Welfare Aspects of Benefit-Cost Analysis

Author(s): John V. Krutilla

Source: Journal of Political Economy, Vol. 69, No. 3 (Jun., 1961), pp. 226-235
Published by: The University of Chicago Press
Stable URL: http://www.jstor.org/stable/1829264 .
Accessed: 22/05/2014 11:23

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.

The University of Chicago Press is collaborating with JSTOR to digitize, preserve and extend access to Journal
of Political Economy.

http://www.jstor.org

This content downloaded from 131.175.12.86 on Thu, 22 May 2014 11:23:52 AM

All use subject to JSTOR Terms and Conditions
 WELFARE ASPECTS OF BENEFIT-COST ANALYSIS'
JOHN V. KRUTILLA
Resourcesforthe Future,Inc.
I. INTRODUCTION
IN RECENT years therehas beensub-
stantial interestin developing deci-
sion rules for public expenditures
undera varietyof conditions.To a large
extentin the literaturedealingwithgov-
ernmental expendituresin the United
States the interesthas been confinedto
thefieldofresourcedevelopmentand the
activityknownin generalas benefit-cost
analysis. Benefit-costanalysis can be
characterizedas the collection and or-
ganizationof data relevantby some con-
ceptually meaningfulcriteria to deter-
miningtherelativepreferredness ofalter-
natives (24, Parts II, III). As is typical
of much of economicanalysis,the objec-
tive is to attemptby analysis to indicate
how a particular desideratum can be
maximized-accomplished by comparing
the differences in the relevantcosts and
benefits associated with alternatives
among which choices are to be made.
This activity,ofcourse,does not differin
kind from the economic analysis em-
ployed in reachingdecisionswithrespect
to production or other policies of the
firm.Nevertheless,whilethe analyticac-
tivity does not differin nature,the de-
sideratum and the choice variables on
whichit dependswilldiffer. Essentially,a
1 Acknowledgments
are due to Robert Dorfman,
Otto Eckstein,Francesco Forte, GeorgeHall, Orris
Herfindahl,and Vernon Ruttan for helpfulcom-
ments on an earlier draftof this paper. The sub-
stance of thispaper was presentedas one in a series
of lecturesgivenin the trainingprogramsponsored
by the United Nations Economic Commissionfor
Latin Americaat the Universityof Mexico in the
summer of 1960.
This content downloaded from 131.175.12.86 on Thu, 22 May 2014 11:23:52 AM
All use subject to JSTOR Terms and Conditions
private cost-gains calculus is employed
in deciding private firms'policies; ex-
ternalities and other divergences be-
tweenprivateand social productare neg-
lected. Benefit-costanalysis,on the other
hand, seeks to take account of such di-
vergences as a basis for guiding public
action eitherwhen marketprices do not
accurately reflectsocial value or when,
by virtueof the indivisiblenature of col-
lective goods, no market exists from
which to observe directlyobjective evi-
dence of the community'svaluation of
the social marginal product. Speaking
loosely, while the decision rules of the
theoryof the firmaim at profitmaximi-
zation, the decision rules of benefit-cost
analysis seek to maximize "public bene-
fits"or "generalwelfare"withinthe area
of responsibility(29, p. 3).
In this connectionthe normativena-
ture of the analysis needs to be empha-
sized. Unlike the assumption of profit
maximization,which is a descriptivehy-
pothesis having explanatoryvalue (but
cf. 28, 38), benefit-costanalysis is in-
tendedto be prescriptive.Underlyingthe
analysis is the value judgment that, if
governmentalinterventionis justifiedin
part2by virtueof the market'sfailureto
achieve an efficientallocation of re-
sources (1, 3, 16, 22, 23, 26, 34), public
officialsought to apply decision rules
which tend to improve the allocation,
that is, to improve the generalwelfare
2 That part ofgovernmental interventioncharac-
terizedby Musgrave's "allocation branch" (26).
226
 WELFARE ASPECTS OF BENEFIT-COST ANALYSIS 227

