Science Foundation grant BNS87-00864; the
Alfred P. Sloan Foundation; and the National Bureau of Economic Research.
1. For discussions of the SHS income concept, see Bradford (1986) or Institute for Fiscal Studies (1978).
2. For this purpose, owner-occupiers can be thought of as in the business of providing housing services. Other
household-owned and household-employed capital (consumer durables) is excluded from the NIPA
investment and capital concepts, but that is not my main concern here.
3. To derive the aggregate accounting net worth of the corporate sector, I have added the net worth of
corporate farms (line 46 of the Sector Balance Sheet for the Non-financial Business Sector) to the nonfarm,
nonfinancial total (line 43).
4. The importance of intertemporal prices (interest rates) is often overlooked in assessments of welfare.
Summers (1983) develops a cost-of-living series corrected for interest rate changes, applicable to a person
with a given amount of wealth (and no anticipated earnings).
374
375
5. For examples of more refined intertemporal models see Breeden (1979), Campbell and Deaton (1988),
Ingersoll (1987, Chap. 11), Merton (1971, 1973).
6. For a classic example of such a model, see Ando and Modigliani (1963). For recent examples, see Blinder
and Deaton (1985), Deaton (1987), Hall (1978, 1988), West (1988).
7. See Beaver and Ryan (1985).
8. See Gibson and Frishkoff (1986, 44).
9. See Gibson and Frishkoff (1986, 46).
10. See Foster (1986) for a survey of the accounting literature on the information content of financial
statements.
Â
Page 462
11. For an overview see Brown (1980).
12. See the clear discussion in Hulten and Wykoff (1981).
13. See U.S. Department of Commerce, Bureau of Economic Analysis (1987); also Musgrave (1983, 1986a,
1986b).
14. Using National Balance Sheets data, Bulow and Summers (1984) have emphasized this point in their
discussion of the failure of income tax rules to recognize wealth changes in the form of asset revaluations.
They suggest that the ex ante depreciation allowances for tax purposes should be increased to compensate the
investor for the risk of asset revaluations that are unrecognized by the tax rules.
15. James Poterba has reminded me that, quite apart from discounting, the aggregate market value of wealth
may not fall by the full prior value of the doomed building when the meteor news arrives. The aggregate
value will depend upon the general equilibrium response of all asset prices, even if the asset in question is a
tiny part of the aggregate stock. Bradford (1978) illustrates the point.
16. See Stiglitz (1972, 1979).
17. Summers (1986) has emphasized how difficult it may be to establish the ''rationality" of asset markets,
i.e., to tell whether one can make money by selling short when prices are too high by some internal standard.
But presumably those who would use NIPA saving figures rather than asset market values are not talking
about small, hard-to-detect, effects.
18. Pollak (1975) has worked out the index number theory applicable to an inter-temporal setting. See also
Summers (1983).
19. The figures for household net

worth (sector basis) included in
this paper incorporate an
adjustment to deal with an error
discovered in the course of this
work by Frederick O. Yohn, Jr.,
of the Flow of Funds section of
the Federal Reserve Board. In
the published series, household
claims on noncorporate private
375
376
financial institutions have been

omitted from household net
worth. I have added the
"approximate share of
noncorporate companies" in the
net worth of the private financial
institution sector (line 50 in the
Sector Balance Sheet of Private
Financial Institutions) to the
published household sector net
worth.
20. Boskin, Robinson, and Huber (1987) and Eisner (1986) have developed government real asset series.
21. Perhaps because it is not clear how one would allocate accounting values, the National Balance Sheet's
"total consolidated net assets" of the United States excludes U.S. holdings of foreign equities and makes no
deduction for foreign holdings of U.S. equities (other than via direct investment). The household sector net
worth does include holdings of foreign equities. The two wealth concepts are thus not quite parallel.
22. See Auerbach (1979, 1983b) and Bradford (1981).
23. Auerbach (1983a, 1989) and Auerbach and Hines (1987) show that the capitalized value effects of tax
law changes can be large.
24. Kotlikoff (1984, 1986, 1988) has emphasized a similar point with respect to assessment of the national
debt.
Â
Page 463
References
Ando, Albert, and Franco Modigliani. 1963. The "Life-Cycle" Hypothesis of Saving: Aggregate Implications
and Tests. American Economic Review 53 (May): 55-84.
Auerbach, Alan J. 1979. Wealth Maximization and the Cost of Capital. Quarterly Journal of Economics 93:
433ff.
. 1983a. Corporate Taxation in the U.S. Brookings Papers on Economic Activity, No. 2.
. 1983b. Taxation, Corporate Financial Policy, and the Cost of Capital. Journal of Economic Literature 21:
905-40.
. 1985. Saving in the U.S.: Some Conceptual Issues In The Level and Composition of Household Saving, ed.
Patric Hendershott. Cambridge, Mass.: Ballinger.
. 1989. Tax Reform and Adjustment Costs: The Impact on Investment and Market Value. International
Economic Review 30 (November): 939-62.
Auerbach, Alan J., and James R. Hines, Jr. 1987. Anticipated Tax Changes and the Timing of Investment. In
The Effects of Taxation on Capital Accumulation, ed. Martin Feldstein. Chicago: University of Chicago
Press.
Beaver, William H., and Stephen Ryan. 1985. Do Statement Thirty-three Earnings Explain Stock Prices?
376
377
Financial Analysts Journal 61, no. 5 (September-October): 66-71.
Blades, Derek W., and Peter Sturm. 1982. The Concept and Measurement of Savings: The United States and
Other Industrialized Countries. In Saving and Government Policy. Boston: Federal Reserve Bank of Boston.
Blinder, Alan S., and Angus S. Deaton. 1985. The Time Series Consumption Function Revisited. Brookings
Papers on Economic Activity, no. 2, 465-511.
Board of Governors of the Federal Reserve System. 1988. Balance Sheets for the U.S. Economy, 1948-87.
Washington, D.C., October.
Boskin, Michael J. 1986. Theoretical Issues in the Measurement, Evaluation and Interpretation of Postwar
U.S. Saving. In Savings and Capital Formation, ed. F. Gerald Adams and Susan M. Wachter, 11-43.
Lexington, Mass.: Lexington Books.
. 1988. Issues in the Measurement and Interpretation of Saving and Wealth. NBER Working Paper no. 2633.
Cambridge, Mass., June.
Boskin, Michael J., and Lawrence J. Lau. 1988. An Analysis of Postwar U.S. Consumption and Saving.
Center for Economic Policy Research, Stanford, Calif., August. Mimeograph.
Boskin, Michael J., Marc S. Robinson, and Alan M. Huber. 1987. Government Saving, Capital Formation
and Wealth in the United States, 1947-85. In The Measurement of Saving, Investment, and Wealth, ed. Robert
E. Lipsey and Helen Stone Tice, 287-353. NBER Studies in Income and Wealth, vol. 52. Chicago: University
of Chicago Press.
Bradford, David F. 1978. Factor Prices May Be Constant but Factor Returns Are Not. Economics Letters 1:
199-203.
. 1981. The Incidence and Allocational Effects of a Tax on Corporate Distributions. Journal of Public
Economics 15: 1-22.
Â
Page 464
. 1986. Untangling the Income Tax. Cambridge, Mass.: Harvard University Press.
Brainard, William C., John B. Shoven, and Laurence Weiss. 1980. Financial Valuation of the Return to
Capital. Brookings Papers on Economic Activity no. 2, 453-511.
Breeden, Douglas T. 1979. An Intertemporal Asset Pricing Model with Stochastic Consumption and
Investment Opportunities. Journal of Financial Economics 7 (September): 265-96.
Brown, Murray. 1980. The Measurement of Capital Aggregates: A Postreswitching Problem. In The
Measurement of Capital, ed. Dan Usher, 377-420. NBER Studies in Income and Wealth, vol. 45. Chicago:
University of Chicago Press.
Bulow, Jeremy I., and Lawrence H. Summers. 1984. The Taxation of Risky Assets. Journal of Political
Economy 42: no. 1 (February): 20-38.
Campbell, John. 1987. Does Saving Anticipate Declining Labor Income? Econometrica 55, no. 6
(November): 1249-73.
Campbell, John, and Angus Deaton. 1988. Why Is Consumption So Smooth? Woodrow Wilson School,
Princeton University, January. Mimeograph.
377
378
Deaton, Angus. 1987. Consumers' Expenditure. In The New Palgrave: A Dictionary of Economics, ed. John
Eatwell, Murray Milgate, and Peter Newman, 1: 592-604. New York: Stockton.
Eisner, Robert. 1980. Capital Gains and Income: Real Changes in the Value of Capital in the United States,
1946-77. In The Measurement of Capital, ed. Dan Usher, 175-344. NBER Studies in Income and Wealth, vol.
45. Chicago: University of Chicago Press.
. 1986. How Real Is the Federal Deficit? New York: Free Press.
. 1988. Extended Accounts for National Income and Product. Journal of Economic Literature 26 (December):
1611-84.
. 1989. Divergences of Measurement and Theory and Some Implications for Economic Policy. American
Economic Review 79, no. 1 (March): 1-13.
Eisner, Robert, and Paul J. Pieper. 1989. The World's Greatest Debtor Nation. Typescript. Department of
Economics, Northwestern University, January 13.
Financial Accounting Standards Board. 1978. Objectives of Financial Reporting by Business Enterprises.
FASB Statement of Financial Accounting Concepts no. 1. FASB, Stamford, Conn.
Foster, George. 1986. Financial Statement Analysis, 2d ed. Englewood Cliffs, NJ: Prentice-Hall.
Gibson, Charles H., and Patricia A. Frishkoff. 1986. Financial Statement Analysis, 3d ed. Boston, Mass.:
Kent Publishing.
Goldsmith, Raymond W. 1982. The National Balance Sheet of the United States, 1953-1980. Chicago:
University of Chicago Press.
Hall, Robert E. 1978. Stochastic Implications of the Life Cycle-Permanent Income Hypothesis: Theory and
Evidence. Journal of Political Economy 86 (December): 971-87.
. 1988. Intertemporal Substitution in Consumption. Journal of Political Economy 96, pt. 2: 339-57.
Â
Page 465
Hicks, John R. 1946. Value and Capital, 2d ed. Oxford: Clarendon.
Hulten, Charles R., and Frank C. Wykoff. 1981. The Measurement of Economic Depreciation. In
Depreciation, Inflation and the Taxation of Income from Capital, ed. Charles R. Hulten, 81-125. Washington,
D.C.: Urban Institute.
Ingersoll, Jonathan E. 1987. Theory of Financial Decision Making. Totowa, N.J.: Rowman & Littlefield.
Institute for Fiscal Studies. 1978. The Structure and Reform of Direct Taxation: The Report of a Committee
Chaired by Professor J. E. Meade. London: George Allen & Unwin.
Kotlikoff, Laurence J. 1984. Taxation and Savings: A Neoclassical Perspective. Journal of Economic
Literature 22 (December): 1576-1629.
. 1986. Deficit Delusion. The Public Interest 84 (Summer): 53-65.
. 1988. The Deficit Is Not a Well-Defined Measure of Fiscal Policy. Science 241 (August): 791-95.
378
379
Lipsey, Robert E., and Irving B. Kravis. 1987. Is the U.S. a Spendthrift Nation? NBER Working Paper no.
2274. Cambridge, Mass., June.
Lipsey, Robert E., and H. S. Tice, eds. 1989. The Measurement of Saving, Investment, and Wealth. NBER
Studies in Income and Wealth, vol. 52. Chicago: University of Chicago Press.
Merton, Robert C. 1971. Optimum Consumption and Portfolio Rules in a Continuous-Time Model. Journal
of Economic Theory 3, no. 4 (December): 373-413.
. 1973. An Intertemporal Capital Asset Pricing Model. Econometrica 41, no. 5 (September): 867-87.
Musgrave, John C. 1983. Fixed Reproducible Wealth in the United States, 1972-82. Survey of Current
Business 63, no. 8 (August): 62-65.
. 1986a. Fixed Reproducible Tangible Wealth in the United States, 1972-82, Revised Estimates. Survey of
Current Business 66, no. 1 (January). 51-75
. 1986b. Fixed Reproducible Tangible Wealth in the United States, 1982-85. Survey of Current Business 66,
no. 8 (August): 36-39.
Obstfeld, Maurice. 1986. Capital Mobility in the World Economy: Theory and Measurement.
Carnegie-Rochester Conference Series on Public Policy 24 (Spring): 55-103.
Peek, Joe. 1986. Household Wealth Composition: The Impact of Capital Gains. New England Economic
Review (November/December): 26-39.
Pollak, Robert A. 1975. The Intertemporal Cost of Living Index. Annals of Economic and Social
Measurement 4, no. 1 (Winter): 179-95.
Poterba, James M. 1987. Tax Policy and Corporate Savings. Brookings Papers on Economic Activity, no. 2
(December): 455-503.
Poterba, James M., and Lawrence H. Summers. 1987. Recent U.S. Evidence on Budget Deficits and National
Savings. NBER Working Paper no. 2144. Cambridge, Mass., February.
Â
Page 466
Ruggles, Richard, and Nancy Ruggles. 1981. Integrated Economic Accounts for the United States,
1947-1980. Institute for Social and Policy Studies Working Paper no. 84. Yale University, New Haven,
Conn., October.
Shleifer, Andrei, and Lawrence

H. Summers. 1987. Breach of
Trust in Hostile Takeovers.
NBER Working Paper no.
2342. Cambridge, Mass.,
August.
Shoven, John B 1984. Saving in the U.S. Economy. In Removing Obstacles to Economic Growth, ed.
Michael L. Wachter and Susan M. Wachter, 187-223. Philadelphia: University of Pennsylvania Press.
Stiglitz, Joseph E. 1972. On the Optimality of the Stock Market Allocation of Investment. Quarterly Journal
of Economics (February): 25-60.
379
380
. 1979. The Inefficiency of the Stock Market Equilibrium. Princeton University, April. Mimeograph.
Summers, Lawrence H. 1983. Observations on the Indexation of Old Age Pensions. In Financial Aspects of
the U.S. Pension System, ed. Zvi Bodie and John B. Shoven, 231-51, 257-58. Chicago: University of Chicago
Press.
. 1985. Issues in National Savings Policy. NBER Working Paper no. 1710. Cambridge, Mass., August.
. 1986. Does the Stock Market Rationally Reflect Fundamental Values? Journal of Finance 41, no 3 (July):
591-601.
Summers, Lawrence H., and Chris Carroll. 1987. Why Is U.S. National Saving So Low? Brookings Papers
on Economic Activity, no. 2, 607-35.
U.S. Department of Commerce, Bureau of Economic Analysis. 1986. The National Income and Product
Accounts of the United States, 1929-82. Washington, D.C.: Government Printing Office, September.
. 1987. Fixed Reproducible Tangible Wealth in the United States, 1925-85. Washington, D.C.: Government
Printing Office, June.
Â
Page 467
16
What is National Saving? Alternative Measures in Historical and International
Context
Introduction
Among policymakers and

commentators there is a virtual
consensus today that the United
States saves too little, and there is
only scattered disagreement
about whether something can and
should be done about it. But
journalistic analysis, if not
professional opinion, is actually
based on a rather unreflective
reliance on one, or perhaps two,
statistical measures of national
saving behavior. It would assist
intelligent discussion of these
issues to have in view the
multiple meanings of the term
saving. My object here is to lay
out some of the main alternative
conceptions of saving and their
interpretation and to present data
comparing recent U.S. saving
380
381
behavior with its own past and

