Vernon Smith An Experimental Study of Competitive Market弗农史密斯市场竞争的.ppt

Vernon Smith An Experimental Study of Competitive Market弗农史密斯市场竞争的.ppt

《Vernon Smith An Experimental Study of Competitive Market弗农史密斯市场竞争的.ppt》由会员分享，可在线阅读，更多相关《Vernon Smith An Experimental Study of Competitive Market弗农史密斯市场竞争的.ppt（13页珍藏版）》请在冰豆网上搜索。

VernonSmith:

AnExperimentalStudyofCompetitiveMarketBehaviorEconomics328Spring2005HowtoWinaNobelPrizein3EasySteps.Somepapersareimportantasmuchforwhattheystartedasforwhattheyare.Smiths1962paperonmarketsisanarchetypicalexampleofthis.Itwasntthefirstpapereverwritteninexperimentaleconomics.Thurstonehadwrittenaninfluentialexperimentalpaperonindifferencecurvesin1933andChamberlin,Smithsadvisor,wroteanimportantexperimentalpaperonmarketsin1948.However,thesepaperswerelargelyisolatedevents.AfterSmithswork,experimentaleconomicsbecameanongoinglineofresearchwithineconomics.“ColumbusisviewedasthediscovererofAmerica,eventhougheveryschoolchildknowsthattheAmericaswereinhabitedwhenhearrived,andthathewasnoteventhefirsttohavemadearoundtrip,havingbeenprecededbyVikingsandperhapsothers.WhatisimportantaboutColumbusdiscoveryofAmericaisnotthatitwasthefirst,butthatitwasthelast.AfterColumbus,Americawasneverlostagain.”AlRothChristopherColumbusResearchQuestionSmithsexperimentsweredesignedtostudytheneo-classicaltheoryofcompetitivemarkets.Thisisthesimplemodelofsupplyanddemandcurvesthateveryeconomicsstudentlearnsinthefirstfewlectureofprinciples.Inspiteoftheimportanceofthecompetitivemodeltoeconomics,therewaslittledirectevidencepriortoSmithsworkthatthetheoryactuallywouldwork.Fielddataistoodynamictoseeifequilibriumisbeingachieved.Chamberlins（1948）earlierworkusingadecentralizedmarketmechanismfoundthatgenerallythepricesweretoolowandthevolumestoohighascomparedtocompetitiveequilibriumpredictions.Smithsexperimentsweredesignedtogivethetheoryitsbestchancetoworkthisreflectsadesiretoestablishiftherewereanycaseswerethemarketwouldequilibrateaspredictedbythetheory.Smithwasinterestedinstudyingwhatconfigurationsofsupplyanddemandweremost（orleast）likelytoleadtoequilibrium,andwasalsointerestedinthedynamicsthatledtoequilibriumExperimentalDesignandProceduresSessionswereruninclassroomsforhypotheticalpayoffs.Theresultswerelaterreplicatedusingmonetarypayoffs.Inonesessionthesubjectsweregraduatestudentsinaclassoneconomictheory.InevaluatingSmithswork,itisimportanttorememberthathewasfirst.Subjectsweredividedintotwogroups,buyersandsellers.Toinducedemandandsupplycurves,eachbuyerwasgivenacardwithhis/hermaximumwillingnesstopayandeachsellerwasgivenacardwithhis/herminimumreservationprice.Eachsubjectwasallowedtobuy/selloneunitofthegoodpertradingperiod.Tradingtakesplacethroughadoubleoralauction.UnlikeChamberlinsearlierwork,thisisahighlycentralizedmarket.Buyersandsellersareawareofallbid,asks,andtransactionprices.SummaryofSessionsThesessionsfocusprimarilyontheeffectofvarytheshapeofthesupplyanddemandcurves.Someattentionisalsopaidtotheeffectofchangingmarketinstitutionsandmakingtradersmoreexperienced.Test1:

BasicsupplyanddemandTests2and3:

Variessteepnessofsupplyanddemandcurveswithoutchangingequilibriumprice.ThisallowsSmithtostudytheprocessthatleadstoequilibrium.Test4:

Flatsupplycurve.Thisleavesnosurplusforthesellers.Itisinterestingtothinkofthisinlightofmuchlaterexperimentsonequityinmarketgames.Test5:

Studiestheeffectofanincreaseindemand.Giventhatsubjectsdontknowhowdemandhaschanged,youmightexpectthistodisruptconvergence.Test6:

Equilibriumgivesaverylargesurplustothesellersbyusingasupplycurvethatgoesvertical.Onceagain,thisisinterestinginlightofthelaterexperimentsonmarketgames.Test7:

Verysteepsupplycurverelativetothedemandcurve.Test8:

Buyerswerenotallowedtomakebidsinearlyperiods.Thiswassupposedtosimulateretailmarkets.Thequestioniswhetherthispreventsconvergencetoequilibrium.Test9and10:

Eachindividualisallowedtomaketwotransactions,doublingtheamountofexperiencereceived.Thisisexpectedtospeedconvergence.InitialHypothesesMarketsareexpectedtoconvergetothecompetitiveequilibrium.Accordingtothe“Walrasian”hypothesis,thisconvergenceshouldbefasterwhenthereislargerexcessdemand（orsupply）.So,forexample,convergenceshouldbefasterinTest2thaninTest3.The“excessrent”hypothesisfocusesonunrealizableprofitsatapricethehighertheseare,thefasterpricesshouldadjust.Historically,thishasnotbeenanimportanthypothesis.Allowingmoreexperienceshouldspeedupconvergence,whilechangingmarketinstitutionsshouldslowconvergence.ResultsConvergencetoEquilibriumThedoubleoralauctionstendtoconvergestronglytowardsequilibriumandachievehighlevelsofefficiency.Forexample,theresultsforTest1areshowntopright.Thisistrueevenwheneitherdemandorsupplyshiftsovertime.SeeTest5results,bottomright.Thisisaverybasicresult,butisthemostimportantresultinthepaper.Competitivemarketequilibriumisacentralconceptineconomictheory,butgenerallycantbeobservedinthefield.Theseresultsprovethatcompetitiveequilibriumcanwork（butnotthatitmustwork）.ResultsUnevenSplitsofSurplusTest4andTest7bothfeatureuneven（predicted）splitsofthetotalsurplusbetweenbuyersandsellers.Thesesessionsareamongtheworstintermsofconvergencetoequilibrium.Test4priceswereconsistentlyaboveequilibrium,givingsomesurplustothesellers.Test7,whichisntquiteasextremeasTest4,onlyconvergesveryslowlytoequilibrium.Pricesareconsistentlytoolow（comparedtocompetitiveequilibrium）givingsomesurplustobuyers.Theseresultscanbeviewedasaprecursortothesortsofresultsonfairnessthatwehavestudiedextensively.ResultsSpeedofConvergenceComparingTests2and3,thesupplyanddemandcurvesareflatterinTest2thaninTest3.Thismeansthatasmallchangeinpricesfromtheequilibriumleadstolargerexcessdemand（amountdemandedamountsupplied）inTest2theninTest3.Ifthespeedofadjustmentisrelatedtothesizeofexcessdemand,weshouldseefasteradjustmentinTest2.Thisisexactlywhatisobservedinthedata.SmithfindsmoresupportfortheexcessrenthypothesisthanfortheWalrasianhypothesis.Thisresultislargelyofhistoricalinteresttheconvergenceresultsabovearetheimportantresults.ResultsInceasingSubjectsExperienceTests9Aand10replicateTest7,butallowtraderstousetheirvalues/coststwiceineachperiod.Bydoublingsubjectsexperience,thespeedofconvergencewillbeincreased.Thisappearstobetrue,especiallyifyoulookatTest10.Thisprobablyisnttherightwaytotestthishypothesisthoughwhatyoureallywantissubjectswhohavealreadybeeninonemarketexperimenttoparticipateinasecondexperimentwithdifferentdemandandsupplycurves.ConclusionsWhileSmithdrawsmanyconclusionsfromthedata,themostimportantconclusionisthemostbasicone:

“Themoststrikinggeneralcharacteristicoftests13579,and10istheremarkablystrongtendencyforexchangepricestoapproachthepredictedequilibriumforeachofthesemarkets.”ItmustberememberedthatSmithdesignedhisexperimentstogivethetheoryitsbestchance.Theseexperimentsdontestablishthatingeneralthetheoryofcompetitiveequilibriumwillhavemuchpredictivepower.Onamoregenerallevel,thispaperplayedanimportantroleinillustratinghowcontrolledlaboratoryexperimentsleteconomistsunderstandphenomenaandtheoriesthatwerehardtoobserveinthefield.Inthefield,oneneverknowswhattheunderlyingsupplyanddemandcurvesare,soyoucannevertrulyknowthatthecompetitiveequilibriumhasbeenachieved.Inthelab,youcandirectlyobservetheemergenceofequilibrium.FurtherResearchonMarketsSmithsgeneralresultsonconvergencehavebeenreplicatedmanytimes.DOAmarketsareremarkablygoodatconvergingtoequilibriumunderawidevarietyofcircumstances.Thisremainstrueevenifsupplyanddemandcurvesareshifting（althoughrandomshiftsdoworsenconvergence）.Convergencecanbequitesensitivetomarketinstitutions.Seeminglysmallchangesintherulesforqueuingcansubstantiallyaffectthespeedofconvergence.Largerchangesintheinstitutionssuchasusingapostedpricemarketcangreatlyslowconvergence,evenleadingtonon-convergenceinsomecases.Marketpowerhasmixedeffectsonconvergencetoequilibrium.Holt,Langan,andVillamil（1986）findpricesthataresignificantlyaboveequilibriumwhensellershavemarketpower,butothershavefoundlittleeffectinDOAmarkets.Moregenerally,theimpactofmarketpowerisgoingtodependonthegamebeingplayed.Whenasingleindividualcaneasilymanipulatemarketpricesorwheninstitutionsreducecompetitionamongsellers,departuresfromequilibriumaremorelikely.Gametheorydoesafairlygoodjobofpredictingwhendeparturesfromequilibriumarelikely,althoughitdoesapoorjobofpredictingthesizeofthesedepartures.NobelPrizeWinnersoftheFuture?