The Evaluation of National Income in An Imperfect Economy (A J) 1969

You might also like

Download as pdf or txt
Download as pdf or txt
You are on page 1of 14

Pakistan Institute of Development Economics, Islamabad

The Evaluation of National Income in an Imperfect Economy


Author(s): James A. Mirrlees
Source: The Pakistan Development Review, Vol. 9, No. 1 (Spring 1969), pp. 1-13
Published by: Pakistan Institute of Development Economics, Islamabad
Stable URL: http://www.jstor.org/stable/41257992 .
Accessed: 05/09/2014 12:55

Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at .
http://www.jstor.org/page/info/about/policies/terms.jsp

.
JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of
content in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms
of scholarship. For more information about JSTOR, please contact support@jstor.org.

Pakistan Institute of Development Economics, Islamabad is collaborating with JSTOR to digitize, preserve and
extend access to The Pakistan Development Review.

http://www.jstor.org

This content downloaded from 85.244.145.75 on Fri, 5 Sep 2014 12:55:20 PM


All use subject to JSTOR Terms and Conditions
The EvaluationofNationalIncomein an
ImperfectEconomy
by
JamesA. Mirrlees*

In thispaper,I shall discusshow nationalincomeshould be measuredin


an 'imperfect' economy,wherefeasiblepolicyinstruments such as taxes,tariffs,
quotas, and controls
quantitative do not operatein a lumpsummanner,and may
be farfromtheiroptimumlevel. In particular,I wantto examinethemeaning
of nationalincome,and its parts,when they are measuredin termsof 'world
prices'; or, morepreciselyand moregenerally,in termsof theaccountingprices
thatmightbe computedas a guide to particularproductiondecisions,in the
publicsectorand elsewhere. I have foundthisset of questionsconfusing,and
it may be thatothersdo too, and will welcomean attemptat clarification.

We would like to be able to use nationalincomestatisticsfora variety


of purposes. Ideally, we should arrangethat changes in aggregatenational
incomereflectchangesin social welfare;that changesin the incomegenerated
in each sectorreflectchangesin its contributionto welfare;and that changes
in theincomeof a group,suchas wage-earners or agriculturalists,
reflectchanges
in the welfareof that group. Naturally, none of these ends can be achieved
withoutfirstspecifyinga standardof group and social welfare,and people dis-
agree about such standards. But it is worthemphasizingthat changesin na-
tionalincomeas usuallymeasuredneed not,and usuallydo not,reflectchanges
in welfare. This is not to say thatbettermeasuresof nationalincomeand its
componentscannotbe devised. CertaingeneralprincipleswhichI take to be
of considerableimportancecan be stated: thereis a standardtheoryof how
nationalincome should be measuredat any rate for a 'perfect'economy1. I

ofEconomicsin theUniversity
♦Theauthoris Professor ofOxford. He wishesto thank
Dr. A. R. Khan forusefulcomments.The subjectof the paperwas suggested bya seminar
givenby Mr. M.FG. Scottin Oxford. The researchwas carriedout at thePakistanInstitute
of Development Economics.
iAmongthemore important papersare: [2 ; 4 ; 6]. ProfessorSamuelson'spaper [6]
butnotintheformsketchedhere.
accountofthetheory,
givesa definitive

This content downloaded from 85.244.145.75 on Fri, 5 Sep 2014 12:55:20 PM


All use subject to JSTOR Terms and Conditions
2 ThePakistanDevelopment
Review

shall beginby outliningthistheory. By a 'perfect9


economy,I shall mean an
economywhichis normallyin competitiveequilibrium,and in whichproperty
that no conceivabletransfers
is so distributed of propertywould increasewel-
fare.
It is assumed that welfareis assessed in a way that respectsindividual
preferences so thata moveto a positionpreferred by one individual,whileothers
remain in the same position,implies an increase in welfare. Then we can write
social welfare as a function of the 'utilities'
of individual.