with the saving behavior of other
nations.
National Income,
Consumption, and
Saving
A good place to start is with the identity ℜϒ = C + S, income equals the sum of consumption and saving. The
simplicity of this identity is apparent but is not real. All it accomplishes is to establish a relationship among
three measures of economic performance and thereby to imply that when an empirical or accounting content
is given to any two of the concepts, the empirical or accounting content of the third is implied. National
income accounting generally starts with definitions of income and consumption (by households and
governments) from which saving is defined by subtraction. All beginning students of economics learn that
saving is a residual.
Analysis of U.S. national saving is quite likely to start with the income data and definitions of the U.S.
Department of Commerce's national income and product accounts (NIPAs). 1 To make life
Â
Page 468
confusing, the term income in the NIPAs is attached to factor payments and distinguishes between taxes that
fall on factor payments and taxes that do not (indirect business taxes). It is doubtful that there is an
economically meaningful distinction between taxes that bear on factor payments and those that do not. When
speaking of the nation, I here use the terms income and product interchangeably to mean "product" (gross or
net) as measured in the NIPAs. Instead of income, consumption, and saving, the three related notions in the
NIPA context are product, consumption, and saving. Which two of the three are fundamental (in determining
the third as a residual) in the case of the NIPAs is not immediately obvious.
National income and product accounts are primarily focused on measuring output, of which there are two
major conceptions: gross and net. The beginning student's deceptively simple truism applies to the concepts
of gross national income (or product) and gross national saving. Gross national product (GNP)"the market
value of the goods and services produced by labor and property supplied by residents of the United States"
2and consumptionpersonal and governmentcan reasonably be described as fundamental ideas. Together (by
subtraction) they define gross saving.
Gross national product may perhaps be of behavioral interest because of its relationship to the aggregate
employment of labor. But it has always been recognized, and more advanced students understand, that a
gross measure of output is defective as a measure of performance because it ignores the using up of national
capital and that gross saving is purely an accounting residual concept with no normative or behavioral
interpretation. As measures of economic performance, more advanced students look to net national product
(NNP) (or income) and net national saving. To reach net product, net investment, and net saving, it is
necessary to subtract an allowance for the using up of the reproducible capital stock. Here, the defining ideas
are consumption and net investment (or saving), the latter understood as the increase in the nation's capital
stock. We will have it about right if we think of net product (income) as definitionally equal to the sum of
consumption (personal and governmental) and saving (the change in the reproducible capital stock owned by
U.S. residents).
It is thus true that net national saving is equal to the difference between net national product (income) and
national consumption, but it is misleading to describe net saving as a residual because net
381
382
Â
Page 469
saving has a normative and behavioral interpretation as the change in the nation's capital stock. The
interpretation is normative because we regard more capital as desirable, and it is behavioral because we think
people and businesses seek to optimize the level of capital. Indeed, it is reasonable to describe net national
product (income) as the piece of the identity that is defined residually (as the sum of consumption and net
saving), even if this is disguised by the way the statisticians describe the derivation of net national product as
gross national product less the depreciation of the capital stock (capital consumption allowances).
Whereas gross saving does not have normative content, net saving does. It measures the increment to the
nation's capital stock. When we speak of saving as an indicator of economic performance, we have in mind
this quality: that it measures a change in a stock of wealth or productive potential. The difference among
alternative measures of saving from a policy perspective consists of differences among the stocks of wealth
or capital to which the saving is an increment.
Wealth and Capital
The Flow of Funds Division of the Board of Governors of the Federal Reserve System compiles balance
sheets for the various private sectors of the U.S. economy. Table 16.1 gives a summary of the figures on the
national net worth for year end 1988. Note that the figures do not include the assets of governments, whereas
the financial liabilities of governments are in effect treated as national liabilities. Another, less obvious,
feature of the data is the treatment of asset location. "Domestic net worth" consists of reproducible assets plus
land located in the United States. The reproducible assets other than consumer durables constitute the stock
that is augmented each year by "net private domestic investment" in the NIPAs. To obtain a figure
representative of the aggregate of the capital owned by U.S. residents, it is necessary, first, to add assets
located abroad but owned by U.S. residents ("direct foreign investment'' plus financial claims on foreigners,
including portfolio equity); second, to subtract the corresponding foreign claims against U.S. assets; and
third, to add the U.S. stock of gold and special drawing rights (SDRs). Note, too, that except for land and net
international portfolio investment in corporate equities, the assets are carried on the national balance
Â
Page 470
Table 16.1
National net worth of the United States, year end 1988
Millions of Percent distributions
dollars
Domestic net worth 14,964,435 104 100
Reproducible assets 11,410,041 76 100
Residential structures 4,235,330 37
Nonresidential plant and equipment 4,364,789 38
Inventories 1,003,997 9
Consumer durables 1,805,925 16
382
383
Land at market 3,554,394 24

Net claims on foreign -558,463 -4
Foreign assets, U.S. owned 738,367
U.S. assets, foreign owned -1,296,830
U.S. gold and SDRs 20,789 0
Total consolidated national net assets 14,426,761 100
Source: Board of Governors, Federal Reserve System, Balance Sheets for the U.S. Economy, Flow of Funds
publication C.9 (October 1989).
sheet on a "financial accounting basis," by which I mean to draw the analogy with the financial accounting
for a business firm and to contrast with estimated market value. 3 "Financial accounting" for assets is often
the only method available, and I do not mean to imply by the term a crude historical cost record. The Federal
Reserve data on reproducible assets are the U.S. Department of Commerce ''current" (that is, replacement)
cost estimates.4 For this reason, the "total consolidated national net assets" figure in Table 16.1 is not
identical to the aggregate wealth represented by those assets. I will say more on this point.
Some of the components of total consolidated net assets can be considered more closely. According to the
Federal Reserve Board, as of the end of 1988, the "net claims on foreign" item was negative. This is roughly
the same as saying that U.S. residents owed more to foreigners than vice versa, but it is important to keep in
mind that the direct investment stocks are valued at replacement, rather than at market, value. Land value is
estimated to have accounted for almost one-quarter of the tangible assets located in the United States. Of the
reproducible assets, housing and industrial plant and equipment are of almost equal importance, together
accounting for 75 percent.
Â
Page 471
Table 16.2
Wealth of U.S. households at market value, year end 1988
Millions of Percent
dollars distributions
Household-owned tangible assets 6,560,104 48 100
Owner-occupied structures and land 4,388,381 67
Other structures and nonprofit tangible 3,185,804 49
Consumer durables 1,805,925 28
Net financial assets 7,205,731 52
Net household financial assets 9,334,094
Total private financial assets 12,600,638 100
Deposits and credit market instruments 4,543,226 36
Corporate equities 2,242,924 18
Life insurance and pension fund reserves 2,906,337 23
Equity in noncorporate businessa 2,654,300 21
Miscellaneous other 253,851 2
383
384
Total private liabilities 3,266,544

Public sector financial assets less liabilities 2,128,363
National wealth at market value 13,765,835 100
Memo: Net household-owned assets 15,894,198
Source: Board of Governors, Federal Reserve System, Balance Sheets for the U.S. Economy, Flow of Funds
publication C.9 (October 1989).
a. Household equity in noncorporate business includes equity in noncorporate private financial institutions,
omitted in the Federal Reserve's household sector balance sheet.
Inventories account for almost 10 percent, and consumer durables, at 16 percent, make up the rest.
Table 16.2 presents data on the aggregate wealth of U.S. households (including nonprofit institutions). The
household totals include all the assets indirectly owned by households via life insurance and pension reserves.
The totals therefore include government financial obligations, directly or indirectly owned, and they make no
allowance for the implicit liability of U.S. taxpayers to foreign holders of U.S. government financial
obligations. A better measure of national wealth is therefore obtained by netting out these financial liabilities,
yielding the figure for "national wealth at market value" shown in Table 16.2.
Where possible, the market values of assets have been used in Table 16.2, in particular, for land and
corporate equity. For this reason, as well as slightly different coverage, the national wealth in Table
Â
Page 472
Table 16.3
U.S. national saving: NIPA definition (decade average percent of GNP)
Period Percent of GNP
1950-1959 7.4
1960-1969 7.9
1970-1979 7.1
1980-1988 3.0
Sources: U.S. Department of Commerce, National Income and Product Accounts of the United States, 1929-82;
Survey of Current Business (July 1987-1989).
16.2, at $13.8 trillion, differs from the consolidated national net asset total of $14.4 trillion shown in Table
16.1.
I noted earlier that the Federal Reserve Board's national balance sheets do not include the assets of
governments. We should remember that the figures also exclude a form of wealth that swamps both private
and public tangible capital in importance: human capital. 5
U.S. Saving in International Context: National Accounts Measures
I turn now to some data on the behavior of national saving over time in the United States and in selected
foreign countries as reflected in national accounts. Table 16.3 presents decade averages of the ratio of
national saving (net national product less government and personal consumption) to GNP for the 1950-1988
384
385
period, taken from the NIPAs. The table makes clear that there was a sharp drop in the NIPA net national
saving rate in the 1980s. Figure 16.1 displays the annual data from the NIPAs, together with its downward
trend line.
National accounting within the Organisation for Economic Cooperation and Development (OECD) follows
rules laid down in the system of national accounts (SNA) developed by the United Nations, which differs in
various ways from the conventions of the U.S. NIPAs.6 The treatment of government expenditure is probably
the most important of these differences. In the case of the NIPAs, all government expenditure on goods and
services is regarded as consumption; the SNA, by contrast, divides government expenditure into consumption
and investment components and keeps track of the associated capital stocks, depreciation, and so on. In fact,
although the U.S. federal government does not distinguish current and capital
Â
Page 473
Figure 16.1
U.S.Â NIPAÂ nationalÂ savingÂ (percentÂ ofÂ GNP).Â Source:Â U.S.Â DepartmentÂ of
Commerce,Â BureauÂ ofÂ EconomicÂ Analysis,Â NationalÂ IncomeÂ andÂ Product
Accounts
expenditures in its budget, the U.S. Department of Commerce nonetheless compiles and publishes the figures.
These are incorporated in the multination compilations of the OECD.
A second difference between U.S. accounting practice and that of all the other OECD countries is in the
denominator typically chosen for measuring economic magnitudes. The U.S. practice is to express aggregates
such as consumption as fractions of GNP, or perhaps NNP. In the other OECD countries, the typical
denominator is gross or net domestic product (GDP or NDP). For purposes of exploring long-term saving
trends, the difference is not important. Table 16.4 suggests the difference implied by the differing treatments
of government expenditure in the two accounting systems. The first column shows the U.S. national saving
rate (as a ratio to net domestic product) according to the SNA definitions; the other column subtracts
government net investment from the SNA definition. Because net government capital formation has been
positive over the time period covered, the SNA saving rates are larger than the "NIPA"
Â
385
386
Page 474
Table 16.4
U.S. national saving: two definitions (decade average percent of NDP)
System of NIPA-type
national accounts measure
1960-1969 10.9 9.2
1970-1979 9.3 8.6
1980-1987 4.3 4.0
Source: Organisation for Economic Co-operation and Development, National Accounts.
saving rates. We can infer from the table that the rate of net capital formation by the U.S. government
dropped from about 1.7 percent of NDP in the 1960s to about 0.3 percent of NDP in the 1980s. (Barry
Bosworth has suggested to me that major influences on the decline have been the completion of the interstate
highway system and the slowdown in construction of new schools attributable to the shifting U.S.
demographic structure.) Figure 16.2 displays the annual time series with trend lines. Again, a sharp decline of
the saving rate is evident in both measures.
A common system of national

accounts preparation permits
international comparisons free
of one possibly significant
source of differences. Table
16.5 compares national saving
rates for six countries: the
United States, the United
Kingdom, Japan, Sweden, West
Germany, and Canada. Large
differences are noticeable,
particularly the high saving
propensity of Japan. National
saving rates in most countries
have declined in recent years.
Figure 16.3 displays annual
national savings rates, based on
the same OECD data, for four of
the countries for which I have
also assembled data on wealth at
market value.
One should not be too quick to draw conclusions from these international differences. Derek Blades and Peter
Sturm have warned that depreciation conventions have not been unified under the SNA, 7 and Barry
Bosworth has expressed to me the view that the Japanese national accounts understate depreciation, which
would tend to increase the numerator relative to the denominator of the Japanese national saving rate. Fumio
Hayashi has suggested that the relatively large place of land value in Japanese wealth also gives a somewhat
misleading quality to a simple comparison of national account saving rates.8
Â
386
387
Page 475
Figure 16.2
U.S.Â nationalÂ saving:Â comparisonÂ ofÂ measuresÂ withÂ trends.Â Source:
Â OrganisationÂ forÂ EconomicÂ Co-operationÂ andÂ Development,Â National
Accounts
Table 16.5
Comparison of national accounts saving rates (national saving as a percent of NDP)
United States United Kingdom Japan Sweden West Germany Canada
1963-1967 11.5 11.1 23.5 17.1 19.0 12.3
1968-1972 9.9 12.2 28.9 15.8 19.0 12.0
1973-1977 9.0 7.2 24.8 13.0 13.9 13.7
1978-1982 7.4 7.3 21.2 5.9 11.0 11.7
1983-1987 3.4 6.4 20.3 6.7 11.5 8.4
Source: Organisation for Economic Co-operation and Development, National Accounts.
Â
Page 476
387
388
FigureÂ 16.3
InternationalÂ comparisonÂ ofÂ nationalÂ accountsÂ saving.Â Source:Â Organisation
forÂ EconomicÂ Co-operationÂ andÂ Development,Â NationalÂ Accounts.Â ForÂ details,
seeÂ AppendixÂ 5.A
Personal Saving and the Flow of Funds
The discussion thus far has concentrated on national saving, for which all accumulation by U.S. residents,
whether directly on their own account or indirectly via the retained earnings of corporations, is aggregated.
The NIPAs also distinguish the income and expenditure of various sectors of the economic system, in
particular, of the household or personal sector (which includes nonprofit institutions as well as households in
the ordinary sense of the term). 9 The underlying idea of personal income is the payment to households for
supplying productive services (plus net transfers), but the idea's implementation emphasizes cash flow to
households as reflected in such transactions as receipt of wages and dividends. In the context of defining
income for purposes of taxation, economists customarily argue that accruing gain is equivalent to, say,
interest payments as a reward to the capitalist who supplies capital services. But accruing gains and losses on
assets held by households are not included in
Â
Page 477
NIPA personal income. Accrual accounting (in the usual financial accounting sense) is applied to directly
owned businesses (and thus income of proprietors is net of depreciation on capital used in the business, for
example).
As mentioned, economics casts doubt on the validity of cash-flow income as a good measure of the
compensation for services rendered by or of the economic opportunities of a household. Most obviously, a
family that owns a readily marketable asset that increases in value during the year is in much the same
position as one that owns a bank account to which the same amount of interest is credited during the year.
(That is the argument for including real capital gains in income subject to tax.) Similarly, a worker whose
employer deposits a certain amount in a defined-contribution pension plan is arguably in the same position as
one who receives the payment in cash and makes a deposit in an individual retirement account. Whether
personal income is a useful concept, either behaviorally or normatively, is thus very questionable.
Be that as it may, the NIPAs provide a measure of a statistically well-defined quantity: personal income. The
NIPA personal income measure includes most transfer payments from the government (Social Security
388
389
benefits and the like). The worker's share of payroll tax "contributions" for social insurance is deducted from
amounts reported as received as wage and salary disbursements by employers. (The employer's share is not
regarded as received by the employee in the first place.)
Of course, employee contributions to social insurance are not the only compulsory outlays by households. So
the NIPAs define "disposable personal income" as what is left from personal income after income taxes
(along with other similar taxes) have been paid. (The line item in the accounts "personal tax and nontax
payments" always amuses students.) Personal saving is disposable personal income less "personal outlays''
(personal consumption expenditures, constituting the vast bulk, plus consumer interest plus personal transfer
payments to foreigners).
The Flow of Funds Section of the Board of Governors of the Federal Reserve System also provides a measure
of personal saving that is alternative to that presented in the NIPAs. 10 This measure differs conceptually
from NIPA personal saving in two respects. First, the flow of funds account (FFA) saving measure treats net
acquisition of consumer durables as a form of saving, whereas the NIPAs regard
Â
Page 478
outlays on consumer durables simply as a component of personal consumption expenditures. Second, the
FFAs adopt a slightly different definition of personal income and count in certain credits from government
insurance programs and capital gains distributions from mutual funds, items excluded from the NIPA
definition of personal income.
There is, however, a further difference between the NIPA and FFA personal saving measures that is in some
ways more important. This difference is the nature of the data. In the NIPAs, personal saving is a residual
between a measure of personal income and a measure of consumption. 11 By contrast, FFA personal saving is
a direct measure of the net acquisitions of assets by households. It should be emphasized, however, that apart
from the two specific differences already mentioned, the NIPAs and FFAs purport to measure the same thing.
That is, both use the same basic concepts of income, consumption, and saving.
The differences in statistical approaches sometimes have resulted in fairly large differences in saving rates.
Table 16.6 presents data on NIPA personal saving side by side with FFA personal saving, both as defined in
the FFAs and as the NIPA personal saving rate would be if derived from the FFAs. Columns 1-3 of the table
show how FFA personal saving is built up from data on net capital expenditures (from the NIPAs, including
purchases of consumer durables less the depreciation on the stock of consumer durables) and acquisitions of
financial assets less increases in financial liabilities. The net of these three forms of saving is FFA personal
saving (column 3). From 1979 to 1988 the FFA personal saving rate thus defined was 8.3 percent, the same
level as in the 1960s. This rate was considerably less than the levels of 9.1 percent during the 1970s and 8.7
percent during the 1950s.
The FFAs provide data from

which we can construct an
estimate of personal saving
according to the NIPA
definitions of income. The
required adjustments are also
shown in Table 16.6 (columns 4
and 5). The NIPA personal
saving rates are substantially less
than the FFA personal saving
rates. The main reasons for the
389
390
difference are the inclusion of

