What Use is Economic Theory?

1 WhatUseis EconomicTheory? CaliforniaatBerkeleyAugust, examinehowneoclassicaleconomictheoryis alsodescribewhatI IsEconomicsBecominga HardScience? 29--30 October,1992,Paris, (inFrench) Cartelier, EconomieDevient-ElleUneScienceDure? worthwhilethingtodo? isa well-constructedeconomicmodelhasanaesthe ticappealwell-capturedbythefollowingline sfromWordsworth: Mightyis thecharmOftheseabstractionstoa mindbesetWithimages,andhauntedbyherselfA ndspeciallydelightfuluntomeWasthatclears ynthesisbuiltupaloftSogracefully. Noonecomplainsaboutpoetry,music,numberth eory,orastronomyasbeing useless, thinkthatonecouldarguea ,whenpressed,mosteconomictheoristsadmitt hattheydoeconomicsbecauseit is thinkpurelyaestheticconsiderationswouldn otprovidea is notjustanintellectualpursuitforitsownsak e,butit apolicyscienceand,assuch, policysciencePartoftheattractionandthepr omiseofeconomicsisthatit claimstodescribepoliciesthatwillimprovep eoples policycomponent, ,ofcourse,hasthepotentialtoimprovepeople s standardsofliving,butthisis reallya , mistaketocompareeconomicstophysics;a ,it isa mistaketocompareeconomicstobiology.

2 A livesbetter; :wevaluewhatis policysubjects ---engineering,medicine,ordentistry---is muchconcernedaboutmethodology,andeconomi sts,byandlarge,aren ,itisquitesurprisingthatthereisn , ,aspracticedinphilosophydepartments,seem stobebasicallyconcernedwithphysics,witha smatteringofphilosophersconcernedwithpsy chology,biology,anda ,philosophyofmedicineorphilosophyofdenti stry---thereisnomodelofmethodologyfora ,inmyview, policyscienceGivenmyviewthateconomicsisa policyscience,ifI wanttodefenda practiceineconomics,thenI mustdefendit froma nodoubtincomplete,andI canhelpfocussomediscussiononwhyeconomist sdowhatthedo, ,forexample,wewanttodeterminehowa marketpricewillrespondtoa regressionofmarketpriceagainsttaxrates, ;taxesjustdon peopleonlycareaboutthetotalpriceofa good,inclusiveoftax---atheory---thenweca nuseestimatedpriceelasticitiestoforecast theresponseofpricetotheimpositionofa theoryaboutbehavior---peoplewillrespondt otheimpositionofa taxinthesamewaythattheyrespondtoa , AchosenwhenfA;Bgis availableandB is chosenwhenfB;Cgis available,thenwecanpredictAwillbechosenw henfA;Cgis certainlya theoryaboutbehavior.

3 It wehaddataonchoicesbetweenallpairsofA,B,a ndC,thenthetheorywouldn ;Cgwewouldsimplylookathowthepersonchosep reviously---thatis, !Butwerarelyobserveallpossiblechoices;ty picallyweobserveonlya ;B;Cgexamplebruteinductionrequiresobserv ing3allchoicestheconsumercouldmakefromth evariouspropersubsetsavailable,whichrequ ires3 theassumptionoftransitivityholds,then2 is duetothefactthatyoucanexplainit toa Congressmaninsixminutesandhecantalkabout it marginaltaxrateof50%requiresa laborsupplyelasticityof1 ;withoutthetheory, ,if , takeit asgiventhatriskis a bad. Thereforewhenwealthgoesup, , benefitsandcostsI afundamentaloneineconomics,andwouldbever ydifficulttousewithouta theoreticalmodelofchoice.

4 A , themodelofbehaviordoesnotapply, ,thenwecangetanideaofwhatis beingoptimizedbylookingatagents ,onemayneeda ,onemaywanttoshowthatonekindofmarketstru cture,ortaxstructure,maydoa betterjobofsatisfyingconsumers ,toindividualchoice,tosocialobjectivesan dsocialchoiceis commontomany, onedescribesa modelina purelymathematicalway,it oftenhappensthattheunderlyingequationswi lldescribea good canbeinterpretedasa physicalcommodityavailableatdifferenttim es,locations, modelofintertemporaltrade,location, formalanalysisoftheproblemofsecond-degre epricediscriminationshowsthatit is ,auctiondesign, irrelevant detailsarestrippedaway,itsbecomesapparen tthatthesameessentialchoiceproblemis , subsidizingborrowing.

