Kuznets's Inverted U-Curve Hypothesis: The Rise, Demise, and Continued Relevance of a Socioeconomic Law Author(s): Timothy Patrick Moran Reviewed work(s): Source: Sociological Forum, Vol. 20, No. 2 (Jun., 2005), pp. 209-244 Published by: Springer Stable URL: http://www.jstor.org/stable/4540893 . Accessed: 01/01/2012 18:39 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 ofcontent 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 [email protected]. Springeris collaborating with JSTOR to digitize, preserve and extend access to Sociological Forum. http://www.jstor.org
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Kuznets's Inverted U-Curve Hypothesis: The Rise, Demise, and Continued Relevance of aSocioeconomic LawAuthor(s): Timothy Patrick MoranReviewed work(s):Source: Sociological Forum, Vol. 20, No. 2 (Jun., 2005), pp. 209-244Published by: SpringerStable URL: http://www.jstor.org/stable/4540893 .
Accessed: 01/01/2012 18:39
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 [email protected].
Springer is collaborating with JSTOR to digitize, preserve and extend access to Sociological Forum.
Kuznets'sInvertedU-CurveHypothesis:The Rise,Demise, and ContinuedRelevanceof a Socioeconomic Law
Timothy Patrick Moran1
This paper provides a historical analysis of the changing significance ofthe most influential statement ever made on inequality and development-Simon Kuznets's "inverted U-curve hypothesis." The shifting interpretationsand appropriations of the hypothesis over time-from its status as a spec-
ulative supposition in 1955, to its rise and fall as a reified socioeconomiclaw, to its contested standing in the social sciences today-demonstrate how
Kuznets's arguments,originally advanced under more limited conditions, be-
came transformed into overarching theoretical, empirical, and political con-structions. This history suggests that even empirically grounded and testable
social science models are contingent on the broader social and political con-
most significant ndconsequentialpropositionsn developmentacademics
andpolicy.In the 30 yearsfollowingits 1955publication, he inverted U-
curve was transformedacrossthe social sciences from a speculative hy-
pothesisto an inevitable and unavoidablesocioeconomic "law" hatpro-vided both scholarsandpolicymakerswith an articulatedworldviewof the
nature of growthandinequality.Kuznets'shypothesisshifted the framingof incomeinequalityas a socialproblemto the examinationof size distri-
butions,spurred he first worldwideefforts to collect suchdata,createda
prolificprogramof research,and helped establishworldeconomicpolicyat a time when policymattered.Characterizinghe prevailingview at the
heightof the U-curve'sascendancy,Srinivasan1977:15) alledthe hypoth-esis "somesort of 'iron aw' of
development."Nownearly50years ater,thetheoreticalandempirical tandingof the
hypothesis s stillambiguous, ontroversial, nd relevant.According o the
Social Science CitationIndex, in the 5 years prior to 2000, nearly 500 articles
from a wide cross-sectionof socialscienceinquiryreferenced he U-curve
article.Withvariousamountsof fervor,manyscholarsnow aim to refutethe
hypothesis,while otherscontinueto arguein its favor. Nielsen (1994:654),for example,concludesthat "despitesome isolatedskepticism... the ex-istence of this curve has been well documentedfor developingsocieties
by laterresearch."Dovring (1991:101)underthe heading"Is the U-CurveInevitable?"states that "theempiricalmaterialdoes, on the whole, sup-port Kuznets'hypothesis. .. [It]has withstoodthe test of time aswell asthatof ongoingresearch."Other scholarswriting n the last decade still re-fer to the U-curveas "anaturaland unavoidable growth-equity)conflict"
In thisarticle, analyze hehistoryof the theoreticalandempirical on-
ceptualizationof the world known as "Kuznets's nvertedU-curve,"with
the aim of accomplishingwo goals.The study'sprimaryaimis to providetime andplacecontextto theshiftingapplications nd nterpretationsf the
U-curvehistorically.Understanding ow the U-curve(andbyextension n-
equality tself) fit within argeracademicandpolicycontexts,andhow this
fit changedthrough ime,will help us to understandbothpastevaluationsof the U-curve as well as the currentambiguities urrounding set of ar-
gumentscentralto manyareas of current ociological nquiry.Second,the
analysishas important mplications or a broaderset of theoretical ssues
concerningthe nature of social scientificpracticeand its location within
largersociopolitical ontexts.In thissense,the storyof the U-curve-a the-ory that triumphedn certain timesand placesandfailedin others,yet isstillhotlycontestedhalf a centuryafter tspublication-is relevant o mod-els of "progress"n social scientificknowledge (as explored,for example,
deal of debatetodaysurrounding amingandclassificationssues nthesoci-
ologyof science ield.I usethephrase"sociology f scientificknowledge"o refer o the workof the constructivistcholarswhoresearch he internalworkingsand contentsof science, he"internal ynamics f science n themaking" Fuchs,1992:3).Thisapproachscognitivelydif-ferentiatedbyitspractitionersrom both"classical"ociologyof knowledge,and Mertonian
and pure theoreticalspeculation.Kuznetsbegan by examiningwhat ten-tativeempiricalevidencewas availableat the time.3Using historicaldataon the size distributionof income within the United States, England,
and Germany,Kuznets found that inequalitywas characterizedby long-term stability,before beginningto decline startingat least around1920
(and probablysince the periodbefore World War I). This trend was ac-
companied by significant ncreases in real income per capita. Thus, he
concludedthat at advancedstages of economicdevelopment,inequalityfirst leveled off then began to decline as these economies continued to
expand.Kuznets was somewhatsurprisedby these findings, noting that "a
long-termconstancy, et alone reduction,of inequality... is a puzzle" (7).More specifically,he wondered how two developmental processes thateconomistshadlong assumedto be fundamentallynegalitarian-the con-
centrationof savingsat the top of the incomedistribution, ndthe simulta-
verging ncomes.Aftersummarizinghe prevailing hinkingon the former,Kuznetsdelved moreanalyticallynto the latterpartof thepuzzle.Througha simplemathematical xercise conductedon a stylized,dual-sectorecon-
suggestedthat the narrowingof inequalityobservedin these three devel-opedcountrieswasactually"relatively ecent andprobablydidnotcharac-terize the earlierstagesof theirgrowth" 18).
Since, according o Kuznets,the intrasectoraldistributionof incomeis necessarilywider in the urban sector than in the ruralsector,a majorshift in populationfrom a sector with low inequalityto one with greaterinequality ncreasesthe weightof the unequalsector,thusraisingoverall
inequality.Perhapsmore significantly, owever,this patternholds even if
the assumptionof wider nequalityn the urbansectoris relaxed.If income
levelsbetweenurbanandruralsectorswiden(due to greaterproductivitygainsin the former),then overallinequalitywill againhave to rise, evenif intrasectoralnequalitiesare assumed o be the same. These two mathe-
matical ruismsprovidethe basisforexplainingncreasingnequality imul-taneous witheconomicgrowth n less developedcountries,andlead to the
"inevitability"f the growth-equity radeoff,as manyauthorshave cometo characterize he "Kuznetsprocess."While either one of these scenar-ios alone would lead to increasing nequality,Kuznetsthoughtthat bothconditionswereprobablypresent n developingeconomies.
3Speaking to the quality of the data he employed, Kuznets (1955:4) characterized the results
as "preliminary informed guesses." This work is subsequently cited by page number in the
If the nature of sectoral imbalances eads to widening inequalityas
populationshifts fromruralto urbanproduction,what causes the reversal
of thistrend?Kuznets'sanswerwasrooted in the dynamicsof the produc-tion process,but he also recognizedthe importanceof political,sociologi-cal, and demographicprocesses.Kuznetsarguedthat "themajoroffset tothe widening nequality...musthave been a rise in the income share of thelowergroupswithinthenonagriculturalector of thepopulation" 17).Thisis accomplished hrough he institutionalandpoliticalchanges nherent n
"thedynamismof a growingand freeeconomicsociety."Thesechanges n-
volve shifts n the economic andpoliticalexpectationsregardinghe "long-term utilityof wide income inequalities," he growthof a "native"urban
population"moreable to take
advantageof the
possibilitiesof
citylife in
preparationor the economicstruggle,"and the growingpoliticalpowerofthisgroup eadingto protectiveandsupporting egislation(9-17). In other
words,the dislocating ffects of the industrial evolutioneventually ead tosocial andpolitical organizationat the bottom,whichultimately ranslatesinto improvingabsoluteandrelativeeconomicpositionsover time.