W = W(ui,u2, . . . ,uH) (1)


whereu1is theutilityof the i-thindividual. The consumptionof thei-thindivi-
dual is denoted by x1= (x* , . . . , x^), so that

ui = ui(x0 = u*(x',...,xl) (2)

We considerwhathappensto welfareas theconsumption bundlesofthedifferent


individualschange. It is convenientto imaginethatconsumptionis changing
continuouslythroughtime. Then the utilityof an individualis changingat a
rate

4 u- gw (3)
whereUjis thederivativeof u withrespectto xj, themarginalutilityof consum-
ing theyth commodity, and xj is the rate of changeof consumptionof theyth
commodity. The assumption of competitiveequilibriummeans that marginal
utilitiesare proportionalto prices. Therefore

= mLqjXj (4)
-^-u

whereq = (qi, . . • , qn) are the prices,and the factorof proportionality,m, is


of
the 'marginalutility income', and is positive. So, for the individual,it is
in
naturalto measurethe change utilityby Zqjxj; that is> bY the change in
2qjXj at constantprices.
We can now deduce the changein social welfare:

J*w = S Wi-1- u» (5)


at i-i at

whereWj is thederivativeof W withrespectto ul. From(4) and (5) we obtain:

= Ljqj LiWrnnxj (6)

This content downloaded from 85.244.145.75 on Fri, 5 Sep 2014 12:55:20 PM


All use subject to JSTOR Terms and Conditions
Mirrlees:NationalIncomein Imperfect
Economy 3

The assumptionthatpropertyis perfectly distributed


impliesthat the marginal
contributionof each man's income to social welfareis the same. That is,
Vf1ml= Wam2= . . . = WHmH= M, say. (I shall call Wjini the 'marginal
welfareof the i-thindividual'sincome'.) Then

= MSjqi 2ix| (7)


4"W

Since ZjXj = xj is the total supplyof the yth commodity,(7) demonstrates


that changes in Sqjxj at constantprices may be taken to reflectchanges
in social welfare2.

I inclineto believethattheassumptionof perfectproperty distributionis,


quantitatively,the weakest link in the standard argument, which I have just
outlined. Whenthemeasureof incomeis restricted to particularincomegroups
in the economy,who mightbe supposed to have sufficiently similarWjini,it
in
seemsquite sensibleto measurethe change group welfare by the change in
groupincome,measuredat constantprices. But, fromthepointof viewof the
theory,actual figuresof nationalincomedo not seemto have muchsignificance.
II

It is universallyrecognizedthat taxes or subsidies may be imposed


on certaincommoditieswherethe individual'sfreechoice is inconsistentwith
themaximization of social welfare,and thatit maythenbe rightto use theprice,
net of tax, to evaluate national income. But I am not aware of a published
proofof thislatterproposition. In any case, one would want to regardmost
commoditytaxes, income taxes, and the like, as interferencewith the 'free
market' requiredto financepublic expenditure,and intendedto improvethe
distribution of income. These taxes existbecause desirablelumpsumtaxes are
impossible(or at least inordinatelyexpensive). Not only are actual taxes far
frombeing lumpsumtransfersas requiredif competitiveequilibriumis to be
optimum:theyare not chosen to maximizelong-runsocial welfare,although
welfareconsiderationsundoubtedlyinfluencethem.
In such an economy,consumerspurchasegoods and servicesat market
prices. At firstsight,it may stillseem reasonableto evaluatenationalincome
in termsof marketprices,so long as we are primarilyinterestedin assessing
changesin real consumption. For any particularindividualwho takes reason-
able care in allocatinghis expenditure,
it is approximately truethathe would be
betteroff in an alternative situation
(verysimilar) ifand onlyifhis consumption,
measuredat constantmarketprices,were higher in that situation. Thus, it is

2Whenthechangeof welfareduringsome intervalof timeis to be measured,it is,of


an indexof diyisiatypein whichchangesfromeach yearto the
course,desirableto construct
nextare weighted ruling,ratherthanthepricesthatruledin somebase
bythepricescurrently
year.

This content downloaded from 85.244.145.75 on Fri, 5 Sep 2014 12:55:20 PM


All use subject to JSTOR Terms and Conditions
4 ThePakistanDevelopment
Review

tempting to assertthatan increasein nationalincome(at constantmarketprices)


means that,in some sense,people are on averagebetteroff. But thisis quite
misleading. We could deduce from such an increasethat someoneis better
off;but that is a conclusionas weak as to be almost totallywithoutinterest.
We cannotdeduce thatit is feasibleto make everyonebetteroff;fortheremay
be no systemof feasibletaxesand subsidiesthatwillallocatetheavailable pro-
ductionas to increaseeveryone'sutility*.The formalargumentstatedin Section
I made it clear thatnationalincomeat marketpricesis a satisfactory measure
of increasedwelfareonlyif a small changein the incomeof any individualhas
thesame effect on totalwelfareas an equal changein theincomeof anyoneelse:
that is, if marginalwelfaresare equal.

