Ray Palgrave Development Economics文献综述Word文件下载.docx
《Ray Palgrave Development Economics文献综述Word文件下载.docx》由会员分享,可在线阅读,更多相关《Ray Palgrave Development Economics文献综述Word文件下载.docx(32页珍藏版)》请在冰豆网上搜索。
March2007.PreparedfortheNewPalgraveDictionaryofEconomics,editedbyLawrence
BlumeandStevenDurlauf.
1Introduction
WhatweknowasthedevelopingworldisapproximatelythegroupofcountriesclassifiedbytheWorldBankashaving“low”and“middle”income.Anexactdescriptionisunnecessaryandnottoorevealing;
sufficeittoobservethatthesecountriesmakeupover5billionofworldpopulation,leavingouttheapproximatelyonebillionwhoarepartofthe“high”incomedevelopedworld.Together,thelowandmiddleincomecountriesgenerateapproximately6trillion(2001)dollarsofnationalincome,tobecontrastedwiththe25trilliongeneratedbyhighincomecountries.Anindexofincomethatcontrolsforpurchasingpowerwouldplacetheselatternumbersfarclosertogether(approximately
20trillionand26trillion,accordingtotheWorldDevelopmentReport(2003))buttheper-capitadisparitiesarelargeandobvious,andtothoseencouteringthemforthefirsttime,stillextraordinary.
DevelopmentEconomics,asubjectthatstudiestheeconomicsofthedevelopingworld,hasmadeexcellentuseofeconomictheory,econometricmethods,sociology,anthropology,politicalscience,biologyanddemographyandhasburgeonedintooneoftheliveliestareasofresearchinallthesocialsciences.MylimitedapproachinthisbriefarticleisoneofdeliberateselectionofafewconceptualpointsthatIconsidertobecentraltoourthinkingaboutthesubject.Thereaderinterestedinamorecomprehensiveoverviewisadvisedtolookelsewhere(forexample,atDasgupta(1993),Hoff,BravermanandStiglitz(1993),Ray(1998),BardhanandUdry(1999),MookherjeeandRay(2001),andSen(1999)).
Ibeginwithatraditionalframeworkofdevelopment,onedefinedbyconventionalgrowththeory.Thisapproachdevelopsthehypothesisthatgivencertainparameters,saysav-ingsorfertilityrates,economiesinevitablymovetowardssomesteadystate.Iftheseparametersarethesameacrosseconomies,theninthelongrunalleconomiesconvergetooneanother.Ifinrealityweseeutterlackofsuchconvergence—whichwedo(see,e.g.,Quah(1996)andPritchett(1997))—thensuchanabsencemustbetracedtoapresumptionthattheparametersinquestionarenotthesame.Totheextentthathis-toryplaysanyroleatallinthisview,itdoessobyaffectingtheseparameters—savings,demographics,governmentinterventionism,“corruption”or“culture”.
ThisviewisproblematicforreasonsthatIattempttoclarifybelow.Indeed,thebulkofmyessayisorganizedaroundtheoppositepresumption:
thattwosocietieswiththesamefundamentalscanevolvealongverydifferentlines—goingforward—dependingonpastexpectations,aspirationsoractualhistory.
Now,afterapoint,thedistinctionbetweenevolutionandparameterisasemanticone.Bythrowingenoughstatevariables(“parameters”)intothemix,onemightarguethatthereisnodifferenceatallbetweenthetwoapproaches.Formally,thatwouldbecorrect,butthen“parameters”wouldhavetobeinterpretedbroadlyenoughsoastobeoflittleexplanatoryvalue.Ahistoricalconvergenceandhistoricallyconditioneddivergenceexpresstwofundamentallydifferentworldviews,andthereislittlethatsemanticjugglerycandotobringthemtogether.
2DevelopmentFromTheViewpointofConvergence
Whyaresomecountriespoorwhileothersarerich?
Whatexplainsthesuccessstoriesofeconomicdevelopment,andhowcanwelearnfromthefailures?
Howdowemakesenseoftheenormousinequalitiesthatwesee,bothwithinandacrossquestions?
These,amongothers,arethe“bigquestions”ofeconomicdevelopment.
ItisfairtosaythatthemodelofecononomicgrowthpioneeredbyRobertSolow(1956)hashadafundamentalimpacton“big-question”developmenteconomics.Fortheory,calibrationandempiricalexercisesthatbeginfromthisstartingpoint,see,e.g.,Lucas(1990),Mankiw,RomerandWeil(1992),Barro(1991),ParenteandPrescott(2000)andBanerjeeandDuflo(2005).Solow’spathbreakingworkintroducedthenotionofconvergence:
countrieswithalowendowmentofcapitalrelativetolaborwillhaveahighrateofreturntocapital(bythe“law”ofdiminishingreturns).Consequently,agivenadditiontothecapitalstockwillhavealargerimpactonper-capitaincome.Itfollowsthat,controllingforparameterssuchassavingsratesandpopulationgrowthrates,poorercountrieswilltendtogrowfasterandhencewillcatchup,convergetothelevelsofwell-beingenjoyedbytheirrichercounterparts.Underthisview,developmentislargelyamatterofgettingsomeeconomicanddemographicparametersrightandthensettlingdowntowait.