acquisitions of consumer
durables and extra receipts
(credits from government
insurance and mutual fund
capital gains distributions) in
income and saving according to
the FFAs. Of these, the
investment in consumer durables
is the larger piece (although both
adjustments are significant).
Like the FFA personal saving
rate,
Â
Page 479
Table 16.6
U.S. personal saving (FFA and NIPA)
(6)
(1) (3) (4) (7) (8)
FFA net (5) NIPA House
Net capital (2) Acquisition of personal Income FFA personal personal Household discrep
expenditure financial assets (net) saving adjustmentsa savingNIPA basis saving discrepancy saving
1949-1958 6.5 2.2 8.7 3.3 5.4 4.5 0.9 20.7
1959-1968 4.8 3.6 8.3 2.8 5.5 4.6 0.9 20.7
1969-1978 4.7 4.4 9.1 3.3 5.8 5.4 0.4 8.0
1979-1988 4.6 3.7 8.3 3.8 4.5 3.8 0.7 20.2
Sources: Board of Governors, Federal Reserve System, and U.S. Department of Commerce.
Note: FAA denotes flow of funds accounts; NIPA denotes national income and product accounts. Figures are the average percentage of GNP for each dec
a. Includes consumer durables.
Â
Page 480
390
391
Figure 16.4
U.S.Â householdÂ discrepancyÂ (ratioÂ toÂ NIPAÂ saving)
the NIPA personal saving rate estimated from the FFAs was down in the 1980s from its high level in the
1970s.
The FFA estimate of the NIPA saving rate differs from the actual NIPA saving rate because of the different
statistical sources used. The NIPA figures are also shown in Table 16.6 (column 6). The difference between
the two (column 7), known as the "household discrepancy," has always been volatile but generally positive
(that is, direct estimates of household net acquisitions of assets exceed the amount implied in the national
income data). As Figure 16.4 shows, the household discrepancy had appeared to be declining as a fraction of
the saving being measured until the 1980s, when its volatility and apparent magnitude increased. As
discussed in detail by John Wilson and colleagues, there is no single explanation for the discrepancy, which
must be regarded as one measure of our uncertainty about household saving behavior. 12
The ratio of personal saving to disposable personal income, often referred to as the "personal saving rate," is
used as a measure of the saving choices of households. The line labeled "NIPA" in Figure 16.5
Â
Page 481
391
392
FigureÂ 16.5
U.S.Â personalÂ saving:Â threeÂ measuresÂ (percentÂ ofÂ disposableÂ personalÂ income).
Sources:Â U.S.Â DepartmentÂ ofÂ Commerce,Â BureauÂ ofÂ EconomicÂ Analysis,Â National
IncomeÂ andÂ ProductÂ Accounts,Â andÂ BoardÂ ofÂ Governors,Â FederalÂ ReserveÂ System,
FlowÂ ofÂ FundsÂ Section,Â FlowÂ ofÂ FundsÂ Accounts.Â Notes:Â FFAÂ denotesÂ flowÂ of
fundsÂ accounts;Â NIPAÂ denotesÂ nationalÂ incomeÂ andÂ productÂ accounts
shows the annual personal saving rate for the United States from 1949 to 1988. The well-known downward
trend in the NIPA personal saving rate since the mid-1970s is dramatically clear. The other two lines in
Figure 16.5 show the estimate of the NIPA concept of the personal saving rate from the FFAs, together with
the FFA personal saving rate expressed as a percentage of disposable personal income. Figure 16.5 suggests
what Table 16.6 may also have led us to expect: that for purposes of assessing the broad trend, it matters little
whether we use the personal saving rate from the NIPAs or the estimate of the same quantity from the FFAs.
It does apparently matter somewhat more whether we look at the NIPA concept or the FFA concept as there
is no down trend in the latter (the top line in Figure 16.5).
Â
Page 482
My own inclination is to prefer (slightly) the FFA personal saving concept, including its extra income
elements and inclusion of consumer durables in saving. But both measures are lacking insofar as they depend
on a somewhat arbitrary definition of what constitutes disposable personal income. As I have suggested, we
may reasonably doubt that disposable income is a good measure of the opportunities of households,
dependent as it is on the degree to which households see through various transactions to underlying economic
reality (such as accruing gains and losses or the change in pension claims). A better measure of the resources
out of which households in the aggregate may be thought to draw their consumption is the amount of national
output that is left over after the government's exhaustive uses.
Laurence Kotlikoff has provided the figures in Table 16.7 to measure the behavior of households in relation
to their real economic opportunities. Table 16.7 shows the behavior over time of two
392
393
Table 16.7
U.S. net national saving rates (corrected measures)
Period (1) (2) (3)
(Y-G-C)/Y (Y-G-C)/(Y-G) G/Y
1940-1949 0.086 0.101 0.271
1950-1959 0.133 0.167 0.203
1960-1969 0.130 0.166 0.215
1970-1979 0.118 0.152 0.223
1980-1985 0.072 0.093 0.230
Source: Laurence J. Kotlikoff, testimony before the U.S. House of Representatives, Committee on Ways and
Means, April 20, 1989.
Notes: Y is national income and product account (NIPA) net, national product plus imputed rent on consumer
durables and government tangible assets (excluding military equipment) less depreciation on the stock of
consumer durables and government tangible assets (excluding military equipment).
G is NIPA government expenditures less government expenditures on tangible assets (excluding military
equipment) plus imputed rent on the government's stock of tangible assets (excluding military equipment).
C is NIPA personal consumption less consumer expenditures on durables plus imputed rent on the stock of
consumer durables.
Imputed rent on an asset is calculated as annual depreciation plus 3 percent times the stock of the asset. Annual
depreciation of consumer durables and government tangible assets as well as the stocks of consumer durables
and government tangible assets are reported in U.S. Department of Commerce, Fixed Reproducible Tangible
Wealth in the United States, 1925-1985 (Washington, D.C.: U.S. Government Printing Office, June 1987).
Â
Page 483
measures of national saving. The first is a version of the usual national saving rate constructed by the addition
to national output estimates of the flows of services from government tangible capital (excluding military
capital) and from consumer durables; similar modifications are made to the NIPA figures for personal and
government consumption. The national saving rate in column 1 of Table 16.7 thus differs from the usual
national saving rate in that it uses these modified income and consumption concepts and has net rather than
gross product in the denominator. This national saving rate is larger than the usual NIPA measure, but it also
displays a sharp downturn in the 1980s. Column 2 presents a modified "personal" saving rate, one that in
effect assumes no substitutability between government and personal consumption and therefore conceives of
households as making their choices from what remains from national output after government consumption
has been subtracted. This modified personal consumption rate shows an even more marked decline in the
1980s, suggesting that households are "to blame" for the decline in savingnot the government, whose claims
have only modestly increased.
National Accounts and Market Value Saving
The national accounts of the United States and other countries generally seek to produce a measure of
"current production" of goods and services. Because the measure of production is value, not physical units,
there are many definitional problems, particularly those associated with what may be generically described as
accruing changes in value. For example, the value of an inventory of wine held from the beginning of the
year to the end typically increases. This increase in value, if it occurs in the context of a wine-producing
business, is presumably economically equivalent to the increase in value of the grapes that are grown in the
field, harvested, and sold to the wine producer. This increase is probably not, however, captured in either
393
394
business or national income accounts. Much the same can be said of most other accruing changes in value of
assets (both up and down).
Accruing changes in value in fact are significant elements of the performance of the U.S. economy. Figure
16.6 displays, for example, the ratio of the market value of the net assets of U.S. nonfinancial corporations to
their replacement value for the period 1945-1988.
Â
Page 484
FigureÂ 16.6
MarketÂ valueÂ ofÂ U.S.Â corporateÂ equityÂ relativeÂ toÂ corporateÂ netÂ worth
(percent).Â Source:Â U.S.Â FederalÂ Reserve
Two aspects are notable. First, the market value of corporate equity is typically considerably less than the
replacement cost of the underlying physical assets, including land. Second, the ratio of the market value to
the accounting value is far from constant, having varied over a range between 37 and 110 percent during the
forty-three-year period.
In the economic theory of household consumption, it is the market value of wealth, not its accounting value,
that plays an economic role. 13 Saving by a household consists of the accumulation of wealth as valued in
asset markets and is measured by the change in the value of wealth between the beginning and end of the
accounting period. The Flow of Funds Division's national balance sheets provide data from which a measure
of the aggregate of household wealth can be constructed. Table 16.8 presents decade averages of the ratio of
national saving to GNP when national saving is defined as the annual change of the aggregate wealth of U.S.
households (including the claims they hold indirectly through pension funds and the like). The first column of
the table shows saving rates when wealth is de-
Â
394
395
Page 485
Table 16.8
U.S. national saving: comparison of measures (percentage of GNP decade averages)
Market value National income and product accounts
Government assets
Excluded Included
1950-1959 14.8 16.6 7.4
1960-1969 10.0 12.0 7.9
1970-1979 6.9 9.2 7.1
1980-1988 7.0 7.1 3.0
Note: NIPA saving is net national saving (see table 16.3); market value saving is the year-to-year change in
national wealth at market value based on the Federal Reserve national balance sheets (see table 16.2), deflated
to 1982 price levels using the GNP deflator and augmented by Department of Commerce estimates of the
replacement value of government capital.
fined as in Table 16.2, exclusive of government assets and with government debt netted out of household
wealth. The second column shows the rates when national wealth is extended to incorporate the net worth of
governments, using the Department of Commerce estimates of government capital stocks. These rates usually
differ from the national saving rates according to the NIPAs (shown in the third column of the table). As will
the NIPA saving rates, a downward trend is suggested by the decade averages.
Figure 16.7 displays the annual data on national market value saving, together with NIPA saving, where
national market value saving is defined to include the value of government assets. Because the value of assets
is highly volatile (the main driving element in the data is the stock market), the market value saving rate
generates a jagged picture. Its sawtooth record contrasts with the smooth path of the NIPA saving rate, but
like the NIPA record, the market value saving rate also drifts downward. The trend in the market value saving
rate is far from sharp, however; the regression on time is negative, but statistically insignificant, even when
1946 is excluded.
Students of taxation generally accept what has come to be known as the Schanz-Haig-Simons (SHS)
definition of income for purposes of income taxation. 14 Rather than treat saving as a residual, the SHS
definition defines income as the sum of the taxpayer's consumption and change in wealth during the year,
where wealth is measured at market value. Having a time series of national wealth at market value puts us in
a position to produce a measure of aggregate SHS
Â
Page 486
395
396
FigureÂ 16.7
U.S.Â marketÂ valueÂ andÂ NIPAÂ savingÂ (percentÂ ofÂ GNP).Â Sources:Â U.S.Â Federal
Reserve;Â U.S.Â DepartmentÂ ofÂ Commerce
income: the sum of the year-to-year change in wealth and the aggregate consumption, where wealth and
consumption are defined consistently. Consistent definition of consumption requires an imputation for the
flow of consumption services for forms of wealth not used for production for market. To construct a series
for aggregate SHS income using private wealth (that is, excluding government assets), we have imputed
consumption out of consumer durables in the amount of the sum of the depreciation of the stock of durables
during the year and the assumed yield on the stock. The latter is calculated as 5.2 percent of the average of
the beginning and end-of-year stocks. (The 5.2 percent real rate of return is the average real rate of return on
owner-occupied housing implicit in the Department of Commerce's imputation of the rental value in the
NIPAs.) Depreciation is the difference between the "gross investment" (outlays on consumer durables) and
the year-to-year increase in real stock. Figure 16.8 displays the graph of the resulting national SHS income
se-
Â
Page 487
396
397
FigureÂ 16.8
U.S.Â nationalÂ income:Â comparisonÂ ofÂ twoÂ measures.Â Sources:Â U.S.
FederalÂ Reserve;Â U.S.Â DepartmentÂ ofÂ Commerce
ries, together with the net national product from the NIPAs. Because the two measures have so much in
common, it should not be surprising that they follow roughly similar paths, but the SHS measure is
considerably more volatile.
Saving, Capital Formation, and Economic Performance
The object of saving is the accumulation of wealth, which in turn supports consumption. Conversely, high
rates of consumption may imply ''too low" rates of saving. In this section I present data on the effects of U.S.
saving behavior as reflected in the accumulation of wealth and measures of consumption. Figure 16.9 makes
clear what is not readily inferred from the study of saving rates: that over a forty-three-year period U.S.
residents achieved a substantial increase in wealth per capita. The data graphed include government tangible
Â
Page 488
397
398
FigureÂ 16.9
U.S.Â realÂ wealthÂ perÂ capita.Â Sources:Â U.S.Â FederalÂ Reserve;Â UÂ S.Â Department
ofÂ Commerce;Â EconomicÂ ReportÂ ofÂ theÂ President,Â 1989,Â Note:Â Includes
governmentÂ tangibleÂ assets
assets in the measure of wealth, which in the case of tangible assets is converted from current-value to
constant-dollar units by deflating at an estimate of the end-of-year GNP deflator. The path labeled "book
value" describes the history of the accounting value of per capita capital, whereas the path labeled "market
value" uses the market value for corporate equity. The "accounting value" here is the "total consolidated
national net assets" figure of Table 16.1, deflated, augmented by the Department of Commerce estimate of
the value of the government tangible capital stock. Land is included at market value in both wealth concepts.
Figure 16.9 reveals that in the aggregate (as well as in the case of equity claims to U.S. nonfinancial
corporations), "book'' value (the quotation marks are to remind us that the data are adjusted for inflation) has
often been above market, substantially so between 1975 and 1985. The data also show a significant contrast
between a slight down trend in "book" wealth per capita since the late 1970s and something more like a
continuation of the historical uptrend in market value wealth.
Â
Page 489
398
399
FigureÂ 16.10
U.S.Â consumptionÂ perÂ capita.Â Source:Â U.S.Â DepartmentÂ ofÂ Commerce,Â Bureau
ofÂ EconomicÂ Analysis,Â NationalÂ IncomeÂ andÂ ProductÂ Accounts
The object of production is consumption, but too much consumption implies too little accumulation. Figure
16.10 shows what happened to per capita consumption, both private and government, over the period
1946-1988. The data in the figure follow the NIPA convention of treating all government purchases of goods
and services as consumption. But purchases of consumer durables are treated two alternative ways: either as
simple consumption (again, as in the NIPAs) or as an imputed rent on a stock (by use of the imputation
method described previously). It is rather striking how little difference it makes which approach is taken to
the measurement of aggregate consumption. The pronounced upward trend in private and in government
consumption per capita is both evidence that the economy is delivering the goods to U.S. residents and a
signal that their saving may be in some sense deficient.
Â
Page 490
399
400
FigureÂ 16.11
U.S.Â privateÂ consumptionÂ (ratioÂ toÂ nationalÂ wealth).Â Sources:Â U.S.Â Federal
Reserve;Â U.S.Â DepartmentÂ ofÂ Commerce
The theory of consumption behavior models individuals as seeking to smooth their consumption, typically
over their lifetimes or at least over an extended future run of years. This model does not readily generate a
particular relationship between consumption and income, but with certain simplifying assumptions it does
suggest a constant ratio between consumption and wealth, including human wealth. One place to look for a
shift in the consumption behavior of U.S. residents is in the ratio of consumption to the piece of wealth we
can observe. Figure 16.11 plots the ratio of aggregate private consumption (including the imputed rent on
consumer durables) to three different versions of national wealth. The first and smallest measure of wealth is
national wealth excluding both government assets and debt. Use of this measure yields the highest ratio of
consumption to national wealth, as shown in Figure 16.11. This measure is the wealth that would govern the
behavior of households that see through the government debt in their aggregate portfolio to the future
Â
Page 491
taxes they must pay to service and retire that debt but do not see through the veil of government to the assets
their taxes and lending to government may be financing. (Alternatively, they see through to the assets and do
not believe they will have a positive yield.)
The next largest measure of wealth since the early 1950s has been aggregate wealth including the value of
government debt but excluding government assets. This is the wealth most people seem to regard as operative
in household calculations. In other words, government issue of bonds has the effect of increasing the
perceived stock of wealth to be held by someone. It is the wealth of a world of fiscal illusion (in which,
perhaps, we live). Third, and largest, is the stock of wealth including government assets. If government assets
are really productive (at least of reduced future taxes), this is the wealth that should operate in the world of
super-rational households that see through the government veil.
If we work from an assumption of no trend in the relationship between human and nonhuman wealth,
inspection of Figure 16.11 suggests that there has been no very sharp break in recent years with behavior in
the past, as reflected in the chosen ratio of consumption to wealth. I can see in the differences in the curves,
400
401
however, some support for the fiscal illusion view that consumption is regulated to maintain a constant ratio
to the stock of wealth including government. A growing stock of debt will cause the level of consumption to
rise relative to one or the other "rational" perceptions of wealth (to include government). In recent years
certainly, the ratio of consumption to wealth including government debt has risen relative to its ratio to either
of the other two wealth measures. Figure 16.12 offers a magnified view of the portion of Figure 16.11 that
covers the period 1980-1988.
I noted previously that aggregate wealth at book value has been declining or constant for the past few years.
Figure 16.13 presents another facet of the same fact, showing the tangible capital stock (including
inventories, housing, consumer durables, and government reproducible assets) per U.S. worker (represented
by the civilian labor force) annually from 1948 to 1987. The divergence from trend of this very broadly
conceived capital-labor ratio, while not conclusive evidence of suboptimal performance, is quite noticeable.
The failure of book value to track market value is important to this record, and one should not be too quick to
accept the view that it is "real" capital, as measured in the current-value accounts kept by the
Â
Page 492
FigureÂ 16.12
U.S.Â privateÂ consumption,Â 1990-1988Â (ratioÂ toÂ nationalÂ wealth).Â Sources:Â U.S.
DepartmentÂ ofÂ Commerce,Â BureauÂ ofÂ EconomicÂ Analysis,Â NationalÂ IncomeÂ and
ProductÂ Accounts,Â andÂ BoardÂ ofÂ Governors,Â FederalÂ ReserveÂ System,Â FlowÂ of
FundsÂ Section,Â FlowÂ ofÂ FundsÂ Accounts
Department of Commerce, that contributes to the productivity of workers. Certainly, the entrepreneur expects
to get the same extra productivity from the last $1 million spent on any form of capital, be it in the form of
machines, patents, or land.
U.S. Saving in
International Context:
Market Value Wealth
and Saving
401
402
The previous section considered the recent saving behavior of the United States in comparison with its own
past, with reference particularly to the accumulation of wealth measured at market value and to consumption
in relation to the value of wealth. In this section U.S. behavior is compared with that of other countries.
Although comparable national accounting data are available for a large group
Â
Page 493
FigureÂ 16.13
U.S.Â capitalÂ stockÂ (ratioÂ toÂ laborÂ force).Â Sources:Â U.S.Â FederalÂ Reserve;
U.S.Â DepartmentÂ ofÂ Commerce;Â EconomicÂ ReportÂ ofÂ theÂ President,Â 1989
of countries, estimates of national wealth are available for few. We have been able to assemble such data for
three countries in addition to the United States: the United Kingdom, Japan, and Sweden.
There are three figures for each of the four countries, in each case for as long a time series as we have been
able to construct. For each country, the first figure presents wealth per capita, where wealth is defined as the
market value of privately owned assets excluding government debt. The second displays national saving rates
(national saving to GNP or GDP), for saving defined as the change in market value wealth and on a national
accounts basis. The third figure in each case shows the ratio of market value wealth to GNP or GDP, as
suggestive both of possible differences in economic structure across the four countries and of possible shifts
in the character of production within each.
The data reveal a good deal of diversity. In the case of the United States, the good news is that the real wealth
per capita is about on its forty-three-year upward trend line (Figure 16.14). U.S. residents managed to
accumulate about $800 (1988) per person per year over
Â
402
403
Page 494
FigureÂ 16.14
RealÂ wealthÂ perÂ capita:Â UnitedÂ States.Â Sources:Â U.S.Â FederalÂ Reserve;Â U.S.
DepartmentÂ ofÂ Commerce;Â EconomicÂ ReportÂ ofÂ theÂ President,Â 1989.Â Note:
ExcludesÂ governmentÂ tangibleÂ assets
that period. The saving behavior described in Figure 16.15 is a repeat of an earlier display. The wealth-output
ratio in the United States has wavered around 2.8 over the whole period and shows no obvious trend (Figure
16.16).
The British story, for which we have a much shorter record, reveals a drop in wealth per capita in the early
1970s, just as in the United States, but a rather longer and stronger upward trend since then (Figure 16.17).
(A more careful international comparison will follow.) The path of the market value saving rate has even
wider swings than that of the U.S. rate along a similarly downward-drifting national accounts measure
(Figure 16.18). At a fairly steady 3, the wealth-output ratio is close to that of the United States (Figure 16.19).
In Japan, wealth per capita has