That thereare nevertheless 'prices'in termsof whichit would be sensible


to measure national income may be made plausible as follows. Granted a
definitionof social welfare,a smallincreasein theavailabilityof one commodity
shouldmakepossiblean increasein social welfare,providedsuitabletax changes
are made. If we assumethatbeforeand afterthechange,tax and otherpolicies
are such as to bringabout the maximumfeasiblewelfare,the changein social
welfarearisingfroman increasein the availabilityof a commodityby one unit
may be takento be its 'accountingprice',measuringits trueworthto the eco-
nomy*. If one knew the accountingpricethatwould be impliedin the opti-
mum feasiblestate of the economy,one would want productionto maximize
profitsin termsof theseaccountingprices. For an increasein profitsin one
line of productionwould implyan increasein nationalincome5,and therefore,
sinceforthe momenteverything is measuredin termsof accountingprices,an
increasein social welfare.

This simpleargumentsuggeststhatsuitablepricesforevaluatingnational
incomecan, in principle,be defined,and shows that the accountingpricesso
definedare the pricesone would want for decentralizing
productiondecisions.
But two serious worriescry out for comment. First, the definitionof the
accountingpriceswas made on the assumptionthat social welfarewas being
maximized. Second, it is far fromclear how such accountingprices could
possibly be estimated. Nevertheless,it has been clearly demonstratedthat

3Cf. [6, pp. 18-19]Samuelsonuses the term 'feasible*to describe institutionally


possiblearrangements of the economy.
The argument is notwatertight,fortheremaybe no choiceof taxesthatwillcreatean
additionaldemandfor the specifiedcommodity, whileleaving otherdemandsunchanged;
althoughwelfare couldbe increasedifat thesametimetheavailability
ofsomeothercommodity
was slightlyreduced.A rigorousproofof theexistenceof accounting prices(or ideal'producer
prices')undergeneralconditionsis givenin [1].
*Givenfactorinputs. The argument is presentedin a formthatappearsto assume
thatfactorsuppliesaregiven,butthisis merely an expositoryconvenience.If thereis a change
in,say,thelabourservicessuppliedbyhouseholds, thevalueof thischangeat constantprices
shouldbe subtractedfromthevalue of thechangein productionin orderto estimatethe
changein socialwelfare.

This content downloaded from 85.244.145.75 on Fri, 5 Sep 2014 12:55:20 PM


All use subject to JSTOR Terms and Conditions
Mirrlees:NationalIncomein Imperfect
Economy 5

nationalincomein termsofmarketpriceshas no specialclaimon theeconomist's


attention.

I shalldeal withthetwo worriesin reverseorder. First,letit be assumed


that the governmentis indeed makingthe most of the policy instruments
availableto it. It is readilyseen that,in some cases, the accountingpricesmay
be quite obvious. The most importantcase is that in whichall commodities
can be traded with foreigncountriesat fixedprices,the same whetherexported
or imported. Then thetradeprices,or 'world prices',mustbe the accounting
prices. Commodity1 can be substitutedforcommodity2 in the ratiopt : pa
(wherepl5 p2, . . . are theworldprices). Therefore,the policies of govern-
mentmusthave continuedthis substitutionup to a point wherethe marginal
gainsto welfarefromthetwo commoditiesare in the same ratiopx : pa.

The situationmay be portrayedsimplyin a diagram(Figure 1). Two


commoditiesare produced. PP is the productionfrontierof the economy(in
the absence of trade). The curvesWx, W2, W3 are equal-welfarecontours,
connectingproductionplans that give rise,withoptimumgovernment policies,
to equal welfare. It will be noted that these contourswill not in generalbe
convex;buttheydo not, in general,intersectone another. On the reasonable
assumptionthat more of eithercommoditymakes possible an increasein wel-
fare,it can be seen thatX is the optimumpoint in the absence of trade. The
line TT has slope - Pi/pa: it representsthe additionalopportunities provided
by trade. Then Y is the optimum point once trade is possible. The tangent
to thewelfarecontourat Y givestheaccountingprices,by our earlierargument.