In keepingwithhis uniquescientificphilosophy,Kuznetswas deliber-atein recognizinghe limitationsof both theavailabledata and his theoret-ical ventures.He openlycautioned hat"noadequateempiricalevidence s
available orchecking hisconjecture" 19),and he prefacedhisconcludingremarkswiththe caveatthat the invertedU-curve was based on "perhaps5%empiricalnformationand95%speculation, omeof it possiblytainted
bywishful hinking" 26).Butevokinghisgenuineconcern or the scientific
enterprise,Kuznetsargued hat"aslongas (thisresearch) s recognizedasa collection of hunchescallingfor further nvestigationrather than a setof fully tested conclusions, ittle harm andmuchgood may result"(26).4The U-curve, however, was quickly appropriatedby practitionerswhowere much less cautious,an academicprocessof which Kuznets was well
aware:
The timelag maybe longbetween he establishment f empirical indingsand their
absorptionnto a testedtheoryof reliably nvariantpatterns... But the timelag isshortbetweennew findingsandtheiruse to enrichthe background gainstwhichbroadcurrentproblemsare considered. ndeed, hedanger s not that such indingswill be usedforpracticalpurposes,but rather hat the resultswillbe eagerlyseized
uponto yielda spateof hypotheses hat claimtoo muchgenerality, .. (that) mayprovidea startingpointfor distorteduse or for unwarranted ogmaticgeneraliza-tions.(Kuznets,1961:13)
4Kuznetspracticeda distinctiveand nearly nvariantapproach o social scientificresearch.As Lundberg 1984:528) ummarizesn his excellentreview of Kuznets's cholarship, The
specialfeaturesof Kuznets'sanalysisare derived romhis attempts o combinequantitativeprecisionwith total overview, ncluding hought-provokingpeculationson differentwaysandmeans o identify ecular nterconnections."
In the firsttwo decades after its publication, he invertedU-curvehy-
pothesisthatKuznetshadcautiouslywarnedwas basedon "95percentspec-ulation"wastransformed nto an unproblematic blackbox"so widelyac-
ceptedthroughouthe social sciencesthatits existencewas seen as a matter
of fact, and the processesthat created it were considered awlike.Usingthe languageof LatourandWoolgar(1986),black boxes areestablished-or ordinary tatementsturned nto universal acts-through the "deletion
of modalities" uch as thoseidentifying ime, context,or other factors thatbound statementsby particularities. or the U-curve,such modalitieswere
demic and political consensusin the nascent field of developmenteco-nomicsin the 1960s and 1970s: he empiricaldiscoursesurrounding ewlycollected socioeconomicdata; he theoreticaldiscourse urrounding ualis-
One of the mostprofoundandlasting mpactsof Kuznets'sarticlewasa new sense of urgencyregarding he need to collect and analyzereliablecross-nationaldata on income inequality.Inspiredby Kuznets'sprelimi-naryresearch,economistsbeganto moveawayfromconceivingof inequal-ity as the functionaldistributionbetween factorsharesand startedto as-semble data on the size distributionof personalincome withinnations.5Kravis(1960),a studentof Kuznets's,was the firstto applysuch data tothe U-curvehypothesis.He found a positive (althoughcomplex)relation-
shipbetweenincomeinequalityand incomeper capita,and in generalcon-firmed most of Kuznets'sarguments.Oshima(1962), in a study of four
Asian countries, henconcluded hatalthoughchanges n income distribu-tion arenot a result of economicgrowthper se, the relationships indeedcurvilinearand explainedby the processof dualisticdevelopment.Usingintersectoral ncome distributiondata collectedby Kuznets(1963)himself,
Cutright 1967:578) onductedone of the firstquantitative nequality tud-ies in sociologyand endedbystating hat "the economistmaybe heartened
5This movement represents a significant intellectual shift. Although the study of personal in-come distributions dates back to Pareto and earlier, social scientists prior to Kuznets concep-tualized inequality as differentiation between the factors of production more than people or
households. In fact, many thought the latter to be an uninteresting if not unimportant social
that his interpretationof inequalityas a function of the level of economic
developmentalso receivedsolidempirical upport."The numberof studiesaddressingKuznets'shypothesissurgedin the
1970s.Scholarsaffiliatedwithvarious nternationaldevelopmentagencies(primarily he WorldBank) began a concerted effort to collect and con-solidatesignificant mountsof nationalsurveydata,and the appearanceofseveral arge aggregatesof cross-national ataby the mid-1970sbegotsub-stantialamountsof quantitativeiterature.6The cross-sectionalasopposedto longitudinal)characterof this earlydataproved important or evalua-tions of the U-curve.Duringthis period, in nearly every dataset and for
nearly everytime period,high-incomecountriesshowed the lowest levels
ofinequality
ollowedby
the low-incomecountries,
and the middle-income
countriesdisplayed evels of inequalityconsiderablygreaterthanthose of
the othergroups.Thus when levels of economicdevelopmentwereplottedon the horizontalaxisagainst evels of inequalityon the vertical, he resul-tantshapewas a clear invertedU with the fittedregression unction nvari-
ably revealinga negative,statistically ignificantquadraticerm.Althoughthe shortcomingsof such data were well known at the time, and althoughmost recognizeda degree of scatter aroundthe curve,scholarsproceededto takethisplace-in-time napshotand infer that thecountries orming he
invertedU-shapein cross-sectiongot to thesepositionsby followingan in-vertedU-trajectoryof risingthen fallinginequalitiesover time. In retro-
spect,Fei andRanis(1997:324)describethis traditionof researchas "pro-vidingverificationof the seeminglyincontrovertiblenature of Kuznets'scross-sectionalvidence,accompaniedbysomeplausible, f loose,hypothe-ses as to its causes."
Majorstudies of theperiodincludeAdelmanand Morris 1973, 1978),Ahluwalia (1974, 1976a,b), Chenery and Syrquin (1975), Jain (1975),Paukert(1973), Srinivasan 1977), and Weisskoff(1970). In perhapsthe
most influentialof these papers,Ahluwalia(1976b:38) ertifies hat"thereis strong supportfor the propositionthat relativeinequality ncreases inthe earlystagesof development,with a reversalof thistendency n the later
stages."In a rareinstancewherelongitudinaldatawas available,Fishlow
(1972, 1973) found that the Brazilianhigh-growth"miracle"of the late1960s and early 1970swas not egalitarianbut detrimental o low-income
groups. n the late 1970s,Ranis(1977:44) eviewed he assembledempirical
6Themostprominentdatasets o emergeduring hisera werecompiledbyMontekAhluwalia
JacquesLecaillonetal. (1984)with the InternationalLabourOrganization;MalcolmSawyer(1976)withtheOECD;andanindependent ffortbyIrmaAdelmanandCynthiaTaftMorris
research and concluded that "it is fair to say that the overwhelmingevidenceto datepointsin the directionof aninevitable,and rathersevere,conflictbetween most measuresof equityfora given societyanditsgrowth
performance."