increased steadily and rapidly
(Figure 16.20). The comparative
behavior of market value and
national accounts measures of
national saving rates is
qualitatively
Â
Page 495
403
404
FigureÂ 16.15
U.S.Â savingÂ rate:Â comparisonÂ ofÂ measures.Â Sources:Â U.S.Â FederalÂ Reserve;Â U.S.Â DepartmentÂ ofÂ Commerce;Â
EconomicÂ ReportÂ ofÂ theÂ President,Â 1989
Â
Page 496
FigureÂ 16.16
Wealth-GNPÂ ratio:Â UnitedÂ States.Â Sources:Â U.S.Â FederalÂ Reserve;Â U.S.Â DepartmentÂ ofÂ Commerce;Â
EconomicÂ ReportÂ ofÂ theÂ President,Â 1989
404
405

Page 497
FigureÂ 16.17
RealÂ wealthÂ perÂ capita:Â UnitedÂ Kingdom.Â
Sources:Â SeeÂ AppendixÂ A
similar to that in the United States and the United Kingdom (Figure 16.21). The different feature in the
Japanese record is the strong upward drift in the wealth-output ratio, from about 3 at the end of the 1960s to
more than 5 in 1987 (Figure 16.22)possibly a sign that assets are "overvalued."
Sweden presents the most surprising picture. After twenty years of steady increase, from the mid-1950s to the
mid-1970s, per capita wealth turned sharply down, falling by 1985 to about its 1965 level (Figure 16.23).
Naturally, the saving rates were consistent with this path, with large negative saving in the late 1970s to the
mid-1980s (Figure 16.24). For Sweden, even the national accounts saving rate was negative in 1986 (when
the market value saving rate was significantly positive). As there has been no sharp contraction in Swedish
national output, it follows that the wealth-output ratio also fell sharply in the 1977-1984 years (Figure 16.25).
As in the case of Japan, one wonders if the market is "improperly" valuing Swedish assets, in this case
undervaluing them. It is also of interest that even in the period when it was fairly steady, the wealth-output
ratio in Sweden was, at about 1.8, substantially less than that observed in the other
Â
Page 498
405
406
FigureÂ 16.18
BritishÂ savingÂ rate:Â comparisonÂ ofÂ measures.Â
Â
Page 499
FigureÂ 16.19
Wealth-GNPÂ ratio:Â UnitedÂ Kingdom.Â
Â
406
407
Page 500
FigureÂ 16.20
RealÂ wealthÂ perÂ capita:Â Japan.Â
countriesperhaps a signal of undiscovered differences in coverage of the data.
It is of interest to compare the four countries' wealth accumulation. The comparison is facilitated here by the
fact that the wealth-per-capita amounts were presented in constant real units of the respective currencies.
Figures 16.26 and 16.27 make the attempt. Figure 16.26 plots on a single graph the relative paths of per
capita wealth, indexed to 1970, the earliest year in the data for Japan. The remarkable rate of accumulation in
Japan, compared with the three other countries, jumps out of the figure, as do the strong recent growth in
wealth in Britain and the slump in wealth per capita in Sweden.
Figure 16.27 uses GDP purchasing power parity exchange rates compiled by the OECD to convert all four
wealth-per-capita series to 1982 U.S. dollars of purchasing power. Naturally, we have to take with a grain of
salt the idea of selling a house in Japan to acquire U.S. consumer goods. Nevertheless, the rapid growth in
Japanese per capita wealth (from almost the bottom to the top of the group)
Â
Page 501
407
408
FigureÂ 16.21
JapaneseÂ savingÂ rate:Â comparisonÂ ofÂ measures.Â
Â
Page 502
FigureÂ 16.22
Wealth-GNPÂ ratio:Â Japan.Â
Â
408
409
Page 503
FigureÂ 16.23
RealÂ wealthÂ perÂ capita:Â Sweden.Â
is striking, as are the recent tendency toward convergence of the British and U.S. levels and the low level of
Swedish wealth. Given the evident high living standard in Sweden, we must again wonder whether there is an
undiscovered difference in coverage of the data.
In view of the exceptional character of Japan's economic performance and the high degree of current U.S.
interest in Japan, two additional figures comparing the two countries are provided. Figure 16.28 displays
annual data on the fraction of national private wealth consisting of land value. It is widely known that real
estate prices in Japan are very high, but it may not be generally appreciated that land value accounted for
nearly 90 percent of the country's wealth as early as 1970 and did not account for less than 70 percent during
the period 1970-1987. The comparable fraction for the United States drifted upward from about 20 percent in
1970 to about 25 percent in 1987.
It might be argued that high wealth owing to high land prices should be interpreted as bad, rather than good, a
signal of crowding and agricultural protection. Figure 16.29 presents wealth per capita
Â
Page 504
409
410
FigureÂ 16.24
SwedishÂ savingÂ rate:Â comparisonÂ ofÂ measures.Â
Â
Page 505
FigureÂ 16.25
Wealth-GDPÂ ratio:Â Sweden.Â
Â
410
411
Page 506
FigureÂ 16.26
InternationalÂ comparisonÂ ofÂ realÂ wealthÂ perÂ capitaÂ index.
Â Sources:Â FiguresÂ 16.14,Â 16.17,Â 16.20,Â andÂ 16.23
in the United States and in Japan, converted to 1982 U.S. dollars of purchasing power using the OECD
purchasing power parity exchange rates. In wealth-less-land per capita, Japan has substantially less than the
United States.
Concluding Remarks
In devoting so much attention to matters of measurement, I have perforce had to slight the development of
policy assessment and policy analysis. I therefore offer some only loosely supported reflections.
The major question addressed in this book is why the United States is saving so little. I would rephrase that
question and ask, Why is the United States consuming so much? On my reading of the record of the behavior
of wealth in the sense that most people mean the term, the recent consumption (and therefore saving)
behavior of
Â
Page 507
411
412
FigureÂ 16.27
InternationalÂ comparisonÂ ofÂ realÂ wealthÂ perÂ capita.Â
Sources:Â ConversionÂ basedÂ onÂ GDPÂ purchasingÂ powerÂ partyÂ exchangeÂ ratesÂ
fromÂ OrganisationÂ forÂ EconomicÂ Co-operationÂ andÂ Development,Â AnnualÂ
NationalÂ Accounts,Â 1987
the U.S. public is not glaringly out of line with its past behavior. A reasonable standard of consistency with
past behavior would be for consumption to maintain its past ratio to wealth. Figure 16.30 presents a further
excerpt from Figure 16.11, in this case showing the trend in the consumption-wealth ratio from 1946-1979
and its projection through 1988, where wealth is interpreted to include the public's holding of government
debt and to exclude government assets. The figure shows that the consumption-wealth ratio in the 1980s was
very close to its historic trend value. (Obviously, a plot of one or two variables against time is not an
analytical model. I indicated previously that these reflections would be loosely supported. In any case, most
argumentation on this subject is at this level.)
Â
Page 508
412
413
FigureÂ 16.28
U.S.Â andÂ JapaneseÂ land-wealthÂ ratios.Â
Sources:Â U.S.Â dataÂ areÂ fromÂ theÂ U.S.
FederalÂ ReserveÂ andÂ U.S.Â DepartmentÂ ofÂ Commerce.Â
SeeÂ AppendixÂ AÂ forÂ JapaneseÂ dataÂ sources
From a policy perspective, of course, the regularity of consumption suggested in Figure 16.30 is not welcome
insofar as it reflects private prudence but not public prudence, which might rather imply a steady ratio
between consumption and wealth exclusive of government debt and perhaps inclusive of government assets.
It is thus possible to find U.S. consumption behavior explicable in terms of what people perceive as their
circumstances but too low by reference to their actual circumstances.
By ''actual circumstances" I mean the

market assessment of the aggregate
wealth of the country. There is a
further question whether we should
rely on the market assessment in
thinking about policy and whether
U.S. consumers have historically
"believed" the financial market results
in the way that the projection in Figure
16.30 implies they might. It is
notoriously hard to outguess the
market. But Patric Hendershott
suggests that the measured
consumption-wealth ratio has tended
to do that, being lower in stock market
booms than in busts. If we presume
we are presently in a boom,
consumption is unexpectedly high,
relative to wealth, in comparison with
past
Â
413
414
Page 509
FigureÂ 16.29
U.S.Â andÂ JapaneseÂ wealthÂ perÂ capita.Â
Sources:Â U.S.Â dataÂ areÂ fromÂ theÂ U.S.Â FederalÂ
ReserveÂ andÂ U.S.Â DepartmentÂ ofÂ Commerce.Â
SeeÂ AppendixÂ AÂ forÂ JapaneseÂ dataÂ sources
behavior, and the basis for policy concern is all the greater. The hypothesis seems to me well worth further
exploration.
We should recognize that even if the U.S. public has shifted its behavior toward a higher consumption, lower
accumulation path, it is not self-evident that that fact alone provides a basis for a policy to reverse that
choice. Nor is it immediately clear that it should be a matter of policy concern that Americans choose to
accumulate less rapidly than the Japanese do.
We might argue that most people are insufficiently reflective about their saving behavior. Most people
apparently never accumulate anything beyond a bare minimum of financial assets (although it is not clear that
they rationally should, given Social Security and other retirement resources). But some of the studies of
individual retirement accounts, for example, indicate that people are not much
Â
Page 510
414
415
FigureÂ 16.30
U.S.Â privateÂ consumption:Â comparisonÂ ofÂ actualÂ andÂ extrapolatedÂ trendÂ (ratio
toÂ nationalÂ wealth).
Â Sources:Â U.S.Â FederalÂ Reserve,Â U.S.Â DepartmentÂ ofÂ Commerce.
Note:Â IncludesÂ governmentÂ debt.Â GovernmentÂ assetsÂ notÂ included
motivated by a calculus of rate of return and are much motivated by advertising. Although philosophically
not very attractive, a dose of paternalism sometimes seems rather sensible.
I do think, however, that there are reasons quite consistent with respect for the rationality of our fellow
citizens to advocate significant changes in policy toward accumulation. By and large, existing policy
discriminates against accumulation. (I was interested to see the survey evidence gathered by John Immerwahr
that indicated that the U.S. public regards accumulation as socially bad. 15 It would be easy to compile a long
list of discriminatory policies, from the income tax to asset tests for receipt of health insurance benefits or
scholarships. In many such cases, the antiaccumulation feature may be a necessary screen to focus some sort
of help on the truly needy, but I wonder how carefully the matter has been considered. To be sure, there are
exceptions, such as the heavy government investment in schooling and the income-tax-free imputed return on
investment in consumer durables and owner-occupied housing. But even some of the apparent exceptions,
such as tax-sheltered retirement saving,
Â
Page 511
are often organized with features (for example, the contribution ceiling) that negate their effects at the critical
margin.
Steps to ameliorate some of the many penalties on accumulation might reasonably be justified on grounds of
a presumption in favor of neutrality toward saving. But we could argue that policies should not merely be
neutral but should promote saving because it has positive externalities. To the extent that each of us places a
positive value on the current standard of living of others, it is in our interest to encourage each other to save.
Appendix A
Notes on the International Data
General Considerations
415
416
Except where specifically indicated to the contrary, the figures reflect the following conventions, adopted in
part because data needed to improve upon them are lacking or unreliable:
1. National government tangible assets are ignored.
2. All government spending is classified as consumption.
3. Local government assets and liabilities are ignored; local governments are treated as having zero net worth.
4. Consumer durables are counted as part of national wealth.
United Kingdom
All publications referred to in this section are by the Central Statistical Office in London. Data are
reproduced by permission of the Controller of Her Majesty's Stationery Office.
GNP Deflator
The deflator used for an end-of-year stock is the average of the fourth and following first quarter figures.
1979-1987 Monthly Digest of Statistics (March 1989), p. 6.
1957-1978 Economic Trends Annual Supplement (1987), pp. 5-6.
Real GNP
Nominal GNP has been taken from the same sources as the GDP deflator.
Nominal GNP is converted to real terms using GDP deflators.
National Accounts Saving
United Kingdom National Accounts (1988), pp. 36-37.
Â
Page 512
Population
Monthly Digest of Statistics (October 1989), p. 19; and Annual Abstract of Statistics (1987), p. 6. Midyear
estimates for the current and following years are averaged to approximate the year-end figures.
Personal Sector Net Worth
1976-1987 United Kingdom National Accounts (1988), p. 87. The stock of consumer
durables, detailed in the same source, are added in.
1975 C. G. E. Bryant, "National and Sector Balance Sheets," Economic Trends (May
1987) (pp. 92-119), p. 102.
1966-1974 Central Statistical Office, "Personal Sector Balance Sheets," Economic Trends
(January 1978) (pp. 97-107), p. 102; C. W. Pettigrew, "National and Sector
Balance Sheets for the United Kingdom," Economic Trends (November 1980)
(pp. 82-100), p. 92. The former source is annual but does not include nonprofit
416
417
organizations, whereas the latter has balance sheets at three-year intervals. The
figures used in the chapter are from Pettigrew, with the gaps filled using the
Central Statistical Office report to indicate approximate year-to-year changes.
Public Net Worth
Public net worth is public corporation net worth plus national government financial assets minus national
government liabilities.
1976-1987 United Kingdom National Accounts (1988).
1975 C. G. E. Bryant, "National and Sector Balance Sheets," Economic Trends (May
1987), pp. 92-119.
1966-1974 C. W. Pettigrew, "National and Sector Balance Sheets for the United
Kingdom," Economic Trends (November 1980), pp. 82-100. Because these
balance sheets are at three-year intervals, interpolation has been used. The
public debt series in Financial Statistics (1978-1982), Table sll, have been
used to estimate year-to-year changes.
Japan
Private Net Worth
Kokumin Keizai Keisan Nempo (Annual Report on National Accounts; hereafter KKKN) (Tokyo: 1989),
Economic Planning Agency, pp. 346-353, line 6. Consumer durables have been added using KKKN (1989), p.
413; (1988), p. 445; (1986), p. 387; (1983), pp. 628-629. A change in definition of the consumer durable
measure between 1983 and 1986 forced a modification of some of the figures for consistency. The difference
is insignificant in comparison to total wealth.
Â
Page 513
Government Net Worth
KKKN (1989), pp. 364-365; (1988), pp. 384-385; (1983), pp. 522-526. The figure is the net financial assets of
the central government and social security fund.
GNP Deflator
KKKN (1989), pp. 130-133; (1988), pp. 504-505. For 1980-1987, the given figure is the average of the fourth
and following first quarter figures. Before 1980, the given figures are the average of the annual figures for the
current and following years.
Real GNP
Keizai Tokei Nempo (Economic Statistics Annual) (Tokyo: Bank of Japan, 1976-1988). The pages for 1988
are 337-338. Nominal GNP is deflated using the GNP Deflator.
National Accounts Saving
417
418
KKKN (1989), pp. 82-83; (1988), pp. 82-83; (1983), pp. 10-11.
Population
Japanese Statistical Yearbook (Tokyo: Statistics Bureau, Management and Coordination Agency, 1987), p.
24; Keizai Tokei Nempo (Economic Statistics Annual) (Tokyo: Bank of Japan, 1988), p. 300. Populations are
as of October 31.
Sweden
Private Net Worth; Price Index; National Accounts Savings
Estimates provided by Lennart Berg.
National Debt
Statistisk Arsbok for Sverige (Stockholm: Statistiska CentralbyrÃ¥n, 1958-1988). In 1988, Table 284, p. 249.
Population
Statistisk Arsbok for Sverige (Stockholm: Statistiska CentralbyrÃ¥n, 1963, 1981, 1988). In 1988, Table 241,
p. 231.
International Comparisons
Purchasing Power Parity
For purposes of international comparison, wealth per capita is converted to U.S. dollar units using the
purchasing power parity ratios published in
Â
Page 514
the National Accounts Main Aggregates Volume 1, 1960-1987 (Paris: OECD, Department of Economics and
Statistics, 1989), pp. 150-151. The U.S. dollar figures are converted to constant 1982 dollars using the GNP
deflator.
Notes
For helpful discussions and leads on data. I particularly thank Albert Ando, Barry Bosworth, Angus Deaton,
R. Glenn Donaldson, Elizabeth Fogler, Charles Horioka, Laurence J. Kotlikoff, Robert E. Lipsey, James
Poterba, Jan SÃ¶dersten, and Frederick O. Yohn. Michael Williams of Princeton University provided quick
and ingenious research assistance.
1. For a clear exposition of saving concepts in the U.S. national income and product accounts, see Thomas M.
Holloway, "Present NIPA Saving Measures: Their Characteristics and Limitations," in Robert E. Lipsey and
Helen Stone Tice, eds., The Measurement of Saving, Investment and Wealth (Chicago: University of Chicago
Press, 1989), pp. 21-100.
2. U.S. Department of
Commerce, Bureau of Economic
Analysis, National Income and
Product Accounts of the United
States, 1929-82 (Washington,
D.C.: U.S. Government Printing
418
419
Office, 1986).
3. For an extensive discussion of the issues, see David F. Bradford, "Market Value vs. Financial Accounting
Measures of National Saving," NBER Working Paper No. 2906 (Cambridge, Mass.: National Bureau of
Economic Research, March 1989). For a good introduction to financial accounting concepts, see George
Foster, Financial Statement Analysis, 2nd ed. (Englewood Cliffs, N.J.: Prentice-Hall, 1986).
4. U.S. Department of Commerce, Bureau of Economic Analysis, Fixed Reproducible Tangible Wealth in the
United States 1925-85 (Washington, D.C.: U.S. Government Printing Office, June 1987). See also John C.
Musgrave, "Fixed Reproducible Wealth in the United States, 1979-82," Survey of Current Business 63, no. 8
(August 1983), pp. 62-67; John C. Musgrave, "Fixed Reproducible Tangible Wealth in the United States,
1972-82, Revised Estimates, Survey of Current Business 66, no. 1 (January 1986), pp. 51-75; and John C.
Musgrave, "Fixed Reproducible Tangible Wealth in the United States, 1982-85," Survey of Current Business
66, no. 8 (August 1986), pp. 36-39.
5. For recent studies of the magnitude of these two forms of investment see Dale W. Jorgenson and Barbara
M. Fraumeni, "The Accumulation of Human and Nonhuman Capital, 1948-84," in Lipsey and Tice, eds., The
Measurement of Saving, Investment and Wealth, pp. 227-282; and Michael J. Boskin, Marc S. Robinson, and
Alan M. Huber, "Government Saving, Capital Formation and Wealth in the United States, 1947-85," in
Lipsey and Tice, eds., The Measurement of Saving, Investment and Wealth, pp. 287-353.
6. See Derek W. Blades and Peter Sturm, "The Concept and Measurement of Savings: The United States and
Other Industrialized Countries," in Federal Reserve Bank of Boston, Saving and Government Policy (Boston:
Federal Reserve Bank of Boston, 1982), pp. 1-30.
7. Ibid.
8. See Fumio Hayashi, "Is Japan's Saving Rate High?" Quarterly Review of the Federal Reserve Bank of
Minneapolis (Spring 1989), pp. 3-9.
Â
Page 515
9. The data for personal income and outlays are presented in the Table 2 series in Survey of Current Business.
10. For a clear discussion, see