ratherthan theirpersonalor speciallyin- matic terms the welfareimplicationsof

terestedclientele'swelfare.3 supramarketallocation of resourcessug-
gested by benefit-costcomparisons.
II. INITIAL CONDITIONS AND SIDE
EFFECTS III. IMPLICATION OF REDISTRIBUTIVE
EFFECTS FOP, THE MEASUREMENT
While interventionis requiredto cor-
OF WELFARE
rectdivergencesbetweenprivate and so-
cial product and cost, both the initial Consideringcondition(a), we may rec-
conditionsand the associated side effects ognize that the practical possibility of
of intervention are ofrelevancein assess- multipartpricingor correspondingspe-
ingthe welfareimplicationsofsupramar- cial assessmentsis limited; and thus we
ket allocationsof resources.Public inter- must retreat to weaker positions. This
vention to redirectthe use of resources retreat will require, at best, that the
involves costs. Assumingthat the gross analyst rely on interpersonalcompari-
benefitachieved exceeds associated op- sons of utilitybeforehe can hazard any
portunitycosts,if,in addition: (a) oppor- judgmentwith respectto welfare,plus a
tunitycosts are borne by beneficiariesin degree of faith that the redistributional
such wise as to retain the initial income consequences for the measurement of
distribution,(b) the initial income dis- welfare are of the second order of sig-
tributionis in some sense "best," and nificance,or (at worst) that the analyst
(c) the marginalsocial rates of transfor- abstain fromsaying anythingabout the
mation between any two commodities magnitude or even the directionof the
are everywhereequal to theircorrespond- change in eitherwealth or welfare,that
ing rates of substitutionexcept for the is, about the change in the size of the na-
area(s) justifying the intervention in tional real income or its welfareimplica-
question, then welfarecan be improved tions.
by such intervention.And, to the extent To claim that welfarehas increased,
that the objectiveis pursuedto the point whenthe ex antedistributionis not auto-
wherethe social marginalrates of trans- maticallypreservedby the mechanismof
formationbetween commoditiesin this intervention,requires that those who
area and othersectorsare likewiseequal gain are able to and do compensatethose
to the rates of substitution for corre- who lose, and still have somethingre-
spondinglypaired commodities,welfare maining. However, if we are uncertain
is maximized. about the "goodness" of the originaldis-
However, since condition (a) is only tribution of income (condition b), we
partiallyfeasiblein the majorityof cases cannot contend that failureto compen-
dealt with by benefit-costanalysis, the sate would not resultin a greatergain in
likelihoodof condition(b) has been sub- welfare.Such failuremeans onlythat the
ject to considerablequestion, and in the issue cannot be resolvedwithoutmaking
world of realitycondition(c) is improb- interpersonalcomparisons(31).
able, consideration of these stringent If we are contentto attach no greater
conditionsis necessaryto assess in prag- normative significanceto the result of
"maximizing" decision rules, however,
IFor a positive theory of political behavior, than to aver that the national incomehas
which may more accurately mirroractual rather
than "desirable" behavior of public officials,see been increased,taking
the distributionas
Downs (4). given,we can envisagea broaderapplica-

This content downloaded from 131.175.12.86 on Thu, 22 May 2014 11:23:52 AM

All use subject to JSTOR Terms and Conditions
228 JOHN V. KRUTILLA
tionofbenefit-cost decisionrules.We are formeasuringthe changein real income.
then contentto accept the "production- Consideration of only the latter and
distribution"or the "efficiency-ethics" lesser question will sufficefor our im-
dichotomyof the Kaldor-Hicks-Scitov- mediate purpose. Scitovsky's critical
sky line of development (9, 10, 15, 36, point can be summarizedsomewhat as
37). We say that, ifthose who benefitby follows:an indicatednet increasein real
virtue of the increase in productioncan incomewhen valued in termsof prevail-
overcompensatethose who sufferlosses ingpricesmay not prove so when valued
(but do not actually make the compen- in termsofpricesreflecting the attendant
satingpayments),the "aggregatereal in- incomeredistribution. Admittedly,fora
come" has been increasedirrespectiveof structuralreorganizationof the magni-
its distributionand, accordingly,of its tude implicit in the repeal of the corn
welfareimplications.4 laws, the effectof the income redistribu-
Kaldor's production-distributiondi- tion on the constellation of relative
chotomyand the resultingtest of an in- prices cannot be ignored. On the other
crease in real incomeappear supportable hand, the relative magnitudeof the re-
for the more or less marginal adjust- distributionassociated with investment
mentsforwhichbenefit-cost criteriawere decisionsforwhichbenefit-costexpendi-
originally developed and typically ap- ture criteriahave been traditionallyem-
plied in the United States. This remains ployed will be of the second order of
true, I believe, despite Samuelson's and smalls in terms of its implication for
De Graaff'scriticismof the proposition measuringthe change and, forpractical
in general (33, pp. 10-11; 8, p. 90 ff.).At purposes,can be ignored(16, p. 50).5
themostfundamentallevel is De Graaff's Thus, whilethe absolute size of the na-
criticism,namely,if we wish to base our tional incomemay not be independentof
economics on an individualistic rather its distribution,a relativelysmall change
than an organic conceptionof the com- in its size, forpractical purposes,can be
munityand its welfare,and if there is consideredindependentlyof its redistri-
morethan one commodity,thenthe "ag- butionalconsequencesin determining the
gregatereal income" cannotbe evaluated magnitude and directionof the change.
withoutweightingcomponents,whichin The distinctionforbenefit-costanalysis,
turn is not independentof the income of course,is significant, as it providesthe
distribution.We may resortto the "size- theoretical basis for benefit-costprac-
distribution"dichotomy,but it has no tices. For, if the more simple criteriaof
operational significance,for "we do not Kal lor and Hicks had to be supple-
know the size unless we know the dis- mented even by Scitovsky's extension
tribution"(8, p. 92). 6 This is not necessarilymeant to implythat the
De Graaff'sconcernwith the distribu- redistributiveimplicationsforwelfare(discussed in
the followingsection) can be similarlyignored.Such
tional implicationsforaggregatereal in- welfare effectsmightbe important,especiallyif re-
come correspondsto Scitovsky'sconcern distributiveeffectsfromexpendituresfor develop-
Faith the redistributionalconsequences mentof resourceswerecumulative.But cumulation
hardlyseemslikelyin the area ofsuch expenditures.
4 Of course,it is not to be inferredthat Hicks Additional annual expendituresof the Bureau of
ignoresthe welfareissue, since he attempteda de- Reclamation,forexample,do notbenefitindividuals
fenseof such an interpretation of the rise in social previouslyfavoredbecause such additionalexpendi-
income.Amongothers,Hotelling(14) precededand turesrepresentpredominantlyextensionof the pro-
Wantrup(2) followedHicks in providinga defense gram to new areas and, accordingly,to different in-
of such welfareimplications. dividuals.