The U-Curveas anImplicitAssumption
While the U-curve emerged as an empiricalregularity n the earlycross-nationaliterature,plentyof seeminglyconfirmed deas do not ascend
to lawlikestatus; hatis, theydo not become closedblackboxes. As Yonay
(1998:20)explains, t is not enoughfor a contribution o be recognizedas
"valid" nd then set aside; t must "becomepartof a largerapparatuswhich
canbe usedregularlywithoutneed tojustify(its)use."For the U-curve, he
largerapparatuswas the academicandpoliticalconsensus hat definedthe
postwardevelopmentalist egimefrom the 1950s o the 1970s.Much has been writtenfrom variousperspectiveson the researchand
politicsof the formativeperiodof developmenteconomics ollowingWorld
War II (Datta, 1986; Escobar,1995;Hunt, 1989;Leys, 1996;Meier and
Seers, 1984;Sachs,1976).Nearlyall agree,however,to the highdegreeof
paradigmaticonsensus hat defined heperiod.Newlyidentified"develop-
ment economists"combined a strong practicalorientationwith real opti-mism that intellectualefforts to understand he "viciouscircle of poverty"(as Nurkse(1953) famouslystated the problem)would translate nto the-
oreticallyinformed,state-ledplanningthat would yield rapidresults. As
Martin(1991:27)recalls,by the end of the 1950s"somethingike a main-streamtheoryof EconomicDevelopmenthademerged,"a set of proposi-tions and a commonlogic of approachemanating romEnglandand theUnited States,rooted in classicaltraditions, hat equated"development"withrisingoutputgrowth,agrarianransition,and industrialization.
Hunt (1989:86)calls the early developmentconsensusthe "paradigmof the expandingcapitalistnucleus,"because at the timenearlyeveryoneagreed that the initialgoal for poor countriesseeking to jump-start he
processof industrializationwas to nurturea class of investors that could
slowly accumulatecapitaland transferunderutilizedabor out of agricul-ture into industry.At the discursivecore of this paradigmwas agrariantransformationbased on the conceptof the dualeconomy,best exempli-fiedbyArthurLewis's 1954,1958)highly nfluential"surplusabormodel."
Dualismreferred o the asymmetricaloexistenceof (andconflictbetween)
a "modern" somewhat ndustrialized) ector with a "backward"mostlyagricultural)ector.Accordingto Escobar(1995:78), he conceptof dual-ism "pervadedhe developmentviewof mosteconomistsandinternational
In Lewis'smodel, later elaboratedmathematicallyby Fei and Ranis
(1964), industrialproductionslowly expandson the basis of agricultural
surplusavailableas wage goodsandunlimited"surplusabor"drawn rom
the countryside.The key ingredient n thisprocess, emphasizedby Nurkse(1953),Lewis(1954),andmanyothers,was the role of savingsandspecifi-callyits relativecontribution o nationalproduct.As Lewis(1954) explains,
The centralproblem n the theoryof economicdevelopments how to understandthe processby whicha communitywhichwaspreviously avingandinvesting4 or5 percentof its national ncome or less,converts tselfinto aneconomywherevol-
untary aving s running t 12to 15percentof national ncomeormore.
Theemphasis
onsavings
n the dualistic rameworkprovedimportantfor the ascendenceof the U-curve because it implied a necessarylink
in the early stages of development between the goal of rapid capitalaccumulation nd the need to (re)distributencome to those who can bestsave and invest. This "growth-equityradeoff" s one of the oldest tenetsin economics,andit provedto be a central dea in the earlydevelopmentconsensus.Developmentscholarsheld a fundamentalbelief in the classical
presumptionhat,sincewagesarefullyspenton consumption,he savingsrate of workers is alwaysassumed to be zero. Kalecki's(1942) aphorism
terselyarticulated he idea: "Workerspendwhatthey earn andcapitalistsearn what they spend." Since only capitalistssave and invest (at leastto a significantdegree), economicgrowthrequiresa changein the classdistributionof control over these resources;hence, the wealth-generatingadvantagesof inequalityare not only to be expected, but inequalityis
actually een as a necessarypreconditiono growth indeed,thiswasone ofthe reasonsKuznetsexpressedsurpriseat his initialfindings). nequality'srole in the developmentprocess s explainedquite clearlyby Lewis(1954):"We are not interestedin the people in general,but only say in the 10
percentof them with thelargest ncomes .. theremaining 0percentof thepeople nevermanageto save a significantractionof theirincome.... Thecentralfact of economicdevelopment s that the distribution f incomes isaltered n favorof the savingclass."
While the growth-equitytradeoff Kuznets discovered was theoreti-
cally built into the structureof early models of development,empiricalresearch at the time also seemed to confirm the intersectoralassump-tions of Kuznets'smodel; thus,it becamean intuitivebelief that the exis-
tence of the invertedU-curvecouldbe explainedby theproblems nherent
in dualisticdevelopment.Weisskoff(1970) and Swamy(1967),for exam-
ple, showed thatinequality n low-incomecountries s generallygreater n
the industrial ectorthan in the agriculturalector. Adelman and Morris
(1973)also confirmed hat economicgrowthusuallyproceeds n a dualistic
paribus,national ncome becomes more concentratedn the earlystagesofdevelopmentand more equallydistributed ater on. Robinson(1976:437)derivedthe U-curvefrom a simple, two-sector econometricspecificationand concludedthat the hypothesis s a "necessarymplicationof the work-
ingsof such models"(emphasisadded).Combinedwith such overwhelmingempiricalsupport,Kuznets'shy-
pothesisthusprovidedan eloquentand coherentsynthesisof the analyti-cal andtheoreticalconsensuses hat defined the field of developmenteco-nomicsin the 1960s and 1970s.It affordedan empiricallygroundedmodel
of theprevailing xpectation hat intersectoralnequalitiesmust rise as sur-
plus labormoves fromthe countryto the city,andmayonly be mitigatedin an advancedstage of developmentwhen the modern sector becomes
large enoughto continuegeneratingcapitalas wagessimultaneously tart
to rise (i.e., afterthe laborsurplusprocessruns its course).The U-curvedemonstrated he inevitable results of dualisticdevelopment workingin
practice,and the hypothesiscan be seen as an importantcorollary o this
theoreticalconstruct.Somearguethat the theoreticalconsensusand the U-
curve to which t fitcombined o form a sort of self-fulfilling evelopmentalprophecy,asDattaexplains:
The Kuznets inverted U hypothesis can be derived from the Lewis model and hence,
undermining the inverted U hypothesis also undermines the Lewis model... Nei-
ther in its original elaboration nor in its prevalent implementation has the Lewis
model been different in spirit or approach from the perception of the Kuznets hy-
pothesis, so much so that in writing a critical survey Ranis (1977) himself lumpedKuznets and Lewis together in his paper. (Datta, 1986:18)
As a resultof its paradigmatic fit"within the largerapparatus f the
developmentalconsensus, he invertedU-curve ascended n statusuntil,asSrinivasan1977)and Robinson(1976:437) tate,it had"acquiredhe forceof economic law."By the end of the 1970s,the invertedU-curveprocesswasuniversally onsidereda generalstatementof thehistoricaland univer-
sal relationshipbetween incomeinequalityand economicgrowth.As earlyas 1974,Ahluwalia(1974:17)refers to the invertedU-patternas "a char-acteristicof the developmentprocess affectingunderdeveloped ountries."FrankandWebb(1977:10)call the hypothesisa "natural ffect"causedbydualisticdevelopmentand reinforcedby governmentpolicy.In an influen-
tial andwidelycited reviewof the distribution nddevelopment iteraturein whichthey call the U-curve a "stylized act,"Adelmanand Robinson
(1989:960)write, "The initial decline in the shareof income of the pooris inevitable and arises throughthe introductionof a small high-income
islandin a largelow-incomesea;.., overall the tendencyis for inequalityto increasefor a considerable ime."It is during his time that the scopeof
the U-curve'sstaturespreads romdevelopmenteconomicsto widespread
recognitionacrossthe social sciences.In sociologyandpoliticalscience,aswell as economics,scholars topped testingfor thepresenceof the U-curveandbeganto debate the relativeweightof different ndependentvariablesinvolved n itspersistence n cross-sectional ata.