John F. Wilson, James L.
Freund, Frederick O. Yohn, Jr.,
and Walther Lederer,
"Household Saving
Measurement Recent Experience
from the Flow of Funds
Perspective," in Lipsey and Tice,
eds., The Measurement of
Saving, Investment and Wealth,
pp. 101-152.
11. For a clear display of the NIPA concepts in this respect, see Table 2.1 in Survey of Current Business.
12. John F. Wilson et al., "Household Saving Measurement Recent Experience from the Flow of Funds
Perspective."
13. A number of authors have emphasized this point, developed at length in Bradford, "Market Value vs.
Financial Accounting Measures of National Saving."
419
420
14. David F. Bradford, Untangling the Income Tax (Cambridge, Mass.: Harvard University Press, 1986).
15. John Immerwahr, "Saving: Good or Bad?" (New York: Public Agenda Foundation, 1989).
Additional References
Federal Reserve System, Board of Governors. Balance Sheets for the U.S. Economy 1949-88. Washington,
D.C.: Board of Governors of the Federal Reserve System, 1989.
. Flow of Funds Accounts. Washington, D.C.: Board of Governors of the Federal Reserve System, various
dates.
Japan Economic Planning Agency. Annual Report on National Accounts, 1989. Tokyo: Economic Planning
Agency, 1989.
. Report on National Accounts from 1955 to 1969. Tokyo: Economic Planning Agency, 1989.
Lipsey, Robert E., and Irving B. Kravis. "Is the U.S. a Spendthrift Nation?" NBER Working Paper No. 2274.
Cambridge, Mass.: National Bureau of Economic Research, June 1987.
Lipsey, Robert E., and Helen Stone Tice, eds., The Measurement of Saving, Investment and Wealth. Chicago:
University of Chicago Press, 1989.
Organisation of Economic Co-operation and Development. Annual National Accounts, Paris, OECD, various
issues.
Â
Page 517
SOURCES
1. Reprinted from Joseph Pechman, ed., What Should Be Taxed, Income or Expenditure (Washington, D.C.:
The Brookings Institution, 1979), pp. 77-113. Reprinted by permission.
2. Reprinted from Charls E. Walker and Mark A. Bloomfield, eds., New Directions in Federal Tax Policy for
the 1980s (Cambridge, Mass.: Ballinger, 1983), pp. 229-252. (This is a revised version of a paper that
appeared in Tax Notes 16, no. 8 [August 23, 1982], pp. 715-723.) Reprinted by permission.
3. Reprinted from Charls E. Walker and Mark A. Bloomfield, eds., New Directions in Federal Tax Policy for
the 1980s (Cambridge, Mass.: Ballinger, 1983), pp. 243-261. (This is a revised version of a paper that
appeared in Tax Notes 39, no. 3 [April 18, 1988], pp. 383-391. Reprinted by permission.
4. Originally published by AEI Press, Washington, D.C. This is a revised version of "Consumption Taxes:
Some Fundamental Transition Issues," in Michael J. Boskin, ed., Frontiers of Tax Reform (Stanford, Calif.:
Hoover Institution Press, 1996), pp. 123-150. Reprinted with the permission of the American Enterprise
Institute for Public Policy Research, Washington, D.C., and Hoover Institution Press. Â© 1996 by the Board
of Trustees of the Leland Stanford Junior University.
5. Reprinted from Henry J. Aaron and Michael J. Boskin, eds., The Economics of Taxation (Washington,
D.C.: The Brookings Institution, 1980), pp. 281-298. Reprinted by permission.
420
421
6. Reprinted from George M. von Furstenberg, ed., The Government and Capital Formation (Cambridge,
Mass.: Ballinger, 1980), pp. 11-71. Reprinted by permission.
7. Reprinted from Charles R. Hulton, ed., Depreciation and the Taxation of Income from Capital
(Washington, D.C., The Urban Institute Press, 1981). Reprinted by permission.
8. Reprinted from Journal of Public Economics 15 (1981): 1-22. Â© North-Holland Publishing Company.
Reprinted with permission from Elsevier Science.
9. Reprinted from Social Choice and Public Decision Making Essays in Honor of Kenneth J. Arrow, vol. 1,
ed. Walter P. Heller, Ross M. Starr, and David A. Starrett (Cambridge: Cambridge University Press, 1986).
Â© 1986 Cambridge University Press. Reprinted with the permission of Cambridge University Press.
10. Reprinted from Charles R. Hulton, ed., Depreciation and the Taxation of Income from Capital,
Washington, D.C.: The Urban Institute Press, 1981, pp. 251-278. Reprinted by permission.
Â
Page 518
11. Reprinted from James Poterba, ed., Tax Policy and the Economy, Volume 12, Cambridge, Mass.: The
MIT Press, 1998, pp. 151-172. Reprinted by permission.
12. Reprinted from North Carolina Law Review 76, no. 1 (November 1997), pp. 223-231. Â© 1997 by the
North Carolina Law Review Association. Reprinted with permission.
13. Reprinted from Economic Effects of Fundamental Tax Reform, Henry J. Aaron and William G. Gale, eds.
(Washington, D.C.: The Brookings Institution, 1996), pp. 437-460, 463-464. Copyright Â© 1996 by The
Brookings Institution. All rights reserved. Reprinted by permission.
14. An earlier version of this chapter appeared as NBER Working Paper Number 5754. It is reprinted from
Tax Law Review 50, no. 4 (1995): 731-785. Copyright Â© 1995 by the New York University School of Law.
All rights reserved. Reprinted by permission.
15. An earlier version of this chapter appeared as NBER Working Paper Number 2906. Reprinted from
National Saving and Economic Performance, ed. B. Douglas Bernheim and John B. Shoven (Chicago:
University of Chicago Press, 1991), pp. 15-44. Copyright Â© 1991 by University of Chicago Press. All rights
reserved. Reprinted by permission.
16. Reprinted from Charls E. Walker, Mark A. Bloomfield, and Margo Thorning, eds., The U.S. Savings
Challenge: Policy Options for Productivity and Growth, Boulder, Colo.: Westview Press, 1990, pp. 31-75.
Reprinted by permission.
Â
Page 519
INDEX
A
421
422
Aaron, Henry J.
on financial services, 366
on inflation, 34, 38, 240
on taxation, 32, 119, 517
on value-added tax, 83
Ability to pay, 29, 61
Accelerated Cost Recovery System (ACRS), 65, 76
Accelerated depreciation, 169, 239.
See also Depreciation
consumption tax and, 195
equilibrium and, 214-218
inflation and, 230
marginal rates and, 192
Accounting.
See also NIPA
capital gains and, 9
cash flow, 159
complexity and, 334
consumption tax and, 4-6
corporate, 54
differences in, 96-97
inflation and, 149
national, 472-474, 483-487, 487-492
net worth and, 442-444
of real investment, 229
realization, 371-428
records and, 7
savings and, 55, 431-466
scheme for, 59-62
tax prepayment and, 51
value-added tax and, 95
Accretion, 25
422
423
Accrual
correctness and, 375, 379-380
income tax and, 53
measurement of, 49, 168
put-call parity theorem and, 380-381
Warren on, 372
wealth changes and, 8-10, 15-16
Accumulation, 46, 151-152.
See also Capital; Wealth
Cobb-Douglas preferences and, 158-159
equivalence and, 160
rate of, 205
revenue constraint and, 161-162
tax structure and, 157-167
utilitarian theory and, 26-27
Adjustment costs, 107
Administrability, 43
Adrion, Harold L., 366, 368
Age
demands of old, 101-102, 258-259
demands of young, 256-258
structure, 161-163
transitional incidence and, 80-81
Aggregate income, 145-146
Aggregate saving, 47-48, 96, 150, 157-158
Aggregate wealth, 491
Cobb-Douglas preferences and, 162
government policy and, 191
labor supply and, 164
low accumulation rate and, 205
423
424
market value data and, 448-452
measures of, 166, 434-438
savings and, 163, 190
Agriculture, 148
Allocation, 243-248
for endogenous variables, 252-256
model for, 248-252
parameter changes and, 261-264
Â
Page 520
Altig, David, 328-329
American Economic Association, 431
American Finance Association, 182
Andersen, Arthur, 116, 119, 365, 368
Ando, Albert, 461, 463, 514
Andrews, William D., xix, 31, 33, 197, 265
on consumption tax, 38, 143, 199
on measuring wealth, 168-169
Arbitrage, 18-19, 191, 207, 214
bonds and, 170
correctness and, 379-380
equilibrium and, 222
formulae for, 214, 218, 224, 230, 232-233, 385
ITC and, 224
marginal tax rates and, 220, 381-401
market equilibrium and, 377-378
mark-to-market and, 405
model for, 322
multiple tax rates and, 398-401, 408
424
425
OID bonds and, 378
price changes and, 386
profits and, 227
regulation of, 404-408
restriction of, 223
symmetry and, 389-392, 397
Archer, Bill, 85, 343
Armey, Richard, 85, 343
Arrow, Kenneth J., xviii-xix, 151, 199
on market equilibrium, 274, 279
Arrow-Debreu contingent claim markets, 442
Asimakopulos, A., 265-266
Asset Depreciation Range (ADR), 9, 147
Assets, 6, 50
ACRS and, 65-66
age and, 80-81
changes in, 7
depreciation and, 8, 214-218, 220-221
double-dipping and, 223
durability and, 124, 131-132
effective tax rate and, 296, 301
exponential decay and, 137
housing and, 10-11
income tax and, 147
inflation and, 227, 318
ITC and, 212
lump-sum tax and, 112
market value and, 431-466
measurement of, 49-50
national, 467-472, 487-492
NIPA saving and, 434
425
426
retirement and, 9-10
subtraction method and, 109
tax defects on, 15-16
transition effects and, 103-105
useful life of, 123
write-offs and, 206
Atkinson, Anthony B., 31, 35-38, 198
analysis with Stiglitz, 28, 156
on interest, 153
model with Sandmo, 26-28, 153, 198
on welfare, 199
Auerbach, Alan J., 102, 239, 424, 427, 461
Auerbach-Jorgenson Scheme and, 234-237
on capital, 36, 463
on corporate tax, 83
on efficiency, 209
on imputed interest, 407-408
on inflation, 240
on investment, 118, 329
methods of, 87, 408, 416, 428
National Balance Sheet problems and, 456
present-value depreciation and, 314
realization and, 376
on saving, 111, 431
on tax policy, 84, 119, 328, 425, 462
Averaging, 13-14, 53
tax defects and, 20
Bailey, Martin J., 182, 199-200, 202, 307
426
427
Bankman, Joseph, 339, 424, 427
Banks. See Financial services
Barham, Vicky, 368
Barro, Robert J., 200, 246, 265
Battle, Frank V., Jr., 425
Beaver, William H., 461, 463
Bequests, 6, 58
accounting scheme for, 61
averaging and, 20
distributional effects and, 78
income tax and, 146
life cycle model and, 151-152
savings and, 163-165
tax bases and, 44-45
X-Tax and, 71
Bernheim, B. Douglas, 119, 330
Bilateral financial instrument, 373
Bittker, Boris I., 426
Â
Page 521
Blades, Derek W., 463, 514
on depreciation, 474
Blinder, Alan S., 35, 38, 461, 463
on income distribution, 152, 200
Bloomfield, Mark A., 329, 517
Blueprints for Basic Tax Reform, 40
on capital gains, 9
on Cash Flow Tax, 113, 335, 339-341
on consumption tax, 3, 62-63, 83-84, 119, 143, 194-196
427
428
on corporations, 10, 180, 265
on savings, 197, 199, 203
on tax defects, 20
on wealth measurement, 168
Boadway, Robin, 329, 368
Bonds, 17-18, 166, 259.
See also Stocks
certainty model and, 270-274
consistency and, 399
correctness and, 375
discount, 378-380, 388-392
distribution model and, 250-251
equilibrium and, 192, 256, 260-261
GRD and, 410
income tax and, 147
long term discount, 417-421
OID, 372, 375
put-call parity theorem and, 381
realization accounting and, 388-392
tax shelters and, 169-171
time and, 382-385
uncertainty model and, 274-278
value-added tax and, 97-98
zero coupon, 386-387
Bookless land, 449, 452
Book value, 488, 491
Borrowing. See Loans
Boskin, Michael J., 38, 339, 517
on capital, 200
on cash flow tax, 328
428
429
on consumption tax, 31-32
NIPA saving and, 436, 514
on savings, 198, 239, 241-242, 431, 461-463
Bossons, John, 32, 38
Bosworth, Barry, 474, 514
Bradford, David F. (notes and other papers by), xix, 34, 222, 279
on consumption tax, 62, 168, 194, 200, 241, 330, 339
on corporate distributions, 265, 267-269
on dividends, 201
on inflation correction, 318
on marginal rates, 222
on savings, 119, 187, 239-240, 307, 461-463
on tax policy, 35-35, 38, 63, 83-84, 181, 200, 340, 368-369, 426-427, 515
on two-tiered cash-flow tax, 311, 313
on X-tax, 328
Brainard, William C., 464
Brannon,
Gerard M.,
140
Break, George F., 32, 38, 197, 200
Breeden, Douglas T., 461, 464
Brookings Institution, 4, 199
corporations and, 10
income measurement and, 12
Brown, E. Cary, 239-240, 330
theorem of, 325
Brown, Murray, 462, 464
Browning, Edgar K., 118-119
Bruce, Neil, 329, 368
Brumberg, Richard, 152, 202
Buchanan, James M., 265
Budget constraint, 22, 27, 152
429
430
demand functions and, 156
government saving and, 158
individual, 159-160
investment credit and, 208
temporary equilibrium and, 256
uncertainty model and, 276
wealth accumulation and, 157
Buiter, Willem H., 265
on savings, 164-165, 197-198, 200, 203
Bulow, Jeremy I., 462, 464
consumption tax and, 34, 40
on inflation, 331
Burbidge, J. B., 265
Bureau of
Economic
Analysis, 445,
449
Business, 89-90.
See also Corporations
household, 94-95
income tax and, 147
loans and, 101
one-time tax and, 99-100, 109-110
retail sales tax and, 86
transaction tax and, 91
two-tier cash-flow tax and, 313
value-added tax and, 88, 97-98
X-Tax and, 67
Â
430
431
Page 522
Business Transfer Tax (BTT), 66
X-Tax and, 74-75
Campbell, John, 461, 464
Cannan, Edwin, 202
Capital.
See also Accumulation; Wealth
accumulation and, 151-152, 159, 205
ACRS and, 65
complexity and, 334
cost of, 284-287
effective tax rate and, 287-291, 295, 304
exponential depreciation and, 125
formation of, 157-160, 206, 487-492
labor and, 26-27, 150
national, 467-472
old, 109-115
revenue and, 145
transitional incidence and, 78, 80-81
U.S. labor and, 166
vs. savings, 142
Capital gains, 8-9, 17, 50
agricultural sales and, 148
averaging and, 20
complexity and, 338
consumption tax and, 50, 169, 195
431
432
corporate tax and, 181-183, 185, 193
endogenous marginal rates and, 178
fixes for, 392-396
GRD/GTRs and, 409-416
income tax and, 146
indexing of, 12
inflation and, 19, 51, 149
NIPA and, 433-434, 445
Revenue Act of 1978 and, 143
vs. savings, 142
symmetry and, 388-392
tax defects on, 15-16
tax rule failure and, 205
tax shelters and, 170
Carter Commission, 10, 180
Cash flow, 8, 125-126
accounting of, 159
before realization, 409-417
capital and, 9, 285
consistency and, 398-401
corporate tax and, 184
distribution model and, 250
estimating future, 49
financial services and, 357-359
government and, 161
inflation adjustment and, 228-229
intermediate, 416
investment credit and, 129
personal saving and, 476-483
432
433
Ricardian behavior and, 246
taxation of, 66, 267, 335
two-tiered, 311
Warren on, 371-372
yield to maturity method and, 373
Cash Flow Tax (CFT), 55-57, 66, 267, 335
financial services and, 359
insurance and, 363-365
mortgage loans and, 361-362
transition to, 311-331
Ceilings, 222-223
Center for Operations Research and Econometrics (CORE), 265
Certainty model, 270-274
Ceteris paribus increase, 188
Chant, John F., 366, 369
Checking services, 348-351
Chia, Ngee Choon, 369
Christian, Ernest S., 365
Cnossen, Sijbren, 366, 369
Cobb-Douglas preferences, 28, 156, 159
aggregate wealth and, 164
special case for, 157-158
Summers' model and, 162
tractability of, 198
total wealth and, 158
utility function and, 23, 27
wealth accumulation and, 162-163
Commodities, 151-152, 156
Compensation tax, 67, 313
Complexity, 333-337
433
434
Comprehensive Income Tax (CIT), 55-58.
See also Income tax
Consistency, 382, 398-401
Constant flow output, 124, 134-135
Consumed income tax, 66, 86
Consumption, 4-5, 21-22, 43-44
accumulation and, 510
aggregate, 486
higher rates of, 509
household, 484
Â
Page 523
national, 467-474, 489-490
national saving and, 508
production and, 489
yield and, 19
Consumption demand function, 158
Consumption tax, 82.
See also C-tax
advantages of, 10, 115-116, 194-195
assets and, 178-179
averaging and, 14
complexity and, 333-341
concept of, 4-6, 43-45
corporations and, 53-54
definitions of, 58-62
description of, 66-67, 143-144
design and, 41-43
distributional effects and, 78-79
efficiency and, 20-25, 47-48
434
435
equity and, 25-31, 45-47
fundamental issues of, 85-116
gifts and, 6
graduated rates and, 174-175
grandfathering and, 315
Haig-Simons income and, 179
housing and, 94-95
hybrid rules and, 173-176
implementation of, 7-14, 49-51, 54-58, 195-196, 311-331
incentives and, 213-223
incidence analysis of, 65-66, 75-78
inflation and, 13, 51-52, 232-233
interest and, 220-221
moderating effects of, 108-111
old capital and, 111-115
pensions and, 169
profits and, 91-92
reforms and, 89-90
regressivity and, 72-73
savings and, 50, 66-67, 189, 195-197, 212-226
SHS concepts and, 438-439
S-I incentives and, 211-212, 226
tax defects and, 15-19
transitional effects of, 80-81, 98-108
wages and, 93-94
X-Tax and, 67-71, 73-75
Contingent returns, 372, 377
435
436
Continuous discounting, 423-424
Corlett, W. J., 28, 37-38
Corporate distributions
incidence/allocation and, 243-246
model for, 248-252
negative, 267-268
Corporations, 10, 16-17, 50, 53.
See also Business
allocation effects and, 243-266
effective tax rate and, 288-289, 291, 296, 298-299, 303-304
flat rate tax and, 54, 312
Hall/Jorgenson approach to, 282
income tax and, 42, 147, 180-188, 192-193, 196
National Balance Sheet problems and, 453, 455-457
1978 tax on, 145
trapped equity and, 452-453
Correctness, 375-376
inflation and, 379-380
risk-free term structures and, 416-417
time and, 397
Cost of capital approach, 281-284, 303-308
analytical framework of, 284-293
caveats for, 293-303
Cox, John C., 427
Credits. See Investment Tax Credit (ITC)
C-tax, 208-209.
See also Consumption tax
Auerbach-Jorgenson Scheme and, 235
436
437
real investment and, 231
savings treatment and, 212-226
S-I incentives and, 211, 226, 237-239
Cunningham, Noel B., 426
Dasgupta, P., 140
David, Martin, 35, 38
Davies, James B., 83-84, 369
Deaton, Angus S., 461, 463-464, 514
Debt neutrality, 165
Deductions
bad-debt reserve, 148
flat tax and, 89-90
with invoice-and-credit, 89
savings and, 55
Â
Page 524
Deductions (cont.)
transition incidence and, 77
value-added tax and, 92-93, 97-98
wages and, 102-103
Demand function, 155-156
Demand price, 219-220
Department of Commerce. See U.S. Department of Commerce
Depositor-side corrections, 351-353
Deposits, demand, 55, 348-350
437
438
imputed income and, 351-353
treatment of, 360
Depreciation, 8-9, 16, 50, 123, 138, 141
accelerated, 169, 195, 214-218, 230, 239
ACRS and, 65-66
complexity and, 334
cost basis for, 124
cost of capital approach and, 282
defects of, 15
defined, 445
economic, 127-130, 218
effective tax rate and, 297-298
exponential, 124-125, 137, 219, 232, 322-328
Hall/Jorgenson approach to, 282
increase of allowances for, 232
indexing of, 12
inflation and, 19, 51, 149, 226-227, 230
instantaneous, 227
interest and, 316-317, 320-321
investment tax credit and, 131-132
marginal tax rate and, 225
market value and, 443
measuring, 317-319
model for, 216-220, 327-328
NIPA and, 433, 445
present-value, 314
real estate and, 18
438
439
S-I incentives and, 225
tax rule failure and, 205
true, 126-127
Deservingness-to-pay, 29, 43, 53
Diamond, Peter A., 198, 200, 265
growth model and, 26
on national debt, 35, 38
on taxation, 36
Differential incidence, 76
Dildine, Larry, 197
Diminishing returns, 227
Direct grants, 208, 213, 238
incentives for, 223-226
Direct nonwealth receipts, 7
Direct services, 10-12, 55
Discount bonds. See Original Issue Discount (OID) bonds
Discount value, 96
Disincentive effects, 20-21, 208
Distributional effects, 53, 78-79
Distributional model, 246-247, 250
Dividends, 17
corporate tax and, 183-187, 193
double taxation of, 54, 243-244, 264
income tax and, 147
Dolde, Walter, 152, 203
Domar, Evsey, D., 117, 120, 329
Domenici, Pete, 86, 343
Domestic net worth. See Net worth
439
440
Donaldson, R. Glenn, 514
Double-dipping, 223
Double taxation, 243-244, 264
Dynamic efficiency, 26
Earned
income
credit, 73
Economic depreciation. See Depreciation
Economic efficiency. See Efficiency
Economics
aggregate saving and, 48
divergence in, 431
savings and, 141-203, 431-466
tax-exempt bonds and, 18
time and, 77
utilitarian theory and, 25
welfare and, 149-150
Economy, U.S.
See also National saving; NIPA
depreciation and, 8-9
dynamics of, 337-338
Effective tax rate, 281-283, 304-308
caveats of, 293-303
Efficiency, 27, 31
asset durability and, 130-131
440
441