It will be seen that the above argumentrestson thesimpleobservation


that the welfarecontours(or community indifference curves) of the usual
analysis can perfectlywell be replaced by feasiblewelfarecontoursto take
accountof the impossibility of the lumpsumtransfersassumedin thatanalysis.
A rigorousjustificationof thisprocedureis rathertroublesome,however[1].

It is nevertruethat all commoditiescan be tradedat constanttermsof


trade. Some nontradedcommodities^may be producedunderconstantreturns
to scale out of tradedcommodities,however. Indeed, it is knownfromthe
non-substitution theoremthat,so long as thereis only one nonproducedinput
(labour), constant returnsto scale everywhere, and no capital prices can, in
general, be determined uniquely. The presence of capital equipmentcauses
in
trouble: order to include capital costs in the costs of nontradedgoods (as
one must do in the case of nontraded goods) one must know the (accounting)

*Andcommoditiessold to a marketwhosedemandis relatively inelastic,forin these


casesa country
bentuponmaximizing itsown welfareevenat theexpenseof others,willhave
in termsof worldprices)in orderto know
price(or costof production
to knowtheaccounting
howmuchto export.

This content downloaded from 85.244.145.75 on Fri, 5 Sep 2014 12:55:20 PM


All use subject to JSTOR Terms and Conditions
6 Review
ThePakistanDevelopment

Commodity1

Figure 1

This content downloaded from 85.244.145.75 on Fri, 5 Sep 2014 12:55:20 PM


All use subject to JSTOR Terms and Conditions
Mirrlees:NationalIncomein Imperfect
Economy 7

rate of interest,and the rate at whichthe (accounting)prices of capital goods


are changing. If the countryhad access to a perfectcapital market,the ac-
countingrateof interestcould be determinedfromthe worldprice. But other-
wise, it will depend on the particularsocial welfarefunction. Sometimes,ac-
countingpricesmay not be verysensitiveto the particular rate of interest
chosen, of course. Attemptshave been made to estimateparts of national
incomein accountingprices?: the task does not seem impossiblydifficultal-
thoughalternativeprices,dependingon alternativewelfarespecifications, may
have to be calculated.

in
I shall now outlinethe more formalderivationof the accountingprices,
and outlinetherelationbetweenmarketpricesand accountingpricesin an opti-
mallycontrolledeconomyof the type we have been considering. It must be
assumed,and it is not unreasonableto do so, thatthegovernment can, by means
of commoditytaxes and subsidies,arrangefor marketpricesto be anythingit
pleases. As before,marketprices are denoted by q = (q2 , . . . , qn). I
denoteby V (q) = V (qx , . . ., qn) thewelfarethatarisesfromtheoptimumuse
of othergovernment policies,giventhe marketprices. When thesepoliciesare
followed, the aggregatedemands by consumers,D = (Dx , . . . , Dn) are func-
tions of q. The argumentsof the previous section suggested,and is rigor-
ously demonstratedby Diamond and Mirrlees [1], that there are account-
ing prices p = (pl9 . . . , pn) such that optimumproductionx* = (xf,. . . , x*)
maximizes
PX = piXi + p2X2+ . . . + PnXn.

If, when commoditytaxes are optimallychosen, marketpricesare

q* = (qj,...,qp,

optimumproductionis

x* = D(q«). (8)
It is shownby Diamond and Mirrlees[1] thatthepricesp can be chosenso that
"M~k2PjDj(q<)(9)
-4V(q<)
fora constnatM» . It is also shownthat(9) can be interpreted
as sayingthat

7Cf. [3 ; 7]. Methods for obtainingcrude estimatesof accounting prices are


discussedin [5].
sCertainassumptionsnot mentioned hereare required. Theyare rathertechnical,
and
not at all restrictive.

This content downloaded from 85.244.145.75 on Fri, 5 Sep 2014 12:55:20 PM


All use subject to JSTOR Terms and Conditions
8 Review
ThePakistanDevelopment
themarginal on welfare
effects taxrateshouldbe pro-
in different
ofincreases
to themarginal
portional in government
increases revenuearisingfromthetax
changes.