The U-Curveas PoliticalAffirmation
The transformation f the U-curve into a universal"blackbox" en-
tailed a thirdintersectingdiscourse n whichthe U-curve became an im-portantpoliticalmetaphor n the historically adicalandrevolutionarydeathat therichcountriesof the worldcould andwouldfacilitate"developing"theformercolonies of theso-calledThirdWorld.Fromtheverystartof this
project n the 1940s,the conceptof developmentwas reducedto growth n
alesced around heproblemsof limitedcapitalaccumulation nd inefficient
labor andcapitalallocation,and these concernscame to be regardedas the
problemsof
development, rrespectiveof the social, cultural,or
politicalimplicationsnvolved.Inequalityandpoverty n thisworldviewwere either
ignored,or consideredautomaticbyproducts,"undesirable"ertainly,yetnecessarypreconditionsand inevitable outcomes of the successfuldevel-
opment process.As Lewis (1976:26)flatlystated, "developmentmust be
inegalitarian."But the worseningof wages and livingstandardswere not somehow
peripheralo the models andpoliciesof developmentpractice;ndeed,they"belongedto their inner architecture"Escobar, 1995:80).Withinthe du-
alisticframework,t was all too easyto rationalizemeasures o hold downwagesinrural-urbanermsof trade nthe "efficient" llocationof laboranduse of profits.In thissense,the metaphorof the invertedU-curve couldbe
invoked,eitherexplicitlyor by unstated ntuition,to pacifyconcernsoverthe detrimental ffectsof theideologyof economicgrowth-anythingthat's
At the sametime, a real (some have saidnaive) optimismpermeatedthe developmentalconsensusbasedon widespread aithin the inherently
progressivenatureof industrialization. hedevelopmentcommunityof the
7Thepoliticalmessageof the invertedU-curvemetaphor ives on in the various"U-curves"in existencetoday, including he environmentalU-curve and the demographic ransitionU-curve o name two.
calculatedmessagethatthegrowthprocess tself waseventuallyequalizing.Thushigh-growth olicesantithetical, r at leastambivalent,owardequal-
ity would not condemnsocietyto acceptingcurrent nequalities orever-
the institutional ndpoliticalchanges nherent n "thedynamism f a grow-
ing and free economicsociety"would eventually prove to be egalitarian
(Kuznets,1955:9).A commonpractice n the empiricalarticlescitedabove,forexample, nvolvedsolvingthe quadratic quationsyielding he inverted
U-curve shape to obtain "turningpoints"indicatingwhen in fact these
changescould be expected.As Kuznets 1955:17)himselfremarked,"Much
is to be said for the notion thatonce the earlyturbulentphasesof indus-trializationandurbanizationhad passed,a varietyof forcesconvergedto
bolster the economicpositionof the lower-incomegroupwithinthe urban
population."Thus,beyondthe unquestionable mpactof the U-curvehypothesis n
the academy, ts largely unqualifiedacceptancewas also grounded n the
politicsof the internationalpolicyarena.Duringthisperiod,developmentscholarsregularlyworked orinternational rganizationsuch astheUnited
Nations and WorldBankand often advised ndustrializing overnmentsdi-rectly, exerting"enormous nfluenceupon the developmentdiscourse...and thereforeon development n practice" FitzGerald,1991:16).
PART II: THE OPENING OF A BLACK BOX
The unanimitysurrounding he U-curve hypothesis began to face awide challengein the 1980s.By the end of the decade,the interpretation
of the U-curve turned frombeing generally regardedas an "ironlaw ofdevelopment"to a contentioushypothesisassociated with contradictoryfindingsand ambiguousconclusions.At firstglance, this transitionmightseem likeanapt description f thesortof scientific evolutionsKuhn(1962)described-routinizedpuzzle-solving eadingto the discoveryof empiricalanomaliesthatcouldnot be explainedwithin the U-curve framework.Butthe reinstatementof modalities hat werepreviouslyconsidereddeleted-
the transitionof an unproblematicblack box into a limited statementde-
pendenton particular ontexts-needs to be precededby the deconstruc-
tion of the widerapparatuswithinwhich he blackbox was seen as an essen-tial foundation.In the case of the U-curve,thisprocesswasaccomplishedthrough heveryun-Kuhnian ounterrevolutionndevelopmenteconomicsthatby the end of the 1980s eft thefield in completeruins.
academicsand policymakingwas being completelytransformed romtheone that emerged duringthe formativeperiod.As Leys (1996:20)charac-terizesthe shifting imes,"The world in whichKeynesianpolicymaking-and its offshoots,developmenteconomicsanddevelopment heory-madesense had changed fundamentally."Duringthis decade,the developmen-talistprojectbeganto draw irefrom allpoliticaldirections or two reasons.
First,it becameincreasingly lear that the identification f socialprogresswith economicgrowthwas not solvingthe long-standingproblemsof in-
equalityand
povertyas the
early developmentalmodels had
predicted.Fishlow's(1972, 1973) celebratedfindingthat the Brazilianhigh-growth"miracle"wasinegalitariano such an extent that thepoor (and especiallythe ruralpoor) had become worse off despite rapid growth,was followed
by Bardham's1973) study,whichfound the same effect in India.Second,a
series of politicaldisastersoccurred n the ThirdWorld,ranging rom civilwars in Africa to the spreadof violent military dictatorships hroughoutLatinAmerica,andthese were takento be, in the words of one develop-mentpioneer, "clearly omehowconnectedwith the stressesandstrainsac-
companyingdevelopmentand 'modernization' (Hirschman,1981:23). nshort,developmentalismwasnot working,andby the end of the 1970sthe
The initialresponsewithin the developmentcommunitywas not toabandon he fieldbut,ironically, o enlarge t through nternalself-critique.The meaningof developmentwasexpanded o includenewobjectives uchas social welfare, environmentalprotection,and women's rights,and in
the early1970s he Robert McNamara dministrationntroduceda povertyagendaat the WorldBank.According o thisnewapproachno tradeoffwas
necessarybetweenthe goalsof outputgrowthandworseningsocial condi-tions such as inequality;he fruits of growth, t wasnowthought,could bebetter distributedwithouteffectingfutureeconomicperformance.n 1974,Hollis Cheneryand his associatesat the WorldBankwrote Redistribution
with Growth,a seminal culminationof work under the new World Bank
approachthat ultimately "changed he course of development thinking"
(Bruno and Pleskovic,1996:2).Cheneryand his colleagues attemptedto
refocus the field on the need to allocatea greatershareof investmentgainsinto formsthat would benefit thepoorest groups n society.
But, as KorzeniewiczandSmith(2000:30)explain,"Noclear andgen-erally acceptedframeworkhad been establishedas yet for theseconcerns;
... the Bank'sstaffwas only weaklycommitted o the McNamaraagenda,and even the institution'sknowledgeaboutpoverty, inequality,and pos-sible policieswas limited."Thus,despiteinternalreevaluationswithin the
development orthodoxy,the black box of the inverted U-curveremainedclosed. The theoretical rameworkunderlyingRedistributionwithGrowth
was not so mucha refutationof the "iron aw,"but a tacit acceptanceofit. Since income distribution hows a naturaltendencytoward increased
polarizationn the firststagesof development,Chenery'sgroupwas left to
arguethatpolicycouldmake a difference n speedingthe Kuznetsprocess
along (Bourguignon, 996;Fishlow,1996).In a separateseries of papersbyMontekAhluwaliaand othersat the WorldBank(1974,1976a,b;Ahluwalia
et al., 1979),often reprintedandwidelycited throughout he field,the U-
curvehypothesiswas stillunderstood o be a centralconstruct.Regardingthe policy implicationsof this series, Anand and Kanbur(1993b:20)say,"Inan influential iterature, hese have become influentialpapers."Basedon this research,the World Bank adopted the invertedU-relationshipin the late 1970s and early 1980sto make projectionsof inequalityand
povertyto the year2000,andpublishedthese forecasts n the 1978, 1979,and 1980 WorldDevelopment Reports (Anand and Kanbur, 1993b).
