20.-Labor-supply-of-New-York-City-cab-drivers-one-

LaborSupplyofNewYorkCityCabdrivers:OneDayataTimeColinCamerer;LindaBabcock;GeorgeLoewenstein;RichardThalerTheQuarterlyJournalofEconomics,Vol.112,No.2,InMemoryofAmosTversky(1937-1996).(May,1997),pp.407-441.StableURL:=0033-5533%28199705%29112%3A2%3C407%3ALSONYC%3E2.0.CO%3B2-JTheQuarterlyJournalofEconomicsiscurrentlypublishedbyTheMITPress.YouruseoftheJSTORarchiveindicatesyouracceptanceofJSTOR'sTermsandConditionsofUse,availableat://@jstor.org.:492007LABORSUPPLYOFNEWYORKCITYCABDRIVERS:ONEDAYATATIME*Life-cyclemodelsoflaborsupplypredictapositiverelationshipbetweenhourssuppliedandtransitorychangesinwages.WetestedthispredictionusingthreesamplesofwagesandhoursofNewYorkCitycabdrivers,whosewagesarecorrelatedwithindaysbutuncorrelatedbetweendays.Estimatedwageelasticit-iesaresignificantlynegativeintwooutofthreesamples.Elasticitiesofinexperi-enceddriversaverageapproximately-1andarelessthanzeroinallthreesamples(andsignificantlylessthanforexperienceddriversintwoofthreesam-ples).Ourinterpretationofthesefindingsisthatcabdrivers(atleastinexperi-encedones):(i)makelaborsupplydecisionsonedayatatimeinsteadofintertemporallysubstitutinglaborandleisureacrossmultipledays,and(ii)setaloosedailyincometargetandquitworkingoncetheyreachthattarget.Dynamicmodelsoflaborsupplypredictthatworkhoursshouldrespondpositivelytotransitorypositivewagechanges,asworkersintertemporallysubstitutelaborandleisure,workingmorewhenwagesarehighandconsumingmoreleisurewhenitsprice-theforgonewage-islow(e.g.,LucasandRapping[1969]).Whilethispredictionisstraightforward,ithasproveddifficulttoverify.Estimatedelasticitiesofintertemporalsubsti-tutionhavegenerallybeenlowandinsignificant,orevennega-tive,whethertheyarebasedonaggregate[Mankiw,Rotemberg,andSummers19851,cohort[Browning,Deaton,andIrish19851,orpanel[Altonji19861data(seealsoLaisney,Pohlmeier,andManythankstoBruceSchaller(NYCTaxiandLimousineCommission)fordataandhelpfuldiscussions;JamesChoi,KimMorgan,andDovRosenbergforresearchassistance;CharlesBrown,JeffreyDominitz,JohnEngberg,JohnHam,SethSanders,andLowellTaylorforhelpfuldiscussions;AndreiShleiferfortheWeberquote;tworeferees,andeditorLawrenceKatzforextraordinarilythoroughcomments,andcolleaguesatmanyworkshops:TheCaltechbrownbaglunchseminar,NBERBehavioralLaborEconomicsmeeting,andBehavioralEconomicsSummerCamp(sponsoredbytheRussellSageFoundation),CarnegieMellonUni-versity'sHeinzSchoolofPublicPolicyandManagement,theUniversityofCali-fornia(IrvineandBerkeley)DepartmentsofEconomics,theMITmarvardBehavioralEconomicsseminar,theUniversityofChicagoLaborWorkshop,theJudgmentDecisionMakingSociety,andtheEconometricSocietymeetings.Ad-dresscorrespondencetothefirstauthoratDivisionofSocialSciences228-77,CaliforniaInstituteofTechnology,PasadenaCA91125,camererQhss.caltech.edu.01997bythePresidentandFellowsofHarvardCollegeandtheMassachusettsInstituteofTechnology.TheQuarterlyJournalofEconomics,May1997.408QUARTERLYJOURNALOFECONOMICSStaat[19921;Pencavel[19861;andcf.Mulligan[19951).However,theseresultsaredifficulttointerpretbecauseactualwagechangesarerarelytransitory,sothehypothesisofintertemporalsubstitutionmustbetestedjointlyalongwithauxiliaryassump-tionsaboutpersistenceofwageshocks,formationofwageexpec-tations,etc.Asaresult,thefrequentlyobservednegativewageelasticitiescanplausiblybeattributedtospecificationerror.Theidealtestoflaborsupplyresponsestotransitorywageincreaseswoulduseacontextinwhichwagesarerelativelyconstantwithinadaybutuncorrelatedacrossdays.Insuchasituationalldynamicoptimizationmodelspredictapositiverela-tionshipbetweenwagesandhoursworked,duetothenegligibleimpactonlife-cyclewealthofaone-dayincreaseinwage(e.g.,MaCurdy11981,p.10741).Suchdataareavailableforatleastonegroupofworkers:NewYorkCitycabdrivers.Driversfacewagesthatfluctuateonadailybasisduetodemandshockscausedbyweather,subwaybreakdowns,day-of-the-weekeffects,holidays,conventions,etc.Althoughratespermilearesetbylaw,onbusydaysdriversspendlesstimesearchingforcustomersandthusearnahigherhourlywage.Thesewagestendtobecorrelatedwithindaysanduncorrelated