This content downloaded from 131.175.12.86 on Thu, 22 May 2014 11:23:52 AM

All use subject to JSTOR Terms and Conditions
 WELFARE ASPECTS OF BENEFIT-COST ANALYSIS 229

alone-not to mentionSamuelson'sall- tion may be shiftedoutward uniformly

possible-distributions test (33)-no ex or welfarepotentiallyincreasedby a pol-
antejudgment withrespectto theantici- icy prescription,the implementationof
pated changein economicefficiency re- the policy may cause such distortionof
sultingfroma contemplated supramar- marginal conditions and "undesirable"
ketallocationcouldbe supported byben- income redistributionthat the utility-
efit-cost analysis.This followsbecause feasibilityfrontiertwistsinward (33, pp.
the analystdoes nothave the powerto 18-21).
manipulatethe distribution of income The firstof these observations is re-
experimentally beforerendering a judg- late I directlyto condition (b), whereas
ment;nordoes thereexistsufficient in- the second bears obliquely on condition
formationregardingindividualprefer- (c).
encemaps to simulateresultsfromhy-
IV. THE WELFARE STATUS
potheticaldistributions.
OF THE STATUS QUO
Samuelson'srequirement that an im-
provement in efficiency mustbe tested Consideringcondition (b), I believe it
not onlyagainstthe ex ante and ex post fairto say that the redistributiveconse-
incomedistributions but againstall pos- quences of small changes of the sort en-
sible hypotheticaldistributions stems counteredin benefit-costanalysis in the
fromtwo partiallydistinctconsidera- United States are negligible. But, as
tions.The firstinvolvesthe degreeto Fisher points out, this does not mean
whichan implicitvalue judgmenthas that benefit-costanalysis is free of dis-
beenmadeineitherKaldor'sandHicks's tributionalvalue judgments.For equat-
criterion or Scitovsky'sdoublecriterion ing incrementalbenefitsand costs in the
in spite of the intendedpreoccupation designingof projects and "scoping" of
solelywithproduction orefficiency.That programsrelies on price data, which in
is, to taketheex ante distribution (or in turnare dependenton the prevailingdis-
thecase ofScitovskythe ex post as well tribution. Accordingly,if for no other
as theex ante) as a datumconfers toosig- reason than this, a judgment is implied
nificant a normative statuson thesepar- regardingthe normativestatus ofthe ex-
ticulardistributions. For, as Fisherhas istingdistribution.
observed: A properquestionto raise at this junc-
tion, however, is: "Can the prevailing
The refusalto makea value judgment. . . is
distributionindicate,as a pragmaticap-
in itselfa value judgment,not onlyin thesense
proximation,the socially sanctionedone
that one is saying that one ought to abstain
frommakingvalue judgments,but also in the
in a democraticallyorganized society?"
sense that the resultsobtained are those that
There are at least threeways in whichto
would result fromglorifyingthe presentdis-
tribution[7, p. 394]. answerthis question. One can replywith
a qualified"yes," followinga line of rea-
The second considerationunderlying soning to be sketched below. One can
Samuelson'spositionseems to be con- maintain that it is reallynot possible to
cernedwiththe distinctionbetweenthe know, but that the answer may not be
"utilitypossibilityfrontier"and the too important,given the level and dis-
"utilityfeasibility thatis, be- tributionof income approached in the
frontier,"
tween potentialand feasible welfare. United States. Finally one mightargue,
func- as does Little, that the prevailing dis-
Here,whilethe utility-possibility