Supposenationalincomeis changing.The changewill be associated


withchangein marketprices. We have:

=MSjPj
1 v=2>-kv(q')qr (I0)
(Eik Dj(q*)qr)
From(8), we seethat

Therefore,

-Ay = MSjPjx/,

as we expect.
Relations(9) definethe optimumrelationbetweenmarketpricesand
accounting canholdevenifincome
prices. It willbe noticedthattheserelations
taxes,etc.,are notbeing chosen optimally. The conclusion(10) holdsevenif
someofthemarket prices arefixedand constantfor somereason,notdeliberately
chosenby government9. Thus,the conditions necessaryfor the conclusion
thatchangesin nationalincomeat constantaccounting pricesreflect
changes
in welfareare that:
taxesare beingchosenoptimally;
a) some commodity
b) all policiesnot chosenoptimally,and all marketpricesnot deter-
minedoptimally, remainconstant.
Ofcoursein anycomparison ofnationalincomeovertime,ithasto be assumed
thatpreferences and thecircumstances ofdemandremainconstant:thiscondi-
- e.g.incometaxrates
themorecircumstances
tionis lesslikelyto be satisfied
- haveto be assumedconstant, determined.
ifnotoptimally
IV

so far impliesonlythatundercertainratherUtopian
Our argument

9lfa tax rateon somecommodity remainsconstant,theproducerpricemaychange,


thataccounting-
and as a resultthemarketpricewillchange. In thiscase thereis no guarantee
pricenationalincomeis a good measure. But in somecases,e.g., whenthereis a fixedtariff
on a commodity importedat a constantprice,theconditionis satisfied.

This content downloaded from 85.244.145.75 on Fri, 5 Sep 2014 12:55:20 PM


All use subject to JSTOR Terms and Conditions
Mirrlees:NationalIncomein Imperfect
Economy 9

circumstancesnamely,that the governmentuses whatevertools are at its dis-


posal withinfinitewisdom- is it possible,in principle,to use accountingprices
to evaluatechangesin welfare. Admittedlywe can assume, alternatively, that
policies not chosen optimallyremainconstant,but this is in practicean un-
reasonable assumption. Governmentsmay have more or less good or admir-
able attentions; but theyvaryparticulardetails of the tax structureonly occa-
sionally,and whenchangesare made, theyare oftendictatedby considerations
of immediateexpediency. Policies are certainlynot (yet)calculatedto obey the
rulesimpliedby a particularsocial welfarefunction. It seemsthat,ifa measure
of thechangesin social welfareis wanted,it will have to be constructeddirectly
frominformation about the changesin consumptionof different incomegroups.

It does seem to me that national income measuredin termsof certain


accountingprices may neverthelessbe givenan importantinterpretation.We
may attemptto estimateaccountingpricesin such a way thata small changein
income at thesepricesshows the change in maximumattainablesocial welfare,
even whenthe government is not pursuingpolicies that allow social welfareto
be maximized(given the instruments at its disposal). For this purpose,it is
necessaryto estimatewhat the accountingpriceswould be if available produc-
tion capacityin the economywere being used in an optimalmanner,so as to
attainthe optimumattainablestate of the economy. In theearlierdiscussion,
the methodsby whichit was thoughtaccountingpricemightbe estimatedwere
perfectly appropriateto thisend. Many of thepricesare determined by trading
possibilities;othersdepend on a more subtle analysis,perhaps by means of
optimizingmodels, in which welfareconsiderationsmay appear directly. At
firstsight,it is hard to see even how accountingpricescan be definedin a non-
optimalsituation(althoughI thinkthat is not necessarilyso hard); but in fact
therewould seem to be some advantagein attempting to estimatethe 'optimal'
accountingprices.

Once this is done, we may imaginethat we are decomposingthe actual


changein social welfareintotwo parts; thechangethatresultsfroma changein
attainablewelfare;and the change that is attributableto changesin the extent
to whichactual taxes,tariffs, quotas and controlsreduce social welfarebelow
the maximum. A new textilemill should bringabout an increasein social wel-
fare; it will- or might- do so by increasingthe value of the economy'spro-
ductionin world-priceterms;or it mightdo so by amelioratingthe adverse
effects of a non-optimaltax on importsof textiles. It does not seemreasonable
to regardthis second effectas an increasein the economy'sproduction,since
theincreasein welfarecould in facthave been obtainedwithoutthe new textile
millhavingbeen set up. The factthatthe desirabletariffchangehas not been
made will usuallybe the resultof inadvertanceor sectionalpressures. If the
adverseeffectshave been lessened,that is good, but in describingthe cause of

This content downloaded from 85.244.145.75 on Fri, 5 Sep 2014 12:55:20 PM


All use subject to JSTOR Terms and Conditions
10 Review
ThePakistanDevelopment

the economy'sdevelopment,it is as well to separatethiseffectfromthe straight-


forwardbenefitof havingextra outputavailable, as a resultof increasingthe
economy'sproductivecapacity.