Lackingthe means for effectiveimplementationas well as experienc-
ing "oppositionto its aims by a broadrange of actors,"the McNamaraagenda was rapidly droppedin subsequentWorld Bank administrations
(KorzeniewiczandSmith,2000:31).Worsestill,the very conceptof devel-
opmentas an intellectualandpoliticalprojectwasin deep crisis.The fieldfacednot only increasingly ocal criticisms romthe old guard (Bhagwati,1989;Lewis, 1984;Sen, 1983;Stern,1989;andthe variouscontributionsnMeierandSeers,1984),but also causticattacks romprofessedcriticswithtitles such as The Poverty of "Development Economics" (Lal, 1985), and
Development Economics on Trial (Hill, 1986). While some were character-
izingthe field as being merelyat an "impasse" Booth, 1985;Schuurman,1993;Sklair,1988;VandergeestandButtell,1988),others, ncludingpromi-nent formerpioneers,werewritingobituariesof the developmentalproject
(Hirschman,1981;Seers,1979).In short,the old consensuswas"witnessingits progressivedissolution"Escobar,1995:94).What came to be known as
project n what became a motivateddestructionof all thatthe developmen-talregimehadcome to be (ArrighiandSilver,2000;Leys,1996;Toye,1993).As ArrighiandSilver(2000:21)argue,"Thiswasno Kuhnian cientificrev-
olution. Itwasa strictlypoliticalcounterrevolution." ustasthe birthof de-velopment nthe 1940sdid not resultfrom a naturalprocessof scientificad-
vancements,neither was its death hastenedby theoretical, nstitutional,or
allydealt withthem.Both wereradical,politicalresponsesby powerfulac-torsin the capitalistworld-economy.By theendof the decade,the researchandpracticeof developmenteconomicsquicklywent "frompromised andto wasteland" ArrighiandSilver,2000:11),until the fieldbecame"aruinin the intellectual
andscape"Sachs,1990:2),a "mined,
unexplorableand"
(Esteva, 1992:22).Theoriginsandprocesses nvolved n the neoliberalcounterrevolution,
andthe consequentcollapseof the postwardevelopmentregime,have notbeen the subjectof muchscholarly nquiry,and the full explanation s ob-
viouslybeyondthe rangeof this study.8Scholarshave largelyfocused onthe effectsof the transformation, nd for our more limitedpurposeshereit is enoughto analyzehow changes n the broaderapparatusn whichtheU-curvepreviously it (inthis case thecompleterejectionof the apparatus)
producedchanges n thewaythe U-curvewasinterpretedn theoryandap-pliedinpractice.Thedevelopmentalparadigmhatemerged n thepostwarperiodhadformed,as Escobar(1995:40)describes,a "system hat allowedthe systematic reationof objects,concepts,andstrategies .. [and]definedthe conditionsunderwhichobjects,concepts, heories,andstrategiescan be
incorporatednto the discourse."Thecounterrevolution estroyed hissys-tem, allowing or the reinterpretationf its componentparts-the openingof previouslyclosedblack boxes.
QuestioninganEmpiricalRegularity
Beginning in the 1980s, scholarsbegan findingdifficultieswith the
overwhelmingempiricalsupportthe U-curve received in the 1960s and
8Therehavebeen a few attempts o explain he counterrevolution. rrighiand Silver(2000)attribute he events to a concertedpoliticalmovement, ed by a radicalchangein UnitedStatesmilitaryand financialpolicies,to respondto the hegemoniccrisis of the 1970s and
(re)consolidate he establishedhierarchyof globalwealth. Cockett(1995)emphasizes heroleof a handfulof economists n slowlydevelopingneoliberal/anti-Keynesianoctrineuntilmaterial orces(similar o thoseemphasized y ArrighiandSilver,2000)changed n theirdi-rection.Similarly klair 1997:524) rgues hatthe counterrevolution as a "socialmovement
for capitalism"edbya "transnationalapitalist lass."
1970s.Thisempiricalncriminationwasbasedon two interrelated ritiquesthatcombined o revealthe U-curveto be a sort of "empiricalmirage"-aself-contradictory atternoften visible in cross-section(and with certain
data and methods) while simultaneouslydisappearing ongitudinally orwith other data and methods). One line of argumentfocused on the
reliabilityof the cross-sectionaldata itself in light of wide intercountryvariation in data-collectionmethods, and possible measurement errors
resultingfrom systematicreportingbiases (e.g., Gagliani,1987;Nugent,1983).As a group,the inequalityestimates assembled n the 1970swereassociatedwith largeinconsistenciesn variabledefinitionsand collection
techniquesrenderingcomparabilitybetween countries(and often within
countriesover time) suspectat best. Plagued by severe limitations n the
documentationof secondarysources and accuracyof the data,inequalityestimateswere oftenextrapolatedrom nformedguessesbased on national
accounts, or establishedusing surveys conducted on nonrepresentativesubsets of the national population (e.g., taxpayers,or urban residents
only). On manyoccasions, t was not altogetherclear who conducted the
originalsurveyor how it wasactuallyadministered.9 aith(1983:367)usedAhluwalia's(1974) material to arguethat the inverted U-curve hasonlybeen confirmedbecauseof statisticaldifficulties, otingthat "evenmarginal
variations"n availabledatasets ead to the "virtualdisappearance"f theinverted U-curve. Fields (1984a) conducted a similarreevaluationusingPaukert's 1973)data,andagainvoiced considerabledoubtconcerning he
commonalityof the U-curvefindings.Another criticaltrackconcerned the use of cross-sectionalresearch
relativelysimple,cross-sectionalmethods that entailed"(1)measuring he
degree of inequality n each (country),(2) measuringother characteristics
of each country(e.g., level of GNP, rate of growth, mportanceof agricul-ture in total product,etc.), and (3) relatingthe level of inequality o that
economy'scharacteristics,singcorrelationor regressionanalysis" Fields,1980:59).The idea was to take theseplace-in-time napshots,almostalwaysdepictingan invertedU growth-equity attern,and infer that the countries
9Overtime these datasetsgrewintojumbledconfigurationsf different ypesof aggregatedestimates.Importantdefinitionaldifferencesbecameblurredor forgottento the point of
Paukert's 1973:124)widelyused dataset s basedon a compilation f 44 countriesproducedby Adelman and Morris"as set forthin variouspapersby these authors."He then deletesfourcountries n whichdata were "so bad that theywereunsuitable," eplaces3 countrieswith "superiordata from othersources,"and adds information or 19 countries romnewsources.
forming he curvilinear hapein time t, got to these positionsby followingthe pathof the U-curve from some timeto.10
Theseprocedurescame underheavycriticismn the 1980s and in time
became the targetof one of the mostdamaging ritiquesof theU-curvehy-pothesis (Datta, 1986;Fields, 1980, 1984a;Gagliani,1987;Papanek,1978;
Saith,1983;Stewart,1978;Sundrum,1990).The argumentwas that while
an inverted U-curvemight accuratelydescribe the growth-equitytrajec-tory of the three nations includedin Kuznets's ongitudinalanalysis,and
even thoughsuch a curvemightalso be observed in cross-sectionaldata,the actualpathof changefor most countriesmightnot be accurately epre-sentedby the U-curvemodel.As Gagliani(1987:323) xplains,"Theavail-
able cross-sectional cattermight perfectly
fit anincreasing
50 line whileeach countryis silently moving along a 450 decreasingone." Or as Tilly
(1984:35)arguesmorebroadly,"There s no logicalconnectionbetween the
cross-sectional elationshipover time." Kuznetshimselfwarned n the late
1960s:
The valueof such(cross-section)vidence orgeneratingomepreliminaryunchescannotbe denied.Butunless nnovationalchangescan somehowbe taken onto ac-count n the use of cross-section ataproper,useof itsresultsmay eadtoerroneousinferencesconcerningpast changes n structuren the processof growth.And thesameapplies,paripassu, o application f cross-section nalysis o projectionsntothe future. Kuznets,1966:436)
When the blackbox wasbeingclosed,thesecritiqueswereoverlooked,often mentioned n footnotes and asasides,and modalities urroundinghe
U-curvecontinuedto be deleted.Not until the social andpoliticalclimateshifted did these samecritiquesbecomeaccepted; heywere moved out offootnotes and into centralarguments,and modalitiesbegan to be addedback on. This suggests,as the constructivist chool claims,that the valueand meaningof empiricalevidence is not objectivelygiven but in some
waysis alsocontingentuponthe time-and-place ontext withinwhichsuchevidence is beingproducedandnegotiated.