This content downloaded from 131.175.12.86 on Thu, 22 May 2014 11:23:52 AM

All use subject to JSTOR Terms and Conditions
230 JOHN V. KRUTILLA
tribution ofincomedoes notenjoya so- A secondpossibleargument, relatedto
cial sanction, so thatanyjudgmentwith thefirst, is that,althoughtheremaybe
respecttoefficiency mustbe inthenature reasonsforrejectingthe prevailingdis-
ofa second-best solutioncontingent also tribution as in somesense"best"reflect-
on a value judgmentthat the resulting inga socialwelfare function, at thepres-
incomedistribution is "good." ent level of incomedistributional (and
It can be arguedthat the prevailing redistributional) questionsare notdomi-
distribution ofincomeis approvedbythe nantconsiderations.8 Thisdoesnotimply
community, since in a democracythe that all membersof societyhave their
community has themeansofchanging it. non-frivolous needsmet equallyamply,
Little discountsthis rationale,appar- butonlythattheassociateddead-weight
entlyonthebasisoftheobserved histori- lossesofmovingtowardgreaterequality
cal tendency towardpersistent reduction wouldbe judgedto exceedthe compen-
of inequalitiesin incomein modernin- satingdistributional gains.
dustrialnations(19,p. 114).Yet thede- A thirdpossibleargumentis thatad-
greeofincomeequalitysoughtbya com- vancedby Little.Littleabandonsinter-
munity maynotbe unrelated to thelevel estina welfare maximum, regarding such
of per capita income,and thereduction an aspiration as utopian.Considering the
of inequalityover timemay be onlya existingdistribution as non-optimal,9 he
functionof technological advancesand focuseson conditions sufficientfor an im-
theincreasein efficiency ofeconomicor- provement in welfareratherthanon con-
ganization. ditionsnecessaryfor a welfaremaximum
Differential rewardsofincomeappear (19,pp. 115-16).To renderan improve-
to be compatible witha Jeffersonian con- mentinefficiency desirable, thatis,forit
ceptionof democracywhichaccommo-
in his discussionof votingmethods
dates an aristocracy of ability.Indirect (26,7 Musgrave,
chap. vi), observesthe numerousdifficulties of
evidencethat the community sanctions implementingan unadulteratedsocial ethicand con-
someinequalityof incomehas been re- cludesthat perhapsmajorityvotingin a democracy
as close to achievingthe desiredend as the
flectedperceptively in the writingsof comes
mechanicsof social organizationwillpermit.
Perlman (27, pp. 164 ff.). Anthony 8 Of course,while such an argumentcan be ad-
Downs, in a whollydifferent fashion, vanced following,or perhapsinterpretingliberally,
providesan interesting rationaleforthe Fisher's argument(7, pp. 407-8), at such levels of
existence ofincomeinequality ina politi- income considerationsof efficiencymayis,also be of
the second order of significance;that freedom
cal democracy(4, pp. 199-200,but also fromextensionofgovernmentalinterventionmaybe
cf.p. 94).6If I interpretDownscorrectly,purchasedat the expenseof some relativereduction
it seemsprobablethat the incomedis- in potential national income. On the other hand,
maximumefficiency may be viewed as a possible
tribution resulting from the explicitre- good in its own rightirrespectiveofthe level ofopu-
distributive activitiesofthegovernmentlenceif the welfarefunctiondependson the relative
is, assumingthat the incidentalredis- rate ofgrowthvis-A-vissome ideologicallycompeti-
tive society.
tributiveconsequencesof othergovern- I
AlthoughLittle rejectsthe notionthat the pre-
mentalactivitiesare non-cumulative, a vailingdistributionofincomeis sanctioned,his will-
reasonableapproximation to thesocially ingnessto relyon pricesas indicatorsof value sug-
sanctioneddistribution.7 gests that its departurefromoptimalityis not suf-
ficientto affectrelativeprices-or else that individ-
B similarthat redis-
For a comprehensivesurveyof recentthoughts uals' preferencesare sufficiently
on egalitarianismwhich,thoughexhaustive,is not tributionswill not affecttheconstellationof relative
conclusive,see Lampman (17). pricesappreciably.

This content downloaded from 131.175.12.86 on Thu, 22 May 2014 11:23:52 AM

All use subject to JSTOR Terms and Conditions
 WELFARE ASPECTS OF BENEFIT-COST ANALYSIS 231