Tobesure,savingsanddemographyarenottheonlyfactorsthatqualifytheargument.Anythingthatsystematicallyaffectsthemarginaladditiontoper-capitaincomemustbecontrolledfor,includingvariablessuchasinvestmentin“humancapital”orharder-to-quantifyfactorssuchas“politicalclimate”or“corruption”.Afailuretoobserveconvergencemustbetracedtooneoranotherofthese“parameters”.
Convergencereliesondiminishingreturnsto“capital”.Ifthisisourassumedstarting
2
point,theshareofcapitalinnationalincomedoesgiveusroughestimatesoftheconcavityofproductionincapital.Theproblemisthattheresultingconcavityunderstatesobservedvariationincross-countryincomebyordersofmagnitude.Forinstance,ParenteandPrescott(2000)calibrateabasicCobb-Douglasproductionfunctionbyusingreasonableestimatesoftheshareofcapitalincome(0.25),butthenhugevariationsinthesavingsratedonotchangeworldincomebymuch.Forinstance,doublingthesavingsrateleadstoachangeinsteadystateincomebyafactorof1.25,whichisinadequatetoexplainanobservedrangeofaround20:
1(PPP).Indeed,asLucas(1990)observes,thediscrepancyactuallyappearsinamoreprimitiveway,attheleveloftheproductionfunction.Forthesamesimpleproductionfunctiontofitthedataonper-capitaincomedifferences,apoorcountrywouldhavetohaveenormouslyhigherratesofreturntocapital;
say,60timeshigherifitisone-fifteenthasrich.Thisisimplausible.Andsobeginsthehuntforotherfactorsthatmightexplainthedifference.Whatdidwenotcontrolfor,butshouldhave?
Thisdescribesthemethodologicalapproach.Theconvergencebenchmarkmustbepittedagainsttheempiricalevidenceonworldincomedistributions,savingsrates,orratesofreturntocapital.Thetwowillusuallyfailtoagree.Thenwelookfortheparametricdifferencesthatwillbridgethemodeltothedata.
“Humancapital”isoftenusedasafirstportofcall:
mightdifferenceshereaccountforobservedcross-countryvariation?
TheeasiestwaytoslipdifferencesinhumancapitalintotheSolowequationsistorenormalizelabor.Usually,thisexercisedoesnottakeusveryfar.DependingonwhetherweconducttheLucasexerciseorthePrescott-Parentevariant,wewouldstillbepredictingthattherateofreturntocapitalisfarhigherinIndiathanintheU.S.,orthatper-capitaincomedifferencesareonlyaroundhalfasmuch(orless)astheytrulyare.Therestmustbeattributedtothatfamiliarblackbox:
“technologicaldifferences”.Thatslotcanbefilledinavarietyofways:
externalitiesarisingfromhumancapital,incompletediffusionoftechnology,excessivegovernmentintervention,within-countrymisallocationofresources,....Allofthese
—andmore—areinterestingcandidates,butbynowwehavewanderedfarfromtheoriginalconvergencemodel,andifatallthatmodelstillcontinuestoilluminate,itisbywayofoccasionalreturntotherecalibrationexercise,afterchoosingplausiblespecificationsforeachofthesepotentialexplanations.
Thismodelservesasaquickandreadyfixontheworld,anditorganizesasearchforpossibleexplanations.Takenwiththeappropriatequantityofsalt,andviewedasafirstpass,suchanexercisecanbeimmenselyuseful.Yetplayingthisgametooseriouslyrevealsaparticularworld-view.Itsuggestsafundamentalbeliefthattheworldeconomyisultimatelyagreatleveller,andthatifthelevellingisnottakingplacewemustsearchforthatexplanationinparametersthataresomehowstructurallyrootedinasociety.
3
Tobesure,theparametersidentifiedinthesecalibrationexercisesdogohandinhandwithunderdevelopment.Sodobadnutrition,highmortalityrates,orlackofaccesstosanitation,safewaterandhousing.Yetthereisnoultimatecausalchain:
manyofthesefeaturesgohandinhandwithlowincomeinself-reinforcinginterplay.Bythesametoken,corruption,culture,procreationandpoliticsareallupforseriouscross-examination:
justbecause“culturalfactors”(forinstance)seemsmoreweightyan“explanation”doesnotpermitustoassignitthestatusofatrulyexogenousvariable.
Inotherwords,theconvergencepredictedbytechnologicallydiminishingreturnstoin-putsshouldnotblindustothepossibilityofnonconvergentbehaviorwhenallvariablesaretreatedastheyshouldbe—asvariablesthatpotentiallymakeforunderdevelopment,butalsoasvariablesthatareprofoundlya