Shifting Theoretical Discourse
Remarkableshifts in the broader social and politicalcontext led tointellectual shifts in the ways inequalitywas being conceptualized, hus
bringingchangesto the realizationof the U-curve and its academicand
policy relevance. New debatesemergedoutsideof economics aroundtherelativeimportanceof social andpoliticalfactorseffectinggrowth-equityprocesses.Some investigatedthe impact of differentialaccess to educa-
tion (e.g., Meyer, 1977; Milner,1987; Teachman,1983).A different line
of interpretationemphasizedthe importanceof nondemocratic orms ofpolitical rule in promoting greater inequality (e.g., Muller, 1985, 1988,1989;Simpson,1990; Stewart,1978).This perspectivewas challengedbya groupof scholarsanalyzing he impactof democracyon inequality(e.g.,
developedaroundthe issue of economicdependencyand a country'spo-sition in the structureof the world-economy.Some arguedthat the dy-namics of "dependent development,"most often operationalizedas the
degree of multinationalcorporation penetrationor the amount of for-
eign direct investment,promotes inequalityin noncore economies (e.g.,Bollen andJackman,1985;Bornschier,1983;BornschierandBallmer-Cao,
1979;Bornschierand Chase-Dunn,1985; Chase-Dunn, 1975;Evans and
Timberlake,1980; Rubinson,1976).11World-systems cholarsquestionedthe utilityof conceptualizing rowthandinequalityas a national-levelphe-nomena in the first
place, arguinghat the axial division of labor between
coreandperipheral ones of thecapitalistworld-economywasthe key pro-cess generatingandreproducing nequalities.
In the shiftingacademic andscape, nequalitywas no longer conceptu-alized as a blackbox simplistically ndmechanicallygeneratedby growthprocesses; t was a socialproblemthat itself needed to be theorized.Re-
gardingthe interpretationof the U-curve,this opened to questionmore
specifically he twinmetaphorsof the U-curveargumentas developedin
theprevious wodecades: 1) that a growth-equityradeoff s economically
necessaryatlow levels of development; nd(2) thatgrowthprocesseseven-tuallyreducethisinequalityat advanced evels of development.
The firstof these theoreticalchallengeswasgeneratedbythediscoveryof theEastAsiandevelopmentexperience. n1979,Feiet al.published heirinfluential volume Growth withEquity: The Taiwan Case, a study that raised
the lid on what would later be called "theEast Asian Miracle" Birdsallet al., 1995;Findlayand Wellisz, 1993;WorldBank, 1993).The develop-mentexperiences n East Asia wereparticularlyhallengingo the U-curve
hypothesis.In the last half of the twentiethcentury,these were the only
poor countriesactually"developing"o any significantandlastingdegree,and they did so via dualisticrural to urbanindustrialization.Yet the de-
velopmentalexperienceof East Asia largelychallenged he theoreticalas-
sumptionsKuznetsmaderegardingdualisticdevelopmentand showedwhy,the U-curveprocessfar frominevitable,could be altogetheravoidable.
UnderlyingKuznets'soriginalanalysiswas an importantpresumptionthat the industrial ector is the only dynamicsectorin the economy.Con-
traryto what Kuznets(andother scholarsof that time) thought, develop-mentin East Asia showed that intersectoral ransitions ouldlead to rising
11Yetthis argument was actively disputed by others (Weede, 1980;Weede and Kummer, 1985;Weede and Tiefenbach, 1981; and later by Crenshaw, 1992; Firebaugh, 1992; and Simpson,1990).
incomesin the agriculturalector andfallingurban-ruralnequalities.Feietal. (1979)foundthemostimportant ontributiono Taiwan's"equitywith
growth"phenomenonwas a more labor-intensive gricultural utputmix,
combinedwitha tremendousexpansionof ruralnonagriculturalctivities.These factorstogether disprovedKuznets'sassumptionsby demonstratingthat: (1) A shift from agriculturalo nonagricultural ctivities "does not
have to entail a shift from a moreequaldistributiono a less equaldistri-
bution if outputmixes and technologiesin the nonagriculturalector are
becoming ncreasinglyabor-intensive, bsorbing he poorestand landless
farmers" Ranis, 1996:51);and (2) income levels in the ruralsector could
improvewithgrowth(as a consequenceof risingproductivityand increas-
inglabor-intensiveechnologies).WhatFei etal. (1979)found in Taiwanbecameincreasingly onfirmed
whenthebroader"miracle"wasanalyzedasa moreregionalphenomenon.In six of the Asian economiesexperiencing apideconomicgrowthduringthe last halfof the twentiethcentury-Indonesia, Japan,Korea,Malaysia,Thailand,and Taiwan-the agriculturalectorsweresubstantial ndhighly
of Green Revolutiontechnology,highinvestments n rural nfrastructure,and limited taxation of agriculturemeantthat rural ncomes andproduc-tivityrose morerapidly n East Asia thanin otherregions.As a result,ru-ral incomesactually mprovedwith industrialgrowth,and urban-rural n-come differentialswere smaller n EastAsia thanin mostotherdevelopingeconomies.13
The East Asian experiencewas not kind to the laborsurplusmodel,nor the conceptsof dualisticdevelopmentand the growth-equity radeoff
upon whichit was based.Theoretically, he conceptof inequalitybecame
inverted.Thinkingof inequalityas a necessaryandunavoidable ide-effectof growthgives wayto a transposedrelationshipwhere initial distribution
of assets determines oris at least apreconditionof) futureratesof growth.Insteadof generatingwealthaccumulation, s in the dualisticmodels,in-
equality s now seen as growth-adverse.Various theoriesnow seek to cap-ture thisphenomenon,whetherby emphasizinghe unequaldistribution f
12HongKongandSingaporeare two countrieswithinthe "EastAsian miracle" hat almost
entirely ackagriculturalectors.
13Accordingo Turnham1993; eproducednWorldBank,1993:34), etween1965and1988,agriculturalncomesgrewat anaverageannual ateof3.2% nEast Asia(comparedo 2.4%in SouthAsia, 2.3% n LatinAmerica,and 1.9% n sub-SaharanAfrica),andproductivityroseby2.2%onaverageannually comparedo 0.6% nSouthAsia,1.55%n LatinAmerica,and 0.3% n sub-Saharan frica).
Bennett,1988).While there is some debate on the extent to whichthisphe-nomenon extends beyond the United States and Great Britain,the idea
thatequality
could worsen ndeveloped
economiesundermines he secondKuznetsmetaphor,that "the dynamismof a growingand free economic
society" s fundamentally ndunendingly galitarian.The theoreticalargumentsused to explainthis effectemphasizeshift-
ing patterns n the location anddynamicsof industrialproductionand theterms of international rade. Morebroadly,so the consensusgoes, the dy-namicsof technological hangehave resulted nthe"deindustrialization"f
wealthycountriesand the simultaneous iseof a new skill-driven conomyin whichhigh-value nformationand technologicalservicesreplace high-
volumemanufacturings the leadingeconomic sector.This shiftin leadingsectorshasdrastically lteredthe relativedemand or andsupplyof skilled
and unskilledworkers,generatingdownwardpressureson thewagesof the
unskilled,while dramaticallyncreasingreturnsto the skilled,thus creat-
ing a newly polarizedearningsdistributionBluestoneandHarrison,1982;Bound andJohnson,1992;Katz andAutor,2000).
PART 3: U-CURVE(S) IN THE TWENTY-FIRST CENTURY
(This) study's indingsoffer broadsupport or Kuznets'shypothesis.
-Randolph and Lott(1993)
The empirical status of the Kuznets hypothesis in ... multicountrycontextsremainsunsettled.
-Ram (1997)
The Kuznetscurve s fiction.
-Rodrik (1998)
What are we to make of Kuznets'sU-curvehypothesis today?Have50 years of continuousintellectualattention resultedin any substantive
trating"necessities on the "tortuouspath"of development-that is, until
the long-run optimismof the U-curvecan be evoked: "It has long been
acceptedthat economicdevelopmentworsens ncomedistribution, t leastin the early stages.... After a certainpoint, the relationshipbetween de-
velopmentand equalityis a win-win situation,but untilthen, results canbe frustratingat best." Yet elsewhere the World Bank (2000:15),once adevout adherentof the U-curve,now concedes that "[e]vidence rom re-cent decades has not validated[Kuznets-based heories],and it now ap-pears likely that growth, equality,and reductions n povertycan proceedtogether."
