Download as pdf or txt
Download as pdf or txt
You are on page 1of 5

TWO ALTERNATIVES IN CONSUMPTION THEORY

The approaches to consumption theory set out in the last section all started
from the intertemporal maximizationmodel. The assumptions of that model
are not sacrosanct, however. Their empirical validity is open to question and
to econometric verification. The consumption literature incBudes two major
alternatives to the pure intertemporal maximizing model, which we will discuss
here. The hrst alternative modifies the budget constraint of equation (3). It
is generally referred to as a "liquidity-constraint" model. The second introduces
an alternative utilityfunction to equation (2) that recognizes that individuals
care about their relative position in the incomne distribution and their relative
consumption levels, as well as the absolute level of consumption. James Due
senberry developed a theory built on this idea in the 1940s. It can be traced
to Thorstein Veblen's ideas about conspicuous consumption which date from
the beginning of the 1900s. Let's begin with tiquidity constraints.

The Role of Liquidity Constraints


In discussing both the two-peiod and many-period intertemporal maximizing
model, we made heavy use of the intertemporal budget constraint of equa
tion (3 ):
T T

(|+r) (1 + r)'
This budget constraint assumes that the consumer can in fact move
resources
through time in both directions. He or she can save today for
consumption

159
269
CONSUMPTION AND CONSUMER EXPENDITURE

tomorrow, oT can borow today to consume against future income. The second
case is the one illustrated in Figure 12.4, Both saving (lending) and borrowing
are assumed to be possible at the same rate of interest. It is obvious, just from
everyday experience, that this assumption does not hold generally at the level
of the individual bousehotd. Many households cannot borrow frecly for con
sumption against future income. They are constrained in their consumption
decisions by current ligquidity--curnent income plus existing assets. Thus, the
term Biquidityconstraint.
There are two basic causes of liquidity constraints in the form of banks'
unwillingness to lend for consumption against repayment out of future income.
One is the uncertainty that both bank and borrower feel about future income.
The second is the risk of default by the borrower. The combination of these
unçertainties causes banks to place credit limits on borrowers, so that 1hey
cannot borow freely against future income. It may be possible to borTOW
against the purchase of durable goods, such as houses and cars, because they
provide collateral, The bank owns the item until the loan is paid of. Banks
will lend against education because in many cases the government guarantees
repayment of the loan. In this case public policy is attempting to break the
liquidity constraint on investment in education.
Generally, however, these limits apply to howfar you can go in borrowing
toconsume. If the borrower goes beyond a reasonable limit in committing
uncertain future income to repayments, he or she faces the possibility of bank
nptcy. Once faced with that situation, the borrOWer may as well go broke on
a grand scale, enjoying life in the short run. This possibility adds to banks*
reluctance to lend for consumption.
Theextreme case of iquidity constraint would be characterized by a level
of consumption limited by current liquidity. For these consumers, the budget
constraint would be
(35)
where y, is current net income (net of past repayraent obligations). In an
economy with a mixture of consumers, some constrained by liquidity as in
(35)and some under the usual intertemporal constraint of equation (3), the
aggregate consumption function would then resemble that of Ando and Mo
digliani:
C, = doY, + aja,, with (36)
Pervasive liquidity constraints would therefore lead us back to a "Keynesian"
consumption function with an important rote for cuTenTwealth.
The existence of liquidity constraints would not eliminate life-cycle pat
terns in consumption and saving, since the constraint is one-sided. It would
constrain borrowing at the young end of the life-cycle. But saving in the middle
years for retirement woukd stillbe optimal. So the consumption path would

160
Scanned with CamScanner
270 SECTORAL DEMAND FUNCTIONS AND EXTENSIONS OF THE STATIC MODEL

Dollars

Figuro 12-10 Liquidity-constrained ite


cycle.

follow income in Figure 12-10 up to point Awhere c, =y. From there on it


would follow the life-cycle path of Figure 12-6. In a cross section of the pop
ulation, we would expect to find liquidity constraints most binding among the
young. This expectation is bornc out by empirical studies, such as the research
of Hubbard and Judd. It also implies that the effect of cycical fluctuations in
income on consumption is felt most sharply by the younger segments of the
working population.