to improvewelfare,the attendant redis- Little (laterjoinedby Meade) attempts

tributionof income must be acceptable. to cometo gripswithproblemsof this
To pursuethe matterto its ultimatecon- nature.These problemsappearto be at
clusion, one could accept the frankly least as greatin practicalimportance as
ascientificapproach suggestedby Meade is theissueofinterpersonal comparisons
(25, chaps. v, vii) and assign distribu- and incomedistribution; and theyare,if
tional weights based on interpersonal anything,less susceptibleof an intel-
comparisonsof welfarein orderto incor- lectually satisfyingsolution. This is
porate redistributionalaspects into a broughthome decisivelyby the pessi-
multidimensionalobjective function.In misticconclusionsof the statementby
a similar,ifmorerestricted,sense thereis Lipseyand Lancasterofthegeneralthe-
the possibility,consistent with Little's oryof the secondbest (21). They have
position,ofmaximizinga one-dimension- demonstrated that,ifanyoneofthecon-
al benefitfunctionsubject to an income- ditionsfora Paretooptimumis not at-
redistributionconstraint." tainable,it is in generalnotdesirableto
V. WELFARE IMPLICATIONS OF NON- achieveanyoftheremainder (18,pp. 11,
OPTIMAL INITIAL CONDITIONS OF 26-27). The following quotationreflects
PRODUCTION AND EXCHANGE the flavorof theirnihilisticconclusions:
The third possibility noted immedi- The problem ofsufficient conditions foran
ately above touches on maximization increasein welfare,as compared to necessary
conditionsfora welfare maximum, is obviously
problemsbest treatedin connectionwith important ifpolicyrecommendations are to be
an evaluation of general condition (c), madein therealworld.Piecemealwelfare eco-
problemsassociated with the non-exist- nomicsis oftenbasedonthebeliefthata study
ence of the necessary conditions for a of thenecessary conditions fora Paretianwel-
Pareto optimum. Here it must be ac- fareoptimum maylead to thediscovery ofsuf-
conditions
ficient foran increasein welfare. In
knowledged that, by reason of market his Critiqueof Welfare Economics, I. M. D.
imperfections and distortionof marginal Little. . . says,". . . necessary conditions are
conditionsowing to governmentactivi- notveryinteresting. It is sufficient
conditions
ties (both explicitredistributionand the forimprovement thatwereallywant."But the
financingof supramarket allocations), theoryof secondbestleads to theconclusion
thatthereareingeneral nosuchsufficient con-
the Pareto-optimumconditions (c) are ditionsforan increasein welfare.Thereare
at best only approximated in practice necessary conditions fora Paretianoptimum.
and at worstare universallybreached,so In a simplesituation there mayexista condition
that the slopes of Samuelson's feasibility thatis necessary andsufficient. Butina general
frontiers have littlerelationto pricesand equilibrium situation,therewillbe no condi-
tionswhichin generalare sufficient foran in-
marginalcosts (33, p. 18). The significant creasein welfare without also beingnecessary
problemthen remainsof evaluating the fora welfare maximum [18,p. 17].
welfareimplicationsof benefit-maximiz-
ing criteriaunder real world conditions. And, to erode furtherthe faithof the
Of the major participantsin the post- innocent, theyconcludethat,in general,
war discussionofwelfareeconomics,only thereis no proofoftheexistence ofa sec-
10To implementthese suggestions,however,the
ond-best solution (18, pp. 27-28).11
benefit-costpractitionerwouldrequiresupplemental 11 ProfessorDorfman,in privatecorrespondence,
legislationsuspendingthe congressionaldirectivein has taken exceptionto this conclusion.In his opin-
the Flood ControlAct of 1936 to the effectthat dol- ion, a second-bestoptimumexists under the same
lar democracyis to prevail. conditionsthat a Pareto optimumdoes, but in gen-