Finding he EmpiricalU-Curve
Despite the empiricalcriticisms n the 1980s,quantitativeassessments
of Kuznets'shypothesishave continued uninterruptedand undeterred.
Notwithstandinghe sheer volume of articlesaddressing he inverted U-
curve(or perhapsbecause of it), researchers ontinue to produceambigu-ous empirical indingsandcontradictory ssessments.Somearecontinuingthe traditionof Fields(1980)and Robinson(1976) by attemptingo model
the "functionalorm"of Kuznets'shypothesis, ncludingeconometriccon-ditions for a turning point in the inequality-developmentrelationship(Anad andKanbur,1993a;Galor andTsiddon,1996;Glomm andRaviku-
mar, 1998;andRam,1995).Manymoreapplyvariousregression echniques
to assortedcross-nationalamples.Ofthese,Dovring(1991),IADB (1998),Jha (1996),Randolphand Lott (1993),Milanovic(1994),Nielsen (1994),andNielsenandAlderson(1995)continueto advocatethe existenceof the
ables, and regressiontechniques) determinethe findings.The invertedU-curverelationships inconsistentwithrespectto econometricspecifica-tion, and statistically ignificantU-curve-relatedparameterestimatescan
still be derivedin variouscross-national amples.For exampleOgwang's
(1995) supportfor the U-curvevariesby both the inequalityand devel-
opment indicatorsused. Deininger and Squire (1996:571)conclude that
analysesof past aggregationsof data,where comparisonsare made with-
out regardto consistentdefinitions,"couldlead to virtuallyany type of
growth-equitypattern."Examplesof this effect in the U-curveliterature
includethe appearanceof largedecreases n inequality n Kenya (Bigsten,1986)andthepresenceof a U-curveeffectinMalaysia Meesook,1975)and
Korea(Kwack,1990).Lastly,muchasin the olddatasets, he latest datacontinues o illustrate
an inverted U growth-equitypatternin cross-section-currentinequalitylevels are usually owerat higherlevels of incomeandhigherinequalitiesare moreprevalent n LatinAmericaand otherpartsof the middle-income
range.Thepersistenceof thecross-nationalU-curvecan be attributedo the
cal inequality evels) that are significantly orrelatedwith current ncomelevels. Thus,authorshave made the cross-sectionalU-curve"disappear,"or at least becomeless robust, n econometricmodelsby simply ncluding
DeiningerandSquire,1998;Ram,1997).14Instudieswherecountry-specificdummyvariablesare addedto models,the U-curvenot only goes away,but
in some cases is even reversed.Higginsand Williamson 1999:10) eem to
agree with this line of interpretation, tatingthat over time "theKuznetscurvedisappearedfromcross-section)whendummyvariables or Asia andLatin Americawere added."
Yet whiledifferenteconometricapproachesandmethodological pec-ificationscan lend various evels of supportto the hypothesisas variouslydefined, the overwhelmingempiricalevidence today does not point to
the existenceof a meaningful nverted U-curvepatterneitherin termsofa by-productof dualisticdevelopmentor simplya pictureof risingthen
falling nequalitiesover time
(Anadand
Kanbur,1993b;Bruno et
al., 1998;Deiningerand Squire,1996, 1998;Kim,1997;Li et al., 1998;Lipton,1997;
Ram, 1997;Ravallion,1995).In terms of empiricalgenerality, he mostro-
bust findingemergingfrom today's quantitative iterature s that no sys-tematic relationshipexists between averageincome levels and/or subse-
quent growthand incomeinequality,either for nations on averageor even
withinaggregatedncomegroupings.As opposedto followingapredictablepattern,trends in the contemporaryworld are characterizedby tremen-dous heterogeneityof growth-equity experience. Deininger and Squire
(1998:261,282) review their comprehensivedatabaseof inequalityesti-matesandconclude hatinthefewcountrieswherea significantinear rendin inequalitycan be detected,"itcontradicts he Kuznetshypothesisalmostas often as confirmingt."In a recent New York Timesarticle,GaryFieldsstates the prevailingview that "the Kuznets's curveis neither a law norevena central endency.15 hepattern s that there isno pattern" Krueger,2002).Insum,the U-curveas acontemporarympiricalandpolicyheuristicis best characterizedby Gustav Ranis (1996:50),who declares,"As I readthe evidence,it suggeststhat it is time to give a decent burial to that fa-
mous 'law,'which was actuallyadvancednot by Kuznets,who was muchtoo cautious,butby Kuznetsians,who were not."
The U-Curveas a Theoretical-Historical tatement
As Ranis implies, Kuznets's formulation of the U-curve hypothe-sis was always more historicallygrounded, sociologically sophisticated,
14This s accomplished via a series of country-specific dummy variables added to the traditional
quadraticmodel. The argument is that these dummy variables, while preserving the commonKuznets structure, allow inequality to differ across countries that are at the same level of
development.150f course, the fact that the inverted U-curve is being discussed in the New York Times in
2002 is itself a testament to the uncommon impact of Kuznets's hypothesis over time.
tribution s not rich with theoretical nsight.As inequalityreemerges n the
social sciences as an importantsocialproblem,the analyticalvalue of theU-curve lies more in its theoretical constructsthan its empirical expec-tations.For example,the cases of Korea and Thailandprovideevidence
thatpoliticalmobilizationof the urbanworkingclassmayhaveproceeded
along patternssimilarto those Kuznetswouldhave predicted(albeit at a
much fasterrate).Conceiqaoand Galbraith 2001:139,160)eek a "macro-economic alternative" o currentinequalityliteratureand use Kuznets-
type methodsto explain risinginequality n the developedworld, in the
tion within a Schumpeterianrameworkandexplorehowthe fundamental
insightsof his researchhelp explainworld-historicalpatternsof inequal-
ity.As Stallingsetal. (2000:106)argue,"Kuznets,ndeed,mayhave had the
economicsof thestoryrighteventhoughempiricaloutcomesacrossregionsandcountriesand withincountriesover time follow no neatpattern."
Recenthistoricalresearchhas also vindicatedKuznets's nsightsabout
the long-term egalitarianeffects of industrialization, specially in the
United Statesand Great Britain.Economic historiansresearchinghe his-
toricaldynamicof the U-curve from the eighteenth centuryonwardhavefound Kuznets'shypothesizedpatternsto be reasonablyaccurate. The
most notable workin this area has been conductedby JeffreyWilliamson
(1991a,b,1997) and Lindert and Williamson(1985). Analyzing the best
available historicaldata on inequalityin Great Britain and the UnitedStates,they findstrongvalidationfor the descendingarc of the U-curve,and some (admittedlymorefragile)evidencefor the early upswing.More
broadly, hey findthat,with the possibleexceptionof Germany, hereis a
cleartrend towardconvergingncomesin the twentiethcentury or a largenumberof industrializedountries.16
As Williamson 1991a:34)s quickto point out, however,"no unam-
biguoustheory of the Kuznets Curveemergesfrom... history."The his-
toricalevidenceisadmittedly ragile,
andeven whereKuznets-type urves
16Note,however, hatwhile Williamson indssupport or Kuznetscurves,he does not fullysupportKuznets's xplanation f whytheyexist.Whilehis basicmodelhasmuch n commonwith Kuznets's ramework, e tends to downplay he social-political actorsthat Kuznets
arguedwouldaccount or the long-run galitarianendency.
lutionsin the earlynineteenthand twentiethcenturies.But Kuznetsalwaysinsistedthathiswasa historicallybondedconclusion,nota forecast hat the
generalpatternwillbe presentwherever ndustrializationrocessesoccur,or wheneverthey take place. As Williamson(1991b:60)argues,"Lavish
attentionto the Kuznetscurvehas tended to deflect our attentionawayfrom assessingthe distributionalmpact of each country's idiosyncraticpolicies and institutions... [that account] for historicaldepartures romthe Kuznetscurve. The issue is not so much whether the Kuznets curveexists in history,but ratherto understand he forces which accountfor its
presenceor absence."