5 Doesit? thetaxbracketsarethesame,forexample, , simpleinsight,butit wouldbeverydifficulttounderstandwithouta theorythatis wrongcanstillyieldinsightPurecompetition is certainlya wrong theorymanymarkets;puremonopolyis a notwhetherornota theoryis truebutwhetheritoffersa examinea true. Therearea hostofreasonswhyconvertinganapartmenttoa ,it focusesourattentionona methodforsolvingproblemsI takethemethodofneoclassicalmicroeconomic stobe1)examineanindividual soptimizationproblem;2)lookattheoptimale quilibriumconfigurationofindividualchoic es;3) maybethatthespecificphenomenonunderexami nationisnotfruitfullyviewedasanoutcomeof optimizing, : Moretrutharisesthrougherrorthanconfusion .

6 Methodologicalindividualismisa limitedwayoflookingattheworld, probablydoesn is , ,forexample, thepriceofgasolineincreasesby25%,a ,it isperfectlypossiblethatmostpeoplewouldn ; , thepricewerea littlehigherora littlelower,theresultsofyouragonizingmig htbedifferent,andthisis shardtoconvincea laymanoftheadvantagesoffreetradesinceit iseasytoseewherethedollarsgo, ;buttheyrarelyencountertheirowncountry thirdexampleis thebiasinperceptionsofinflation:pricemov esareperceivedtobeexogenousfromtheviewpo intoftheindividual, pricesandwagesmoveupbythesameamount, obviousmayshowthatit isn tOneofthecriticismsthateconomistshavetod ealwithisthattheyspenda titobviousthatdemandcurvesslopedownandsu pplycurvesslopeup?

7 Maybeobviousthatdemandcurvesslopedown--- butasthetheoreticalanalysisshows,it ispossibletohavedemandcurvesthatdon mightbepossible---afterall,farmerswouldn tadvancetheclaimunlessit ,theoryshowsusthatthisparticularclaimcan notbetrueif two-personzero-sumgameshowsthatimproving yourbackhandintennismayleadtoyourusingit publicoffertomatchanycompetitor spriceisa competitor spriceis a highlycompetitivetacticcaneasilybeviewed asa , Whenyoucannotmeasureit,whenyoucannotexpr essit innumbers,yourknowledgeis ofa meagreandunsatisfactorykind. 1 Theoreticaleconomicsgivesusa ,butit , verylittlecomputationinsociology,politic alscience, usefulsinceyoucanuseit talwaystherightanswers---thatdependsonwh etherthemodelyouhaveis right.

8 (Or,atleast,whetherit is goodenoughforthepurposesathand.)Buta desideratumofagoodmodelis thatyoucancomputewithit:themodelcanbesol vedtodeterminesomevariablesasa ,it is ,a highlyprolificelectricalengineer,oncegav emesomeexcellentadviceabouthowtowritea ,1 However,thesamepoetwhosepraiseforabstrac tionandsynthesisI quotedintheintroductionalsooncesaid: :::HighHeavenrejectstheloreofnicelycalcu latedlessormore. ,I havetriedtofollowthisadvice,with,I think, amenabletoexperimentalverificationBecaus eneoclassicaleconomicmodelsenablesonetoc omputeanswerstoproblems,it , ,suchassupplyanddemand, ,suchasexpectedutility, tobeexpected---iftherewerenosurprisesfro mexperiments,theywouldn ,concreteandtestable,ratherthantheoriest hatarecomplex,abstract, , ,theywillalsobecomebetteratidentifying naturalexperiments havearguedthatinordertowhyeconomictheori stsbehaveinthewaytheydoonehastounderstan dtheroleofeconomictheory fundamentallya