This content downloaded from 131.175.12.86 on Thu, 22 May 2014 11:23:52 AM

All use subject to JSTOR Terms and Conditions
232 JOHN V. KRUTILLA
Non-existenceof a second-best solu-
tion in a technical sense, however,does
not mean that,ifan additionalconstraint
is imposedon the welfarefunction,there
is no actual relative maximum. Intui-
tively, we perceive that some adjust-
ments in response to the given con-
straints will be better than others and
that there must be an actual peak
whether or not we can stipulate what
conditionsmust obtain at the marginfor
all permutations.But the question re-
mains whetheror not supramarketallo-
cations, guided by marginal equalities
rather than the unknown appropriate
inequalities,will tend to move the econ-
omy furtheraway fromsuch an actual
relative maximum when similar mar-
ginal equalities are absent elsewherein
the economy. While it follows that, in
general,supramarketallocations guided
by marginal equalities will prevent the
economyfromachievingthe constrained
maximum,it does not followthat absten-
tion frominterventionwould permitthe
economy to remain closer to the con-
strainedmaximum.
The possibilityalways exists that an
observed inefficientsituation is domi-
nated by an attainable moreefficientsit-
uation (12, p. 98; 13, p. 208). An ex-
ample drawn fromthe fieldof resources
may illustratethis point. Three alterna-
tive plans of developmentwereproposed
forthe Hell's Canyon reach of the Snake
River,the thirdofwhichwas a privately
eral the familiarPareto-marginalequalitieswill not
holdat a second-bestoptimum.Therefore,underthe
second-bestconditions,an allocation of resources
that satisfiessome of the Pareto-marginalequalities
is not necessarilypreferableto an allocation that
satisfiesnoneofthem.Nevertheless,theredoes exist
a second-bestoptimum,that is, an allocation of re-
sourcesthat satisfiesthe distortingcontraintsthat
make"second-besting"necessaryand that is socially
preferableto any otherallocationthat satisfiesthose
constraints.
This content downloaded from 131.175.12.86 on Thu, 22 May 2014 11:23:52 AM
All use subject to JSTOR Terms and Conditions
advanced plan of development.For the
threeplans of developmentthe following
conditionshold:
01 >02 >03
and
I1 > 13 > 12,
where0 and I referrespectivelyto physi-
cal output and inputs,and the subscripts
to the respectiveplans of development
(16, chap. v). As betweenthe second and
thirdalternatives,it is obvious that the
efficiencyof the formerdominates the
latter. Hence, we can argue by domi-
nance that, subject to an economic de-
mand forthe output,appropriatepublic
interventionis a sufficientconditionfor
an improvementwithoutregardto neces-
sary conditionsfora welfaremaximum.
On the otherhand, sincethefirstalterna-
tive requiresmore inputs to achieve the
greateroutput than does eitherthe sec-
ond or the third,and the value of the dif-
ferencein output is greaterthan, equal
to, or less than the opportunitycosts of
the inputs dependingon a criticalfactor
price,we cannotmake a decisionwithout
recourseto prices. Now, ifprices are not
exact measures of opportunitycosts-a
conditionimplicitin the negative theo-
rem of the general theoryof the second
best-they do not providean unambigu-
ous criterion.
Three partially distinctpositions can
be discernedin the approaches adopted
by benefit-costanalysts in such cases.
McKean takes the followingposition:
Those conditions[Pareto optimum]are not
completelyrealized and moving toward the
achievementof one alone is not necessarilya
step in the rightdirection.However,if a fre-
quency distributionof the possibilitiesis im-
agined,it seemslikelythatincreasedproduction
whereprice exceeds cost would usually be a
step towardefficiency, even thoughthe other
The con-
conditionsare onlypartiallyfulfilled.
clusionhereis that pricesand costs show how
to "maximizeproduction"[24, pp. 130-31].
 WELFARE ASPECTS OF BENEFIT-COST ANALYSIS 233

Eckstein (5, p. 29), followingLittle of its operation.And, while he must rec-

(20, pp. vii-xiv), approaches the prob- ognize the prevalenceof departuresfrom
lem in the followingmanner: ideal conditions,he should not feed such
rInsofaras there are monopoly elements, departuressystematicallyback into the
prices will exceed marginalcosts, but froma optimizingcalculations. For, in contrast
quantitativepointofview,thesedeviationsare to the static situationto whichthe nega-
bothwidely-also perhapsevenly-diffused and tive theorem of Lipsey and Lancaster
relativelysmall,particularlyin the range of
marketsmostrelevantto waterresourcedevel-
applies, such feedbacks can have a cu-
opment.Projectsin thesefieldsproduceoutputs mulative effect,resultingin progressive
whichin largestpart are producergoods,such divergencesfromconditionsof optimum
as rawmaterials,electricenergyand transporta- productionand exchange.
tion services.In theseareas, advertising,con- The choices in which one alternative
sumerloyaltyand asymmetricmarketpower
clearlydominatesanother(or all others),
concentratedon the side of the seller are less
prevalentthan in marketsforconsumergoods. while theymay be numerousabsolutely,
Thus while pricesdo not serve theirfunction must stillrepresenta small proportionof
perfectly,we hold that theyare generallyade- the total of choices which face the eco-
quate for the range of policy decisions with nomic decision-makers.And the posi-
whichwe are concerned.At the same time,in
tions advanced above to deal with the
any applicationofthemethodsofthisstudy,we
mustkeep in mindtheassumptionswhichvali- more representativesituationcan neither
date theuse ofprices,and we mustnothesitate, rest on formalproofsnor claim much by
in certainsituations,to rejectthemin favorof way ofan intellectuallysatisfyingstatus.
othermeasuresof social benefitand cost.'2 Also, while the analyst must be sensitive
A thirdpositionsupportingthe use of to "higher-levelaims," he mustby virtue
prices (adjusted forobvious divergences) of the nature of his problemsand analy-
as measuresof opportunitycosts, and of sis work largely in a suboptimizingor
criteriabased mainlyon thepresumption partial-equilibrium context (24, pp.
of the existenceof marginalequalities is 30 ff.). Conclusions reached by analysis
the following(16, p. 73, n. 32). While the at this level of generalityneed not hold
benefit-costanalyst must recognizethat in more general cases, and may require
he does not institute utopian reforms substantial reconsideration(see, for ex-
of
simply by an act of analysis, he must ample, 21, 30, 32). The application
also recognizethat his criteria,in the dy- criteriafor improvingwelfaretherefore
namic contextof the real world, should cannot be a mechanicalor a compellingly
be consistentwith the higher-levelaims logical activity. Rather, it requiresper-
whichdominatethe workof those public haps more intimate knowledge of the
servantsresponsibleforpolicingmonop- economy, experience,and highlydevel-
oly, improvingmarketperformance,and oped intuitivesense than analysts com-
otherwisemonitoringthe economywith monlypossess-which suggeststhe qual-
the objective of increasingthe efficiency ityofresultsand degreeofprecisionto be
anticipated.
12
Consistentlywith the latterpart of the state-
ment,Ecksteinhas done much originalworkin de- VI. CONCLUSIONS FOR PRACTICAL
velopingbenefitmaximizing criteriasubjectto a vari-
CHOICES "IN THE PUBLIC
ety of constraintsin addition to the resource-tech-
nologyconstraint(5, 6). This effort
has been extend- INTEREST"
ed and generalizedby Steiner (39), Marglin, and
other membersof the IHarvard Water Resources Does the array of positions advanced
Seminar. previouslyprovidean adequate rationale