The Duesenberry Approach: Relative Income


HYPo
The model developed by Duesenberry in 1949 differs considerably from all
the models we have presented in this chapler in that it does not begin with an
entirely private utility function independent of the consumption of other per
sons. Instead, Duesenberry's analysis is based on two relative income hy
potheses.
The first hypothesis is essentially that consumers are not so much con
cerned about their absolute level of consumption as they are with their con
sumption relative to that of the rest of the populatjon. The Fisherian model
is based onthe solution to the problem of consuner choice when the consumer
tries to maximize iu= u(Co, ...,C,, ... , CT) subject to a present value con
straint. In that case, only the absolute level of the individual's consumption
enters the utility function. Duesenberry. however, writes the utility function

U= (37)
R,
where the R's are a weighted average of the rest of the population's consump
tion. This says that utility incrçases only if the individual'sconsumption rises
relative fo that of the average.
This assumption leads to the result that the individual's c/y ratio will
depend on his position /n the income distribution. Aperson with an income
below the average will tend to have a high c/y ratio because, essentially,he is

161 Scanned with CamScanner


CONSUMPTION AND CONSUMER EXPENDITURE 271

trying to keep up with a national average consumption standard with a below.


average ihcome. On the other hand, an individual with an above-average in
come will have alower c/y ratio because it takes asmaller proportion of his
income to buy the standard baskct of consumer goods.
This provides ihe explanation of both the cross-sectional result that
MPC < APC and the tong-run constancy of c/y. I, as income grows along
trend, the relative distribution of income is stable, there willbe no reason for
cly to change. As people earn more along trend, theycan increase their con
sumption proportionatelyto maintain the same ratio between their consump
tion and the national average.
Duescnberry's second hypothesis is that present consumption is not in
fluenced merely by present levels of absolute and relative income, but also by
levels of consumption attained in previous periods. It is much more difficult,
he argues, for a family to reduce a level of consumption once attained than
to reduce the portion of its income saved in any period. This assumption
suggests that the aggregate ratio of saving to income depends on the level of
present income relative to previous peak income, ÷. Mathematically, in Due
senberry's formulation,
= do t (38)

where y is real disposable income. As present income rises relative to its pre
vious peak, s/y increases, and vice versa. We can conveii this Duesenberry
saving function into aconsumption function by observing that ify is disposable
income, c/y = 1-(s/y), so that from (38) we can obtain
C
= (| - ao) - a (39)

As income grows along trend, previous peak income will always be last
year's income, so that y/ù would be equal to 1+ g, where 8, is the growth
rate of real income. If y grows at 3 percent along trend, y/ will be 1.03 and
cly will be constant, as required by the long-nun data of Kuznets.
But as income fluctuates around trend, the c/y ratio will vary inversely
with income, owing to the negative coeficient of y/÷in (39). To compute
the MPC, we can multiply the c/y ratio of (31)by y to obtain

c= (l - ao)y - a,
The MPC, the partial derivative of c with respec to j is then
ôc
|MPC = = (|- ao) - 2a, (40)
dy

162
Scanned with CamScanner
272 SECTORAL DEMAND FUNCTIONS AND EXTENSIONS CF THE STATIC MODEL

Comparison of equation (40) giving the MPC and equation (39) giving the
APC shows that in the short un, with previous pçak incorne fixed, the Due
senberry model implies MPC<APC.
This combination of short-run andlong-run behavior of consumption
gives us the ratchet elect shown in Figure 12-1|. As income grows along trend,
cand y nove up along the long-run function ofFigure 12-11 with a constant
c/y ratio. But if, at some point like Cow'a, income falls off and the econoiny
goes into a recession, c and y move down along a short-run function CoCo with
a slope given by the MPCin cquation (40). Rcçovery of incomeback to its
trend level, which is also the previous peak, will take cand y back up coCo to
the initial CoJo point, where trend growth resumes along the long-runfunction.
II another recession occurs at C,,, consumption and income will fall back
along cC), and rise back to C,' during the recovery. Thus, Duesenberry's
model implies a ratchet effect in that when income falls off. consumption
drops less than it rises as incomc grows along trend. We might note that
this mechanism is formally the same as that suggested by wage rigidity in
Chapter 10.
This completes our survey of the principal theories of the consumption
function. Each theory improves our understanding of the consumption-saving
income relationships. The theories of Ando-Modigliani and Friedman, with
the rational expectations extension by Hall, seem to be more successful than
Duesenberry's, in terms of their present acceptance among economists. The
strength of Friedman's theory isrelated to the acceptance by many economists
of the proposition that people base current consumption-saving decisions on
more than just current and past values of income and assets. The notion,
common to both Ando-Modigliani and Friedman, of a basic permanent con

Long-un
function

Flguro 12-11 The "ratchet effect" in consump


tion.

163 Scanned with CamScanner

You might also like