There can be no unambiguouslyrightway of evaluatingthe national


income,or thecontribution of particularsectorsto thenationalincome. When
an investmentproject is evaluated, it may be necessaryto make alternative
recommendations: it would be best,say,to removean importquota on fertilizers
(whiletemporarily restricting
aggregateinvestment); but ifthatwillnot be done,
factory. In such a case, theecono-
it would be a good idea to set up a fertilizer
mist should also point out the cost of adopting the less satisfactoryplan.
Similarly,one cannot measurethe contributionof a particularsector to the
nationalincomeunlessone knowswhatwould have remainedthesame - parti-
cularlytaxes,rations,quotas and the like - if activityin that sectorhad been
different. It seemsbest not to assume that circumstancesare fixedand un-
changeableif the government could perfectly well changethem. But it is hard
to draw a clear line betweenwhat the government can do and what it cannot.
The suggestionthatnationalincomestatisticians at least assumethatcommodity
taxes could have been different has the advantage,apart fromthe plausibility
of the presumption,that the implied accountingprices can perhaps be esti-
mated!0.

that,in an imperfect
It is unfortunate economy,thepricesappropriateto
measuring the national income are not actual prices,and thereforecannot be
observed directly. They are surelynot, in any presentlyexistingeconomy,
marketprices,for reasons we have already seen. They are not, in general,
factorprices. Factor priceswould indeed be the appropriatepricesto use if
controlof the marketsforgoods and serviceswereexercisedentirelyby means
of uniformtaxes. But thisrequires,in particular,thattariffs and exporttaxes
different from 'optimum' taxes and tariffsare not levied. Revenue or protec-
of
tivetaxation foreign trade is ruled out. Undeniably, it is awkward to want
an economic statisticsthat is not definedby standard proceduresapplied to
observabledata: the 'objectivity'of the statisticsappears to be compromised.
But thereis no way out. Eitherone has a (relatively)meaninglessfigure,or
one employseconomic expertiseand guessworkin the constructionof the
statistics. We should not have expectedanythingelse.

The argumentsof thispaper suggestthatthe situationand development

lODr.A. R. Khan and I are currently


engagedin attempts
to estimateaccountingprices
forthe economyof Pakistan,and to computenationalincomeand its sectoraland regional
components at accounting
prices.

This content downloaded from 85.244.145.75 on Fri, 5 Sep 2014 12:55:20 PM


All use subject to JSTOR Terms and Conditions
Mirrlees:NationalIncomein Imperfect
Economy 11

of an economymightbe ratherinformatively
portrayedby means of thefollow-
:
ing schemeof estimates

Income Income Indirect Income Income Income


(accounting (market taxes-f group group group
prices) prices) profit I II III
J taxes
Sectorsof
theeconomy

Total

Net income: Tax -f -f = (market


prices)
Consump- -I- =
FromabroadGNP : tion:
Net borrowing
from
abroad: „ (accounting
-f + = prices)

Public consumption =
Grosscapitalformation =
= GrossNationalExpenditure =
(accountingprices)

The differentincome groups listedin the table mightbe wage-earners,profit-


earners,and self-employedpeople, or (ideally) groups classifiedby size of in-
come. It mustbe admittedthat it is bound to be hard to estimatethe details
of thesecolumns,interesting
thoughtheyare.
Certainpointsmustbe made about the interpretation of such a formof
national accounts. First,it should be emphasizedthat comparisonbetweena
particularcomponentmeasured in marketprices and the same component
measuredin accountingprices tells one nothing. Accountingpricescould all
be doubled,and would stillbe perfectly correctaccountingprices. The theory
only claims significance
for changesin items,measuredat constantprices,and
small ones at that. It is perhaps not extremelymisleadingto compare,say,
the contributionsof two sectors,each measuredby means of accountingprices,
and to remarkupon therelativeimportanceof thesesectors. But a comparison
betweenitemsmeasuredin different priceshas no meaning. In particular,one
cannotestimate,in an absolutesense,the subsidythata sector'really'receives,
the differencebetweenits actual incomeand its accountingprice. A doubling
of all accountingprices would change all apparent taxes and subsidies,but
change none of the correctdeductionsfromthe figures. One can, of course,
estimatethe differencebetweenthe contributionof a sectorto production on