This content downloaded from 131.175.12.86 on Thu, 22 May 2014 11:23:52 AM

All use subject to JSTOR Terms and Conditions
234 JOHN V. KRUTILLA

forattemptsto evaluate the benefitsand
costs of resource-developmentalterna-
tives? Or are the commentshereintrans-
parentrationalizationswhichleave little
convictionthat analysis of benefitsand
cost and of their distributioncan help
significantlyto improvewelfarethrough
public intervention? One's view, of
course, will differ depending on the
nature of one's experience, one's tem-
perament, and perhaps also one's per-
sonal situation. The academic theorist
without responsibilityfor policy can
affordto (and probably should) be puri-
tanical withoutregardto whetheror not
this is immediatelyconstructive.On the
otherhand, the practicingeconomistin
government,chargedwith responsibility
to act under constraintsof time and in-
formation, will often be grateful for
perhaps even a perforatedrationale to
justify recommendations"in the pub-
lic interest." Since the alternative is
not to retire to inactivity but, rather,
to reach decisions in the absence of
analysis, we may take some comfort
from the belief that thinking system-
atically about problemsand basing deci-
sions on such analysis are likely to pro-
duce consequencessuperiorto those that
would resultfrompurelyrandombehav-
ior. Nonetheless,the utilityand welfare
effectsof benefit-costanalysis are likely
to be viewed differently, depending on
the end of the telescope throughwhich
the affectedparty is privilegedto look.

REFERENCES
1. BAUMOL, WILLIAM J. WelfareEconomics 9. HicKs, J. R. "FoundationsofWelfareEco-
and the Theoryof the State. Cambridge, nomics," EconomicJournal, XLIX (De-
Mass.: HarvardUniversityPress, 1952. cember,1939), 696-712.
2. CIRIACY-WANTRUP, S. V. "Concepts Used 10. . "The Valuationof Social Income,"
as EconomicCriteriafora SystemofWater Economic, N.S., VII (May, 1940), 105-24.
Rights," Journal of Land Economics, 11. . "The RehabilitationofConsumer's
XXXII (November,1956), 295-312. Surplus,"ReviewofEconomicStudies,VIII
3. COLM, GERHARD. "Comments on Samuel- (February,1941), 108-16.
son's TheoryofPublic Finance," Reviewof 12. HITCH, CHARLES. "Suboptimizationin Op-
Economicsand Statistics,XXXVIII (No- erationsProblems,"Journalof theOpera-
vember,1956), 408-12. tioits Reset-rihSorict of .1
,inrirr. i (MAyT.
4. DOWNS, ANTHONY. An EconomicTheoryof 1953), 87-99.
Democracy.New York: Harper & Bros., 13. . "Economics and Military Opera-
1957.
tionsResearch,"pp. 199-209 in "Econom-
5. ECKSTEIN, OTTO. WatcrResourceDevelop-
ics and OperationsResearch: A Sympo-
ment:The EconomicsofProjectEvaluation.
sium," Reviewof Economicsand Statistics,
Cambridge, Mass.: Harvard University
XL (August,1958).
Press, 1958.
6. . "A Surveyof theTheoryofPublic 14. HOTELLING, HAROLD. "The General Wel-
ExpenditureCriteria,"Paper read at Uni- fare in Relation to Problemsof Taxation
versities-NationalBureau Conferenceon and ofRailway UtilityRates," Econometri-
Public Finance,April,1959, University of ca, VI (July,1938), 242-69.
Virginia,Charlottesville. 15. KALDOR, NICHOLAS. "WelfarePropositions
7. FISHER, FRANKLIN M. "Income Distribu- of Economicsand InterpersonalCompari-
tion,Value Judgments and Welfare,"Quar- sons of Utility,"EconomicJournal,XLIX
terlyJournalof Economics, LXX (August, (September,1939), 549-52.
1956), 380-424. 16. KRUTILLA, JOHN V., and ECKSTEIN, OTTO.
8. GRAAFF,J. DE V. Theoretical WelfareEco- MultiplePurposeRiverDevelopment: Studies
nomics.Cambridge:CambridgeUniversity in Applied EconomicAnalysis. Baltimore:
Press, 1957. JohnsHopkinsPress, 1958.