Page 525
competition and, 153
exogenous features and, 42-43
gain measurement and, 24
government and, 154, 158-160
hybrid system and, 173
neutrality and, 124
one-time tax and, 110
output rate and, 134-135
Pareto optimality and, 26
production and, 215-216
savings taxation and, 149-157
S-I incentives and, 209-211
tax base choice and, 20-25, 47-48
Eisner, Robert, 431, 444, 462, 464
on capital shortage, 198, 200, 241
on savings, 239
Elderly. See Senior citizens.
Employee Stock Ownership Plan (ESOP), 147
Endogenous variables, 42-43
accelerated depreciation and, 192
corporate tax and, 182, 185
hybrid system and, 176-178
rational expectations and, 255-256
in temporary equilibrium, 252-255
Energy tax, 75, 80
English, Morley D., 328, 330, 368-369
441
442
on time varying tax rates, 358
Equations. See Mathematical equations
Equilibrium
accelerated depreciation and, 214-218
arbitrage and, 207, 222, 224, 377-378
asset market and, 247-248
bonds and, 192, 420-421
characteristics of, 260-261
durability and, 135
efficiency and, 132-133
financial market and, 267-279
Fisher's law and, 230-231
government and, 151, 267-279
graduated rates and, 174
marginal tax rate and, 387
modeling of, 81
net-of-tax payoff and, 395
social rate of return and, 214, 225
temporary, 251, 252-256
uniqueness of, 260
Equity, 25, 87, 101
arguments of, 114-115
certainty model and, 272
442
443
corporate, 181-182, 187, 484
distribution and, 142-143, 244-245, 250-251
double taxation and, 243
flat rate tax and, 54
horizontal, 29-30, 46-47, 209, 350
implementation of, 47
inflation correction and, 318
market clearing and, 259
nonutilitarian, 28-31
tax base choice and, 45-46
tax complexity and, 333
trapped, 452-453
utilitarian analysis of, 26-27
vertical, 29-30, 46-47, 350
welfare analysis of, 27-28
Equivalence
non-zero taxes and, 153
symmetry and, 382
Estate
tax, 42
Esubi (nonwealth receipts), 5-6
Euler's theorem, 259, 273
Ex ante averaging, 14
443
444
Exemptions, 12, 345-346
Exogenous features, 42-43, 152
Expenditure tax. See Consumption tax
Expensing of investment, 218-220
Exponential depreciation, 124-125, 135, 137, 219, 232
model for, 322-328
Exponentially declining output, 124
Ex post averaging, 14
Â
Page 526
Fabricant, Solomon, 240-241
Fairness, 42
Family status, 30, 41-42
Federal Reserve System, 434, 457, 469-470, 477
Federal sales tax, 85, 343
Feldstein, Martin S., xix, 239, 283, 307
on Cobb-Douglas preferences, 156
consumption tax and, 31, 34-37, 39
on flat rate tax, 12
formulation of, 23-24
growth models and, 151
on income, 149
on inflation, 19, 240-241, 300, 302
on investment, 150-151
savings and, 156, 182, 187, 201
on taxation, 200
on welfare gain, 23
444
445
Ferguson, Bradford L., 426
Fiekowsky, Seymour, 138, 306-307
Financial Accounting Standards Board, 443
Financial
instruments,
371-372, 374, 376,
421-428
arbitrage and, 377-378
bilateral, 373
capital gain fixes and, 392-396
discount bonds and, 378-380
general taxation of, 408-421
risk and, 401-408
tax arbitrage and, 377-378
time and, 381-401
Warren on, 380-381
Financial portfolio, 17, 82
age and, 81
migration and, 220
risk bearing and, 210
of young, 258
Financial services, 366-370
cash flow and, 357-359, 361-362
demand deposits and, 348-356, 360
exemptions, 345-346
existing taxation of, 347-348
other, 360-365
Fisher, Irving, 152, 201
445
446
Fisher's law, 230-231, 306
cost of capital approach and, 286-287
effective tax rate and, 300, 302-303
parameter values and, 292-293
Flat rate credit, 123
Flat rate tax, 22, 30, 86
averaging and, 53
bank treatment and, 343
bequests and, 78
consumption tax and, 66-67, 175-176, 196
corporations and, 54, 312
ease of, 312
equilibrium for, 215
as graduated tax, 311
interest and, 317
model for, 167-168
partial integration and, 244
savings and, 189
social rate of return and, 225
stocks and, 268
subtraction method and, 89-90
tax reform and, 85
on wages, 153
X-Tax and, 67-68, 73
Flow of Funds Account (FFA)
approach to, 281
446
447
Flow of Funds Division, 457, 469, 477, 484
Fogler, Elizabeth, 514
Foley, Duncan K., 369
Formulae. See Mathematical equations
Foster, George, 461, 464
Fraction financing, 220-221
Fraumeni, Barbara M., 287, 307, 514
Freund, James L., 515
Friedman, Milton, 152, 201
Frishkoff, Patricia A., 443, 461, 464
Fullerton, Don, 197
on effective tax rates, 281-308
Gain Reference Date (GRD), 427-428
continuous discounting and, 423-424
Â
Page 527
illustrated approach to, 412-416
short term rates and, 420-421
Gain Tax Rate (GTR), 427-428
continuous discounting and, 423-424
447
448
Gale, William G., 119, 517
Galper, Harvey, 83, 138, 366, 369
Gas expenses, 148
Gentry, William M., 339
Gergen, Mark, 424
Gibbons, Samuel M., 343, 346, 369
Gibson, Charles H., 443, 461, 464
Gifts, 30, 58
accounting scheme for, 61
averaging and, 20
equity and, 47
income tax and, 6, 209
1978 tax on, 145
tax bases and, 44-45
X-Tax and, 71
Gokhale, Jagadeesh, 102, 119
Golden rule, 26-27
Goldsmith, Raymond W., 431, 464
Goode, Richard, 32, 36, 39, 197, 201, 365
Gordon, Roger H., xix, 281
on corporations, 120, 187
on dividends, 201
on effective tax rate, 305, 307
investment and, 138
savings and, 197, 239, 242
Goulder, Lawrence H., 84
Government.
See also National saving; NIPA
budget policy, 191
448
449
corporate distribution and, 243-266
debt and, 27-28
deficits, 267-279
saving and, 27-28, 158-160
stocks, 485
wealth accumulation and, 163-164
Graduated-rates, 174-175
consumption and, 66-67
Graetz, Michael J., 330
Grandfathering, 315, 320
model for, 326-327
rules for, 113
Gravelle, Jane G., 281, 307, 328, 330
Green, Jerry R.
on tax reform, 201
Grieson, Ronald E., 151, 201
Griffith, Thomas, 339
Gross Domestic Product (GDP), 473
power parity and, 500
savings and, 493
Grossman, Gene M., 83
Gross National Product (GNP), 432, 468, 473
capital stock and, 445
deflator, 11, 488
1950-1988 period, 472
wealth and, 496
449
450
Gross tax rate, 288
Grubert, Harry, 369
Hague, D. C., 28, 37-38
Hahn, F. H., 274, 279
Haig-Simons income, 4
accounting and, 7
averaging and, 20
deviation from, 8
equity and, 333
formula for, 5
market value and, 8
measurement of, 23
nonwealth receipts and, 6
principle of, 143
rejection of, 194
taxation of, 168
wealth and, 179
Half-way house, 167-171
Hall, Bronwyn, 461
Hall, Robert E., xix, 461
on capital gains, 284
on depreciation, 307
effective tax rate and, 290-291, 300, 306
Â
450
451
Page 528
Hall, Robert E. (cont.)
financial services and, 344, 366-369
flat tax and, 85, 118, 120, 311, 343
imputed income and, 352
on investment, 139, 198, 239, 242, 283, 308
on life cycle model, 198
on marginal tax rates, 281
models of, 322
on parameter values, 291-293
tax policy and, 162, 330, 464
on transition incentives, 328
X-Tax and, 67
Harberger, Arnold C.
effective tax rates and, 281, 307-308
investment and, 140, 218, 242
savings and, 180-182, 198-199, 202
Hariton, David P., 341, 425
Hayashi, Fumio, 474, 514
Heller, Walter P., 517
Hellmuth, William F., 33
Hellwig, Martin, 265
Hendershott, Patric H., 239, 242, 508
Henderson, Yolanda K., 366, 369
Hewson, Barbara, 197
Hicks, John R., 431, 465
Hines, James R., Jr.
on investment, 329, 463
451
452
on tax policy, 462
Hoffman, Lorey Arthur, 366, 369
Holloway, Thomas M., 514
Home production, 153-154, 156
Horioka, Charles, 514
Horizontal distributional effects, 79
Horizontal equity, 29-30, 46-47, 209
House of Representatives, 85
House Ways and Means Committee, 85, 206, 343
Housing, 10-11, 50
imputed consumption and, 19
income tax and, 94-95, 147
measurement of, 49
reform effects and, 107
X-Tax and, 70
Howitt, Peter, 87, 118, 120, 330
Hu, Sheng-Cheng, 239, 242
Hubbard, R. Glenn, 339
Huber, Alan M., 461-463, 514
Hulton, Charles R., 462, 517
on depreciation, 282, 307-308, 445, 465
Hybrid system, 171-172, 194-195
endogenous marginal rates and, 176-178
graduated rates and, 174-175
Hypothetical project approach, 281, 303-308
caveats for, 293-303
452
453
Ibbotsen, R. G., 425
Ideal tax model, 48
Immerwahr, John, 510, 515
Implementation
accrued wealth and, 8-10
averaging and, 13-14
of consumption tax, 7-14, 49-51, 54-58, 86, 195-196, 311-331
direct consumption and, 10-12
of equity, 47
of income tax, 4, 7-14, 49-51
transitional effects of, 98-108
of uniform taxation, 87-98
Imputed consumption yield, 19
Imputed income, 351-353
Imputed interest, 350-351, 393
marginal tax rate and, 405-407
realization and, 394-396, 407-408, 411
Imputed transactions, 7-11
Incidence analysis, 65-66, 82
differential, 76
distributional effects and, 78-79, 243-266
for endogenous variables, 252-256
steady state effects and, 75-76
time perspective and, 76-77
transitional, 77-78, 80-81
453
454
X-Tax and, 67-71, 73-75
Income
averaging of, 13-14
from capital, 8
definition of, 43
measurement of, 12
Â
Page 529
NIPA definition of, 468
timing and, 372
Income Tax, 3
accrual and, 53
ACRS and, 65
cash flow and, 66
concept of, 4-6, 43-45
corporations and, 53-54, 180-188, 192-193, 196
defects of, 15-20, 209
definitions of, 58-62
demand deposits and, 349-351
design and, 41-43
efficiency and, 20-25, 47-48
equity and, 25-31, 45-46
gifts and, 6, 209
housing and, 94-95, 147
implementation of, 4, 7-14, 49-51
454
455
investment and, 92, 147
savings and, 142, 146-149, 167-171
SHS concepts and, 438-439
tax reform and, 85-116
transactions and, 91, 373-374
transition from, 103, 319-320
value-added tax as, 95-97
X-Tax and, 68
Indexed bonds, 81
Indexing
of real investment, 229
type I, 233
of yields, 12-13
Individual Retirement Accounts (IRAs), 141, 206, 211
Individual tax base, 41, 53-54, 86
judging of, 42-43
Infinite horizon world, 248-249, 270
Infinitely elastic capital, 106
Inflation, 3, 86, 318, 397
adjustment for, 228-230
background on, 226-228
cash flow and, 8
consumption tax and, 52, 232-233
455
456
effect of, 148-149
prices and, 12, 100
S-I incentives and, 206, 211-239
stocks and, 19, 227, 318
tax defects of, 19
tax implementation and, 12-13
transitional incidence and, 81
Ingersoll, Jonathan E., Jr., 427, 461, 465
Inheritance, 6, 30
equity and, 47
taxation and, 42
X-Tax and, 71
Institute for Fiscal Studies, 120, 144, 202, 242
Institutional-side corrections, 351-353
Insulating, 315-317
Insurance, 362-365.
See also Financial services; Life insurance
Intangible expenses, 148
Interest, 16, 125-126
accounting scheme for, 59-62
adjusting treatment of, 220-221
depreciation and, 232, 316-317
effective tax rate and, 293-297, 303
elasticity of, 155
456
457
imputed, 393-397, 400-401, 411
inflation and, 51-52, 228-232
neutrality and, 124
preexisting commitments and, 110-111
tax changes and, 86
transition effects and, 103-108
Warren on, 371-372
Intergenerational links, 163-165.
See also Bequests; Gifts
Intermediate transactions, 400, 405
Â
Page 530
Internal revenue code, 180, 184
Inventory, 205, 313-314
Investment.
See also S-I incentives
capital income and, 92-93
cost of capital approach and, 285-286
defects of, 15
definition of, 207
design issues in, 205-242
effective tax rate and, 291, 298, 304
457
458
graduated rates and, 175
Hall/Jorgenson approach to, 281-282
hypothetical project and, 281-282
income tax and, 92, 147
inflation correction and, 318
mathematical models for, 125-138
NIPA and, 431, 444-446
partial expensing of, 218-220
preference of, 16
special provisions for, 148
tax arbitrage and, 191-192
tax credit and, 15, 123-138
tax defects and, 18-19
tax wedge and, 150-151, 190
transition effects on, 77, 111-112
Investment Tax Credit (ITC), 12, 15-16, 141, 212, 239
direct grant incentives and, 223-226
effective tax rate and, 296
inflation and, 230
mathematical models for, 125-138
neutrality and, 123-138
problem of, 129-130
Invoice-and-credit, 89
458
459
Japan, 500-503, 506
consumption and, 509
sources for, 512-513
Johansson, Sven-Erik, 329-330
Johansson-Samuelson theorem, 324
John M. Olin Foundation, 365, 424, 461
Joint Committee on Taxation, 116, 120
Jorgenson, Dale W.
cash flow tax and, 322, 330
on corporations, 231
effective tax rates and, 281, 283-284, 306-308
on inflation, 300, 302
investment and, 139
savings and, 198, 201, 239, 241-242, 514
Sullivan and, 300-302, 308
Kaldor, Nicholas, 158
Auerbach-Jorgenson proposal and, 240
consumption tax and, 33, 39, 49, 63, 83
Kaplow, Louis, 365
statutory change and, 328, 330
Kau, Randall K. C., 425
Kay, John A., 31, 33, 39, 63
Keogh Plans, 16
459
460
C-tax and, 222
S-I incentives and, 211
tax changes and, 206
Keuschnigg, Christian, 87, 120
Keynesian response, 157-158
King, Mervyn A., xix
corporations and, 247-248, 265
effective tax rates and, 281, 306, 308
savings and, 181, 187, 202, 242
King, Thomas, 307
Klein, William A., 427
Kotlikoff, Laurence J.
cash flow tax and, 328-329, 331
consumption tax and, 83-84, 87
on fiscal policy, 102
savings and, 111, 119-120, 279, 462, 465, 482, 514
Kravis, Irving B., 436, 465, 515
Kronmiller, Thomas, 305-306
Kurtz, Jerome, 365
Kurz, Mordecai, 151, 199
Kyle, Pete, 279
Â
Page 531
Labor
aggregate wealth and, 157-158, 164
Atkinson-Stiglitz analysis of, 28
460
461
endogenous features and, 43
revenue and, 145
savings and, 150
supply, 155-156
tax effects on, 102-103
transitional incidence and, 80
U.S. capital and, 166
X-Tax and, 71-72
Labor supply function, 158
Lagrange multiplier, 178
Last In First Out (LIFO), 205
Lau, Lawrence J., 436, 463
Lederer, Walther, 515
Leisure, 153, 155, 160
Leveraged tax shelter, 169-171, 176
Life cycle model, 151-152, 157, 163, 178
Life insurance, 8-10, 17, 364-365
existing taxes and, 348
income tax and, 146
tax defects on, 16
Lifetime wealth, 30, 46-47
Linearity, 374, 392.
See also Symmetry
Arrow-Debreu markets and, 442
Auerbach method and, 408
tax arbitrage and, 404
461
462
Lipsey, Robert E., 436, 461, 463, 465, 514-515
Liquidity, 152
Litzenberger, Robert H., 187, 202
Loans, 17-18, 55
to self, 179
subtrction method and, 88
Lodin, Sven-Olof, 31, 39, 63
Logue, Kyle D., 369, 427
Long, Russell B., 139
Long Amendment, 139-140
Look-back method, 408
Lugar,
Richard, 85,
343
Lump sum tax, 27-28, 111-112
Lyon, Andrew B., 328, 331
McDaniel, Paul, 365
McHugh, Richard, 35, 40
Mackie, James, 369
McLure, Charles E.
cash flow tax and, 328, 331, 366, 369
Main, Brian G. M., 369
Malkiel, Burton G., 239, 242
Marchand, Maurice, 265
Marginal tax rate, 12
demand price and, 219-220
depreciation and, 225
double-dipping and, 223
462
463
effective tax rate and, 290-291, 304
fraction financing and, 220-221
Hall/Jorgenson approach to, 281-282
imputed income and, 352-353
inflation and, 227
interest and, 229
multiple, 398-401, 408
nominal interest and, 397
OID bonds and, 379
single, 381-398, 405-408
time and, 381-401
Marginal utility of income, 154
Market clearing, 259-260
Market value, 461-466
Arrow-Debreu concept and, 442
GNP and, 435-436
National Balance Sheet problems and, 452-457
NIPA saving and, 434-438, 444-446, 457-460
saving and, 431-434, 446-448, 483-487
time-series data for, 448-452
wealth and, 438-441
Mark to market method, 371-372, 392-393
Mathematical equations
arbitrage profit, 214, 218, 224, 230, 232-233, 385
Auerbach-Jorgenson Scheme, 236
bond repayment, 420-421
463
464
budget constraint, 152, 159, 440-442
certainty model, 271-274
Cobb-Douglas preference, 156-158
Â
Page 532
Mathematical equations (cont.)
consumption sequence, 161
continuous discounting, 423-424
cost of capital approach, 286-287
credit/durability, 224
demands of old, 258-259
demand function, 156
demand price, 219
demands of young, 257-258
depreciation, 229, 327-328
effective tax rates, 287-291
endogenous variables, 177-178, 252-256
equilibrium, 215, 263
Fisher's law, 230-231
grandfathering, 326-327
GRD/GTR, 427-428
inflation, 227
investment, 125-136, 138, 383
market clearing, 259-260, 393
net-of-tax payoff, 395
no profit condition, 221, 226, 390, 392
OID clearing price, 387
population growth, 161
present value of revenues, 160
production efficiency, 215-216
464
465
rate changes, 322-326
rate of return, 214, 219
risk payoff, 405
second-best optimality, 21-23, 155
time and investment, 383
time path, 261
utility function, 154
wealth, 157-158
Matson, Nils, 365
Matzler, L. A., 329
Meade, James E.
consumption tax and, 31, 39, 63, 120
savings and, 144, 202, 242, 465
Meade Committee
inconsistencies found, 144
Measurement, 49-50
of aggregate wealth, 434-438
of efficiency, 24
of income, 12, 23
by market value, 446-448
National Balance Sheets and, 436
of national saving, 429-515
of stocks, 49
of wealth, 165-168, 487-492
Merrill, Peter R., 346, 365-366
Merton, Robert C., 461, 465
Messere, Kenneth C., 369
465
466
Meyer, Laurence H., 83
Microeconomics theory, 151-152
Mieszkowski, Peter, 31, 39
Miller, Merton H.
market equilibrium and, 268, 279
savings and, 143, 181-182, 185-186, 188, 199, 202
Mirrlees, James A., 36, 39, 200
Models.
See also Mathematical equations
analysis without taxes, 322-323
budget constraint, 439-442
cash flow, 161
certainty, 270-274
C-tax incentives, 212-223
depreciation, 216-220, 327-328