This content downloaded from 85.244.145.75 on Fri, 5 Sep 2014 12:55:20 PM


All use subject to JSTOR Terms and Conditions
12 ThePakistanDevelopment
Review

the one hand, and the value, in accountingprices, of the consumption(and


possibly,asset-acquisition)of those who receivethe incomesof the sector:this
mightbe interpretedas thetruetax or subsidyon the sector,but is veryhard to
estimate.

Secondly,publicconsumptionand capitalformationshouldbe measured


in accountingprices too. This is reallya matterof convention:theirvalue,
comparedto the value of consumption,mightwell be substantially greaterthan
at
(or, any rate, different
from) the value suggestedby the in
figures the table.
For thatvalue derivesfromthe finalvalue of the resultsof theseexpenditures
forhouseholds,and on thatno directevidencehas been broughtto bear. Only
if thegovernment's expenditure policiesand the level of savingin the economy
are optimalcan changesin the value of theseitemsbe taken to be measuresof
social welfarechange comparableto changes in consumption. But it is con-
venientto use the same prices;and it is an advantagethat,in principle,a small
reductionin consumption,measuredat accountingprices,would make possible
an equal increasein, say, investment;and vice versa.

Finally,one mightsupplementthe pictureof the economy'sdevelopment


givenby the table for a numberof years,by computingrough estimatesof
changes in social welfarefromyear to year. To do this, one must weight
changesin the consumptionof differentincomegroupsby estimateof the mar-
ginal welfaresof average members of the groups. Weightsinverselypropor-
tionalto averageincomeper head (aftertax), or to thesquare of averageincome
per head, mightbe appropriate. It must be recognized,however,that the
empiricalbasis forthe componentsof such an averageis likelyto be veryweak
at present,quite apartfromthe supposedlyembarrassing precisionof thevalue
judgmentemployed.

VI

The method of evaluatingnational income explored in the paper, if


satisfactory,is so partlybecause accountingpricesin manyeconomiesmay not
be verysensitiveto theparticularassessmentof social welfare. A directevalua-
tionof changesin social welfare,in termsof theconsumptionof differentincome
groups, is hard to do, and on
may depend quite sensitively the of
sepcification
of nationalincome
the welfarefunction. In any case, the practicalsignificance
changes evaluated by means of accountingprices is considerable;since, in a
naturalsense,thesechangesmeasurethe 'achievements'of the economy,con-
sideredas a mechanismforproduction. If growthratesare to be used forgrad-
ing governments, I should like to see themmeasuredat accountingprices.

This content downloaded from 85.244.145.75 on Fri, 5 Sep 2014 12:55:20 PM


All use subject to JSTOR Terms and Conditions
IncomeinImperfect
Mirrlees:National Economy 13

REFERENCES
1. Taxationand PublicPro-
Diamond,P.A. and J.A. Mirrlees,Optimum
duction.WorkingPaperNo. 21. (Cambridge;M.I.T., Departmentof
Economics,
May 1968).
2. 1940.
Hicks,J.R., "The ValuationofSocialIncome",Economica,
3. Lewis,S. R. andGuisinger,
Measuring ina Developing
Protection Country:
The Case of Pakistan. Memorandum No. 20. (HarvardUniversity
Research,December1966).
Projectfor Quantitative
4. Little,I.M.D., 'The Valuationof Social Income",Economica,1949.
5. Manualof Industrial
Little,I.M.D. and J.A.Mirrlees, ProjectAnalysis.
(Paris: Organisationfor Economic Cooperationand Development,
forthcoming).
6. Samuelson,P.A., "ValuationofReal NationalIncome",OxfordEcono-
micPapers,1950.
7. Soligo,R. andJ.J.Stern,"Tariff and In-
ImportSubstitution
Protection,
vestment Pakistan
Efficiency", DevelopmentReview,Vol. V, No. 2,
Summer 1965.

This content downloaded from 85.244.145.75 on Fri, 5 Sep 2014 12:55:20 PM


All use subject to JSTOR Terms and Conditions

You might also like