CONCLUSION: MPLICATIONSFORTHE SOCIOLOGY
OF SOCIAL-SCIENTIFIC NOWLEDGE
In theprecedingpages,I analyzethe historyof anextraordinaryocialscientificargument,and one of the most influential tatementson inequal-ity ever made-Simon Kuznets's nvertedU-curvehypothesis.The historyof Kuznets'stheory is one of shifting interpretationsand appropriationsthroughtime, demonstrating longthe way how the applicability f a setof argumentschangedwith the social and politicalcontextsin which in-
equalitywas conceptualized.A set of conjectures hat the authorhimself
claimedwerebased on "95%speculation"becamea reifiedsocioeconomiclaw, a closed "blackbox,"at a time when the U-curveprovidedan aes-
thetically elegant, empiricallygrounded synthesisof the broad academicand policyconsensusthatprevailed n the developmentalist ra. As such,
the U-curvebecamea foundation tatement n this consensusas anempiri-cal, theoretical,andpoliticalstatementof fact. A widely acceptedargumentthen became contentious,and its existence was paradoxically hallenged
by the veryevidencepreviouslyused to support t, when this larger appa-ratuswas destroyed.In a climateof intellectual ragmentation nd critical
empirical nquiry, he black box was opened.But whatemergedwasnot a
meaningless,discredited heory,butmanyversionsof the "U-curve," ach
interpreted o meansomethingdifferent han the others.That the U-curve
is stillvigorouslydebated and relevant odayreflectsnot onlythe powerof
old blackboxes to remaincompelling,but also the theoreticaland method-
ologicalrichnessof Kuznets'soriginalstudy.The
historyof the U-curve
hypothesishas
implicationsor those in-
terestedin the constructionandprogressionof social-scientific nowledge.
Although, as Randall Collins commentsin the preface to Fuchs'sbook
(1992:xv),the sociology of science has been "in flux" since the 1990s,two general frameworkshave emergedin the discipline: hose imposinga Kuhn-basedapproachand those considering he constructionist spectsof knowledgecreation.Duringthe 1960s and 1970s,Kuhn's(1962)model
wascertainly n vogue, and social scientistsbeganto incorporatehis ideas
despitehis explicitstatementthat social sciencehad not reached the stage
of paradigmaticcience. Hall (1993),for example,contends that Britain's"policyparadigm"ransformation nderThatcheroccurred hrougha pro-cess of "puzzling ut" andpolicy experimentationhat wenton until it pre-
cipitated"a shiftin the locus of authorityoverpolicyandinitiate[d]a wider
the Kuhnianmodel has beenwidelycitedbut,as Babb(2001:3)contends, t
begs an importantquestion:"[I]fneoliberalism works'so well,whywasit
not implemented orty yearsbefore?"This is preciselythe point of departurefor the second general ap-
proach,one which emphasizesthe social constructionof knowledge.AsYonay (1998:18) xplains, he goalis "notto showhow 'good'theorieswin,but to documenthow the view of what s 'good' sbeingconstitutedand then
used to resolve debates aboutNature."Scholars n this traditioncontend
thatthe old debate betweenthe internalistsand externalistsn the studyof
scienceis obsoletebecause the elements involvedneverappearas "purelyinternal"or "purelyexternal."They do not deny the role of objectively
gained empiricalevidence,or that there can be good and bad arguments,but they understand hese processesto be contingent.The rise and fall of
the invertedU-curvehypothesis avorsa constructionisteadingof the his-
tory.Intheend,theconceptualizationf theU-curveovertime was as much
aproductof thelargeracademicandpoliticalclimatesas it was a productof
the original deasthemselvesor the empirical"puzzling ut" of evidence.
Thusfar,mostconstructivist esearchhasconcerned he "micro"work-
ingsof scientific esearch-historicalstudiesandethnographic bservationsof the strugglesamongworkingscientists hat drawattentionto the social
processes that shape the content of theirwork (see Knorr-Cetina,1991,Latour,1987; he contributions n Pickering,1992).Yet the storyof the U-curvesupportsrecentefforts to expandthe constructionist pproach romone that focuses on micro-levelanalyseswithin the scientificcommunityto one thatbetterunderstands he sociologyof social-scientific nowledgewhere outcomesmustbe explainedbywidersocial andpolitical actors hat
historicallycondition its production.This approachwould seek to prob-lematizeboth sides of thistwo-waystreet: hemechanismsbywhichsocial-
politicalcontexts affect the
productionof social-scientific
knowledgeand
the waysinwhichthisknowledge s simultaneously ppropriatedndinter-
pretedin social-politicalcontexts.Indeed,this is aweighty ntellectual ask,but it is in thissense thatthe vein of recent researchcarriedout, forexam-
ple, by Babbs(2001),Escobar(1995), RueschemeyerandSkocpol (1996),andYonay(1998)shouldbecome morecommonplace.
ACKNOWLEDGMENTS
I amgrateful o JamesRule, IanRoxborough,GiovanniArrighi,andPatricioKorzeniewicz orprovidingvaluable ntellectualnsightsatvarious
stagesof thisproject,with the usual caveat thattheybearno responsibilityforthe endproduct. wantalso to thankthe anonymous eviewers ortheircarefulreadingof an earlierversion,and especiallythe editor and staffof SociologicalForum,for the time andenergy they spenton thispaper'sbehalf.
REFERENCES
Adelman, Irma, and Cynthia Taft Morris1973 Economic Growthand SocialEquity
in Developing Countries.Stanford,CA: StanfordUniversityPress.
1978 "Growth and impoverishment inthe middle of the nineteenth cen-
tury." World Development 6:245-273.
Adelman, Irma, and Sherman Robinson1989 "Income distribution and develop-
ment." In Hollis Cheneryand T. N.Srinivasaneds.),Handbookof Devel-
Bank Conference on DevelopmentEconomics 1995. Washington,DC:WorldBank.
Bruno, Michael,MartinRavallion,and Lyn
Squire1998 "Equity and growth in developing
countries: Old and new perspec-tives on the policy issues." In UiroTanzi and Ke-YoungChu (eds.), In-come Distributionand High-QualityGrowth: 117-146. Cambridge,MA:MITPress.
London:OxfordUniversityPress.Chenery,Hollis,and LanceTaylor1968 "Development patterns: Among
countriesand over time." Review ofEconomicsand Statistics 0:391-416.
Cockett,Richard1995 Thinking the Unthinkable: Think-
Tanks and the Economic Counter-Revolution, 1931-1983. London:
HarperCollins.Conceigao,Pedro,and JamesK.Galbraith2001 "Toward a new Kuznets hypothe-
sis: Theory and evidence on growthand inequality."In James K. Gal-braith and Maureen Berner (eds.),Inequalityand IndustrialChange:AGlobal View: 139-160. Cambridge,UK:CambridgeUniversityPress.
Crane,Diana1972 Invisible Colleges: Diffusion of
KnowledgenScientificCommunities.
Chicago:Universityof ChicagoPress.
Crenshaw,Edward
1992 "Cross-nationaldeterminantsof in-comeinequality:A replication ndex-tension using ecological-evolutionarytheory." Social Forces 71:339-363.
1996 "Inequality, poverty, and growth:Where do we stand?" In MichaelBruno and Boris Pleskovic (eds.),Annual World Bank Confer-ence on Development Economics1995:25-39.Washington,DC: WorldBank.
FitzGerald,E. V. K.1991 "Kurt Mandelbaumand the classi-
cal tradition in development the-
ory." In Kurt Martin (ed.), Strate-
gies of Economic Development:Readings in the Political Economyof Industrialization.New York: St.Martin's.
Australia n the nineteenthcentury."In Y. S. Brenner,HartmutKaelble,andMarkThomas eds.),IncomeDis-tribution in Historical Perspective:149-173.Cambridge,UK:CambridgeUniversityPress.