distributional, 246-247, 250
effective tax rate, 281-308
endogenous marginal rates, 178
equity, 245-246
grandfathering, 326-327
GRD/GTRs, 412-416
growth, 151
life cycle, 151-152, 157-163, 178
rate changes, 322-326
tax on distributions, 248-266
uncertainty, 274-278
Modigliani, Franco
exogenous variables and, 152
market equilibrium and, 279
savings and, 202, 461, 463
466
467
Modigliani-Miller theorem, 268
Mortgage loans, 360-362
Much, Kathleen, 461
Multiple tax rates, 398-401, 408
Musgrave, John C., 462, 465, 514
Musgrave, Peggy B., 36, 40
Musgrave, Richard A., 36, 40, 117, 120
Muth, J., 251, 265, 272, 279
Mutti, Jack, 331
Mutual funds, 9
National Balance Sheets
problems of, 452-457
U.S. savings performance and, 457-460
Â
Page 533
National Bureau of Economic Research, 279
National Income Account (NIA), 436
National Income and Product Account. See NIPA
National saving, 429-430, 462-466, 512-515.
See also Savings
account measures and, 472-476
defining, 467-468
market value and, 431-438, 446-452, 483-487, 492-506
National Balance Sheets problems and, 452-457
trends of, 506-511
U.S. performance, 457-461, 472-476, 487-492
467
468
wealth and, 438-446, 469-472
Negative distributions, 267-268
Negative saving, 195
Net Domestic Product (NDP), 473-474
Net National Product (NNP), 468, 473
Net-of-tax basis, 288
Net private domestic investment, 469
Net product. See Income
Net revenue requirement, 27
Net worth, 4-5, 43-44, 434, 512.
See also Savings; Wealth
accounting and, 59-62, 442-444
GNP and, 435-436
national, 469-470
Neubig, Thomas S., 368, 370
Neutrality, 123-124
mathematical equations for, 125-138
NIPA, 431
accounting and, 444
consumption and, 446, 489
definitions of, 432, 467-468
net private domestic investment and, 469
1950-1988 period, 472
NIPA saving, 460-461, 473-474.
See also National saving
investment and, 444-446
National Balance Sheets and, 433-438
468
469
Nonutilitarian equity, 28-31
Nonwealth receipts, 5-6, 29, 44, 46
direct type, 7
equity and, 47
gifts and, 45
measurement of, 50
wages and, 49
Non-zero taxes, 153
Nunn, Sam, 86, 343
Obstfeld, Maurice, 436, 465
Oil expenses, 148
Okner, Benjamin A., 84
Old capital, 86, 314
protecting, 109-110
tax issues of, 111-115
Older generation See Senior citizens
One-time tax, 99-100, 110
equity and, 114
lump-sum character of, 109
reputation of, 112
Optimal commodity tax theory, 28, 45
second-best, 154-156
Organisation for Economic Co-operation and Development (OECD), 472-475, 506
Organizing principle, 143
Original Issue Discount (OID) bonds, 372, 375, 378-380
clearing price of, 387
imputed interest and, 405-406
469
470
risk and, 402
symmetry and, 388
time and, 383, 384
Pareto optimality, 26
Partial expensing, 218-220
Partial integration, 243-244
Paul, Deborah L., 329, 331, 339-341
complexity and, 333-336, 338
Payrolls, 42, 145
Pechman, Joseph A., 517
savings and, 197, 200, 202, 240, 242
Peek, Joe, 431, 465
Pensions, 7-10, 49, 58
ceilings of, 222-223
collateral and, 179
consumption treatment and, 169
income tax and, 146
Â
Page 534
Pensions (cont.)
National Balance Sheet problems and, 455
tax changes and, 206
Percentage deletion, 148
Period-by-period interest, 394-395
Period-by-period sum, 374
Perpetuity, 6, 130
Personal consumption tax, 3, 32-40.
470
471
See also Consumption tax
background for, 4-6
equity and, 25-31
implementation and, 7-14
income tax defects and, 15-20
Pestiau, Pierre, 265
Phelps, Edmund S., 37, 40
Pieper, Paul J., 444, 464
Poddar, Satya N., 366, 368-369
cash flow type tax and, 328-329
on time varying tax rates, 358
Point input investment, 124, 132, 134-135
Polar case, 101
Policy issues
of consumption tax, 3-40
corporate allocation effects, 243-266
savings/investment design and, 205-242
savings treatment and, 141-203
tax structure and, 157-167
Politics, 112-113
tax policy and, 3, 142, 194, 311, 333
tax shelters and, 170, 176
two-tiered cash flow tax and, 321
Pollak, Robert A., 462, 465
Population growth, 251
age structure, 161-163
capital and, 26
golden rule and, 27
Samuelson on, 161
471
472
Portfolios, 17, 82
age and, 81
migration and, 220
risk bearing and, 210
of young, 258
Poterba, James
cash flow type tax and, 328, 330
national saving and, 436, 461-462, 465, 514
Present-value depreciation, 314
Prices
aggregate wealth and, 164
changes in, 12, 100
demand, 219-220
efficiency gain and, 24
marginal tax rates and, 386
market value measurement and, 447-448
OID clearing, 387
tax changes and, 86
Princeton University, 365
Problem of the second-best, 21
Production tax, 90
Profits tax, 91-92
Progressivity, 46, 98
Purchasing power parity, 500, 513-514
Pure consumption, 22
472
473
Pure income, 22
Put-call parity theorem, 380-381
Rabushka Alvin, xix, 118, 330
flat tax and, 86, 311, 343
imputed income and, 352
X-Tax and, 67
Ramaswamy, Krishna, 187, 202
Rate schedules, 12, 14, 53
Rationality of expectations, 251, 255-256
Real estate, 9, 18
Realization accounting, 371-376, 422-428
background of, 377-381
capital gains symmetry and, 388-392
marginal tax rates and, 381-398
multiple tax rates and, 398-401
put-call parity theorem and, 380-381
risk and, 401-417
symmetry fixes and, 392-396
Recapture, 125, 127
Receipts
differential incidence and, 76
inflation and, 51
investments and, 11
Â
473
474
Page 535
nonwealth, 5-7, 29, 44-47, 49-50
X-Tax and, 68
Record keeping, 7
Regressivity, 72-73
Regulation Q, 13
Reiter, Michael, xix
Research and development provisions, 148
Retail sales tax, 75, 86-87
tax reform and, 85
Retirement, 17.
See also Life insurance
rights of, 9-10, 50
savings and, 337
tax defects on, 16
Return-to-scale function, 249
Revenue
agencies, 8
cash flow and, 160-163
1978 taxes and, 145
optimality and, 154-155
second-best optimality and, 156
Revenue Act of 1978, 143, 170
Ricardian behavior, 246
Ring, Diane, 424
Risk, 92-93, 375-376
effects of, 401-403
term structure and, 409-417
474
475
time and, 403-408
Robinson, Marc S., 462-463, 514
Rose, Manfred, 329, 331
Rosen, Harvey S., 197
Ross, Stephen A., 427
Royal Commission on Taxation, 180, 197, 202
Ruggles, Nancy, 431, 466
Ruggles, Richard, 431, 466
Ryan, Stephen, 461, 463
Sales tax, 75, 86-87
tax reform and, 85
Samuelson, Paul A., 270, 329
cash flow type tax and, 331
investment and, 139-140, 239, 242
theorem of, 324
Sandmo, Agar
cash flow type tax and, 331
consumption tax and, 26, 36-38
model with Atkinson, 26-28, 153, 198
Sarkar, Shounak, 87, 118, 120
Savings, 3, 6, 17.
See also National saving
aggregate wealth and, 163, 190
475
476
averaging and, 53
concept of, 4-5, 44, 207
Congress and, 15
consumption tax and, 50, 66-67, 189, 195-197, 212-226
corporation tax and, 193
design issues in, 205-242
determining tax rate of, 149-157
distribution model and, 246-247, 250
economics of, 141-203
flow of funds and, 476-483
government, 158-160
GRD/GTRs and, 409-416
half-way house and, 167-171
international, 492-506
loans and, 18
market value and, 446-448, 483-487
measurement of, 49, 50
national, 431-466
NIPA and, 433-438, 444-452, 460-461, 473-483
protecting, 109-110
retirement and, 337
second life-period and, 248-249
taste for, 79
tax base choice and, 44-45, 47-48, 144-149
tax defects on, 16
transition effects on, 111-112
476
477
treatment of, 6, 46, 55
USA tax and, 86
U.S. performance, 457-460
vs. income tax, 47-48
wealth and, 432-433
yield indexing and, 13
Scarborough, Robert H., 120, 424-425
Â
Page 536
Schanz-Haig-Simons (SHS) income
market value and, 432, 434, 438-439, 445
national accounts and, 485-486
Schenk, Deborah, 424-426
Scholes, Myron S., 143, 185-186, 199, 202
Schutzer, George J., 365
Second-best optimality, 155-156
Second-best
taxation, 21-24
net revenue and, 27-28
Self-averaging, 30, 113
Senate Finance Committee, 139, 206
Senior citizens, 101-102
age structure and, 161-163
demands of old, 258-259
Shapiro, Eli, 241
Shares, 10, 53-54.
See also Bonds; Stocks
Shaviro, Daniel, 314, 328, 331, 424-425
477
478
Shelters, tax, 17-19
paradigm of, 169-171
politics and, 176
Sheshinski, Eytan
Shiller, Robert, 461
Shleifer, Andrei, 456, 466
Short period payments, 29
Short-term rates, 417-421
Shoven, John B., 518
effective tax rates and, 281, 307-308
savings and, 181, 202, 240, 242, 431, 461, 464, 466
Shuldiner, Reed H., 382, 425-427
on consistency, 398
S-I incentives, 206, 238-239
absence of inflation and, 212-226
classifying, 211-212
depreciation and, 225
equity and, 208-209
inflation and, 206, 211-237
simplicity and, 210-211
Simons, Henry C., 143, 202
Simple flat tax. See Flat rate tax
Simplicity
478
479
Sims, Theodore S., 427
Sinn, Hans-Werner
Brown theorem and, 325
Johansson-Samuelson theorem and, 324
Slemrod, Joel
savings and, 149, 182, 201
Smart, Scott, 461
Smetters, Kent, 328-329
Smith, Adam, 202
Social rate of return, 214, 216-217
direct grants and, 224
flat rate and, 225
under various parameters, 294
Social
security,
9-10, 42
consumption and, 509
measured wealth and, 167
tax reform and, 85
Sodersten, Jan, 514
Soerensen, Peter Birch, 331
Source model, 59-62
Special Drawing Rights (SDRs), 449, 469
Specialized production, 81
Stage of production tax, 90
479
480
Standard treatment, 168-169
Starr, Ross M., 517
Starrett, David A., 279, 517
Static efficiency conditions, 26
Steady-state growth, 26-27
Steady-state incidence, 75-77, 130
distributional effects and, 78-79
Stem, N. H., 36, 200
Steuerle, C. Eugene, 35, 40, 242, 426
St-Hillaire, France, 83
Stiglitz, Joseph E.
on Cobb-Douglas preferences, 156
consumption tax and, 31, 35, 37-38
corporations and, 247-248, 265
effective tax rates and, 306, 308
investment and, 140
market value and, 279, 462, 466
savings and, 153, 181, 199, 203, 240, 242
Stocks, 17, 101
Â
Page 537
consumption tax and, 13, 52, 175-176
corporate tax and, 180, 183-188, 244-245, 249-250
double taxation and, 243
effective tax rate and, 289, 291, 305
government, 485
inflation and, 19, 227, 318
480
481
measurement of, 49
NIPA and, 431
trapped equity and, 452-453
Strnad, Jeff, 398, 425, 427-428
Structures
age, 161-163
effective tax rate and, 297
term, 409-421
Sturm, Peter, 436, 463, 474, 514
Subsidies, 27
assets and, 123
cash flow model and, 161
efficiency and, 124
Subtraction method, 87-88, 366
business and, 109
housing and, 95
reforms and, 89-90
stocks and, 88
transaction tax and, 91
value-added tax and, 96, 313
Sullivan, Martin A.
Jorgenson and, 300-302, 308
Summers, Lawrence H.
effective tax rates and, 307
481
482
NIPA saving and, 436, 456, 461-462, 464-466
on inflation, 300, 302
savings and, 149-150, 162-163, 201, 203, 241
Sunley, Emil M., Jr., 35, 40, 137-140
Sweden, 497, 503-505, 513
Symmetry, 382-385
capital gains and, 388-396
imputed interest and, 395
necessity of, 387-388, 396
System of National Accounts (SNA), 472-474
Tabushka, Alvin, 120
Tait, Alan A., 117, 120, 366, 370
Tax arbitrage, 18-19, 191, 207, 214
bonds and, 170
formulae for, 214, 218, 224, 230, 232-233, 385
ITC and, 224
marginal tax rates and, 220, 381-401
market equilibrium and, 377-378
mark-to-market and, 405
model for, 322
multiple tax rates and, 398-401, 408
OID bonds and, 378
price changes and, 386
profits and, 227
regulation of, 404-408
restriction of, 223
symmetry and, 389-392, 397
482
483
Taxation, general
averaging and, 13-14, 20, 53
background on, 226-228
cuts and, 19
design problems, 41-42
double, 243, 264
economic theory of, 181
exempt bonds, 17-18
flexibility in cash flow, 311-331
grandfathering and, 113, 315, 320, 326-327
implementation of, 7-14
lump sum, 27-28, 111-112
neutrality and, 123-138
nonutilitarian analysis of, 28-31
optimal commodity theory, 28, 45, 154-156
personal consumption, 3-40
pitfalls of effective rates, 281-308
prepayment and, 11, 14, 51, 53, 55
rate schedules and, 12
reform of, 85-116, 143, 170, 328, 367
of risky instruments, 401-408
savings/investment design and, 205-242
second-best, 21-24, 27-28
Â
483
484
Page 538
Taxation, general (cont.)
timing of, 267-279
two-tier, 148
utilitarian analysis of, 26-27
welfare analysis of, 27-28
Tax Calculation Account (TCA), 358-359
Tax Reform Act
of 1969, 143
of 1976, 143, 170
of 1986, 113, 328, 367
Tax shelters, 17-19
paradigm of, 169-171
politics and, 176
Tax wedge, 156-157, 190, 209
effective tax rate and, 288-291, 297-300, 303
elimination of, 219
social return and, 216
Temporary equilibrium, 251-256
effects from change, 261-264
10-5-3 proposal, 206
Term structures
before realization, 409-412
Three-martini lunch, 8
Thuronyi, Victor, 331
Tice, Helen Stone, 436, 463, 465, 514-515
484
485
Tideman, T. Nicholaus, 137, 140
Time
capital gain fixes and, 392-396
income tax and, 372
marginal tax rates and, 381-401
risk and, 403-408
tax bases and, 95
zero coupon bond and, 386-387
Tobacco tax, 80
Tobin, James
savings and, 152, 164-165, 198, 200, 203
Toder, Eric, 239, 241
Toll charge, 186-187
Tracing rules, 170
Transactions
declining costs of, 375
financial instruments model and, 372
income tax and, 373-374
intermediate, 393, 405
marginal tax rates and, 386
market, 43
measurement of, 50
risk and, 402
485
486
tax base and, 7, 91
value-added tax and, 91, 353-357
Warren on, 380
Transition issues, 77-78, 82, 86-87, 98
age differences and, 80-81, 101-102
cash-flow-type tax, 311-331
effects of tax shifts, 103-105
labor and, 102-103
moderating of, 108-111
one-time asset tax, 99-100
price changes, 100
wealth tax effect, 100-101
Trapped equity, 452-453
True depreciation, 126-127
Tulkens, Henry, 265
2-4-7-10 proposal, 206
Two-tiered cash flow tax, 340
savings and, 148
variation and, 313-317
Type I indexing, 233
Ullman, Al, 206
Uncertainty model, 274-278
Uniform taxation, 86
implementation of, 87-98
486
487
United Kingdom, 498-499, 503
imputed transactions and, 11
Meade committee and, 3
savings and, 494
tax system of, 144
United Nations, 472
Unlimited Savings Allowance (USA) tax, 85-86, 343, 353
Â
Page 539
Untangling the Income Tax (Bradford), 439
U.S. Congress, 203
Joint Committee on Taxation, 116, 120
savings and, 15
tax cuts and, 19
tax-exempt bonds and, 18
U.S. Department of Commerce
book value and, 491-492
consumption estimates, 445
efficiency and, 350
government stocks and, 485, 488
income data, 467
OECD and, 472-473
U.S. Treasury Department, 3, 203.
See also Blueprints for Basic Tax Reform
Comprehensive Business Income Tax and, 118
Utilitarian analysis, 25-27
Utility function, 153-154, 162
487
488
Value Added Tax (VAT), 42, 66, 86-87
business and, 353-354
corrections and, 359
deductions and, 97-98
demand deposits and, 354-355
exemptions and, 346
income and, 92, 95-97
insurance and, 362-365
mortgage loans and, 361
reforms and, 85, 89-90
transactions and, 91, 353-357
transitional effects of, 98-108
two-tier cash-flow tax and, 313-314
wages and, 93-94, 97-98
X-Tax and, 67, 71, 73-74
Vertical distributional effects, 78-79
Vertical equity, 29-30, 46-47
Vickrey, William, 34, 40
von Furstenberg, George M., 517
investment and, 140
realization accounting and, 426
von Schanz, Georg, 32
488
489
Wages, 5, 102-103
flat tax and, 86
income tax penalty of, 209
1978 tax on, 145
tax ability and, 45
transition incidence and, 78
value-added tax and, 93-94, 97-98
X-Tax and, 71-72
Wait and see approach, 371, 381
Walker, Charls E., 329, 517
Walliser, Jan, 328-329
Wall Street, 337
Walras' law, 259
Warren, Alvin C., Jr.
analysis of, 380-381
realization accounting and, 371-372, 424-426
savings and, 239
Ways and Means Committee, 85, 206, 343
Wealth.
See also Accumulation; Capital
accrued changes and, 8-10
accumulation, 160, 162
ACRS and, 66
bequests and, 78-79
concepts of, 5-6, 438-446
consumption and, 10-12, 508-509
489
490
government lending and, 150
lifetime, 30, 46-47
lump-sum, 165
measurement of, 165-168
model for, 190
national, 485, 487-492
National Balance Sheet and, 452-457
NIPA and, 433-434, 438
tax effect on, 100-101
time-series data for, 448-452
Wedge, tax, 156-157, 190, 209
effective tax rate and, 288-291, 297-300, 303
elimination of, 219
social return and, 216
Â
< previous page page_540
Page 540
Weichenrieder, Alfons J., 370
Weiss, Laurence, 464
Welfare analysis, 24, 27-28, 149-150
Well-off-ness, 45
Wenger, Ekkehard, 329, 331
Whalley, John
490
491
White, William L., 241
Williams, Michael, 514
Willis, Arthur B., 32, 40
Wilson, John F., 515
Windfall profits, 65-66, 78, 81
Wiswesser, Rolf, 329, 331
Workshop on
Economic
Structural
Change, 279
Write-offs, 234-237
Wykoff, Frank C.
effective tax rates and, 282
savings and, 445, 462, 465
X-Tax, 68-70, 82, 328, 340
alternatives, 73-75
bequests and, 71
business and, 67, 74-75
complexity and, 336
distributional effects and, 79
incidence analysis of, 76
Yield, 124, 156
Yield to maturity method, 371
interest and, 394
OID bonds and, 372-373, 398
Yohn, Frederick O., Jr., 461-462, 514-515
491
492
Younger generation. See Age
Zero coupon bonds, 386-387, 417
Zero elasticity, 156
Zero net cash flow, 398-401
Zodrow, George R.
cash flow type tax and, 328, 331
Â
< previous page page_540
492