This content downloaded from 131.175.12.86 on Thu, 22 May 2014 11:23:52 AM

All use subject to JSTOR Terms and Conditions
 WELFARE ASPECTS OF BENEFIT-COST
17. LAMPMAN, ROBERT J."Recent Thoughtson
Egalitarianism,"Quarterly Journalof Eco-
nomics,LXXI, No. 2 (May, 1957), 234-66.
18. LIPSEY, R. G., and LANCASTER,KELVIN.
"The GeneralTheoryof theSecond Best,"
Reviewof EconomicStudies,XXIV (1956-
57), 11-32.
19. LITTLE, I. M. D. A CritiqueofWelfareEco-
nomics.Oxford:ClarendonPress,1950.
20. . The Price ofFuel. Oxford:Claren-
don Press,1953.
21. . "Direct vs. IndirectTaxes," Eco-
nomicJournal,LXI, No. 243 (September,
1951), 577-84.
22. MARGOLIS, JULIUS. "A Commenton the
Pure Theory of Expenditures,"Reviewof
Economicsand Statistics,XXXVII (No-
vember,1955), 347-49.
23. . "Secondary Benefits, External
Economiesand the Justification of Public
Investment," ibid., XXXIX (August,
1957),284-91.
24. MCKEAN, ROLAND. Efficiency in Govern-
mentthrough SystemsAnalysis,withEmpha-
sis on Water ResourceDevelopment. New
York: JohnWiley& Sons, 1958.
25. MEADE, J. E. The Theoryof International
EconomicPolicy,Vol. II: Trade and Wel-
fare. New York: OxfordUniversityPress,
1954.
26. MUSGRAVE,RICHARDA. The TheoryofPub-
lic Finance.New York: McGrawHill Book
Co., 1959.
27. PERLMAN, SELIG. A Theoryof the Labor
Movement. New York: MacmillanCo., 1928.
28. REDER, M. W. "A Reconsiderationof the
MarginalProductivity Theory,"Journalof
PoliticalEconomy, LV (October,1947),450-
58.
29. REGAN, MARK, and TiMMONS, JOHN F.
"Benefit-CostAnalysis," paper presented
beforea jointsessionof theEconomicsand
EngineeringSection of the AmericanAs-
This content downloaded from 131.175.12.86 on Thu, 22 May 2014 11:23:52 AM
All use subject to JSTOR Terms and Conditions
ANALYSIS 235
sociationforthe Advancementof Science,
Berkeley, California,December 27, 1954.
Reproducedby the Committeeon theEco-
nomicsofWaterResourcesDevelopmentof
the WesternAgriculturalEconomics Re-
searchCouncil.
30. ROLPH,EARL,and BREAK, GEORGE. "The
WelfareAspectsof Excise Taxes," Journal
of Political Economy, LVII (February,
1949), 46-54.
31. RUGGLES, NANCY. "The WelfareBasis of
the Marginal Cost PricingPrinciple,"Re-
view of Economic Studies, XVII, No. 2
(1949-50), 29-46.
32. . "Recent Developmentsin theThe-
ory of Marginal Cost Pricing,"Reviewof
EconomicStudies,XVII, No. 3 (1949-50),
107-26.
33. SAMUELSON,P. A. "Evaluation ofReal Na-
tional Income," OxfordEconomicPapers,
N.S., II (January,1950), 1-29.
34. . "The Pure Theory of Public Ex-
penditures,"Reviewof Economicsand Sta-
tistics,XXXVI (November,1954), 387-89.
35. . "Aspects of Public Expenditure
Theories," ibid., XL (November, 1958),
332-38.
36. ScITovsKY, TIBOR. "A Note on Welfare
Propositionsin Economics,"ReviewofEco-
nomicStudies,IX (1941), 77-88.
37. . "A Reconsiderationof the Theory
of Tariffs,"ibid., IX (1942), 89-110. Re-
printedin pp. 358-89 of Readingsin the
Theoryof InternationalTrade,ed. Howard
S. Ellis and LloydA. Metzler.Philadelphia:
BlakistonCo., 1949.
38. SIMON,HERBERT. "A BehavioralModel of
Rational Choice,"Quarterly JournalofEco-
nomics,LXIX (February,1955),99-118.
39. STEINER, PETER 0. "Choosing among Al-
ternativePublic Investments,"American
EconomicReview,XLIX (December,1959),
893-916.