Ouvr - Taxation-Wealth-And-Saving

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Ouvr - Taxation-Wealth-And-Saving

Uploaded by

Copyright:

Available Formats

1

cover next page >

title : Taxation, Wealth, and Saving

Taxation, Wealth, and Saving

Â© 2000 Massachusetts Institute of Technology

This book was set in Palatino by Asco Typesetters, Hong Kong.

Printed and bound in the United States of America.

LibraryÂ ofÂ CongressÂ Cataloging-in-PublicationÂ Data

Bradford,Â DavidÂ F.,Â 1939-

At the Treasury I found myself

establish their own, and on, and

< previous page page_ix next page >

purposes until the asset was sold,

U.S. Government, Department

In the first section of the chapter I

consumption is the base. Whereas

Background: The Concepts of Income and Consumption

where Y, C, and ∆W stand for

medical treatment constitute

The Relative Difficulty of Implementing Income and Consumption Bases

making those imputations simple enough to permit

Accruing Wealth Changes

Wealth Yielding Direct Consumption Services

Making the System Inflation Proof

< previous page page_14 next page >

Defects of the U.S. Income Tax

Accruing Wealth Changes

Although no one knows how to

Reasons for the Provisions

nent of a system relating tax

Presumably, the nontaxation of

Imputed Consumption Yield

Second-Best Taxation in a Life-Cycle Saving Model

The essence of the problem is

L = labor supplied in period 1

tr = tax rate on interest

tw = tax rate on wages.

where t2, the tax on second-period consumption, is related to tr by

Choosing Between Income and Consumption

the economic system in productive performance.

Utilitarian Analysis of the Choice of Tax Base

Welfare Analysis with Government Saving

in the second-best problem

fullest extent possible. This is

Nonutilitarian Equity Arguments

6. Pechman, ed., Comprehensive Income Taxation.

1978); Meade committee report; and Blueprints.

13. For a detailed discussion, see Blueprints, pp. 42-49.

16. Meade committee report, pp. 129-35.

25. Martin Feldstein,

37. David has shown that for

43. Cobb-Douglas preferences for this case can be written

and a3(1 - tw)wH/(a1 + a2 +a3) on leisure, at price (1 - tw)w.

56. Atkinson and Sandmo, "Taxation of Savings."

57. Edmund S. Phelps, Golden Rules of Economic Growth (Norton, 1966).

58. Atkinson and Sandmo, "Taxation of Savings."

60. See note 43.

62. See note 61.

Atkinson, A. B, and A. Sandmo.

. Lectures on Public Economics New York: McGraw-Hill, 1979.

. "The Meade Report from a U.S.

. "On the Theory of Tax

Meade, J. E. See Institute for Fiscal Studies.