搜索
1TimeDiscountingandTimePreference:ACriticalReviewShaneFrederickSloanSchoolofManagementMassachusettsInstituteofTechnologyGeorgeLoewensteinDepartmentofSocialandDecisionSciencesCarnegieMellonUniversityTedO’DonoghueDepartmentofEconomicsCornellUniversityJanuary30,2002Acknowledgements:WethankJohnMcMillan,DavidLaibson,ColinCamerer,NachumSicherman,DuncanSimester,andthreeanonymousrefereesforusefulcomments.WethankCaraBarber,RosaBlackwood,MandarOak,andRosaStipanovicforresearchassistance.Forfinancialsupport,FrederickandLoewensteinthanktheIntegratedStudyoftheHumanDimensionsofGlobalChangeatCarnegieMellonUniversity(NSFGrantSBR-9521914),andO’DonoghuethankstheNationalScienceFoundation(AwardSES-0078796).Mail:ShaneFrederick/SloanSchoolofManagement/MassachusettsInstituteofTechnology/RoomE56-329,38MemorialDrive,Cambridge,MA02142-1307;GeorgeLoewenstein/DepartmentofSocialandDecisionSciences/CarnegieMellonUniversity/Pittsburgh,PA15213-3890;TedO’Donoghue/DepartmentofEconomics/CornellUniversity/414UrisHall/Ithaca,NY14853-7601.Email:shanefre@mit.edu;GL20@andrew.cmu.edu;andedo1@cornell.edu.Webpages:mitsloan.mit.edu/facstaff/;sds.hss.cmu.edu/faculty/loewenstein.html;and—decisionsinvolvingtradeoffsamongcostsandbenefitsoccurringatdifferentpointsintime—areimportantandubiquitous.Suchdecisionsnotonlyinfluenceanindividual’shealth,wealth,andoverallhappiness,but,asAdamSmithfirstrecognized,mayalsodeterminetheeconomicprosperityofnations.Inthispaper,wereviewempiricalresearchonintertemporalchoice,andpresentanoverviewofrecenttheoreticalformulationsdesignedtoaddresstheinsightsgainedfromthisresearch.Economists'attentiontointertemporalchoicebeganearlyinthehistoryofthediscipline.NotlongafterAdamSmithcalledattentiontotheimportanceofintertemporalchoiceforthewealthofnations,theScottisheconomistJohnRaewasalreadyexaminingthesociologicalandpsychologicaldeterminantsofthesechoices.InSection2,webrieflyreviewtheperspectivesonintertemporalchoiceofRaeand19thandearly20thcenturyeconomists,anddescribehowtheseearlyperspectivesinterpretedintertemporalchoiceasthejointproductofmanyconflictingpsychologicalmotives.AllofthischangedwhenSamuelsonproposedthediscounted-utility(DU)modelin1937.DespiteSamuelson'smanifestreservationsaboutthenormativeanddescriptivevalidityoftheformulationhehadproposed,theDUmodelwasacceptedalmostinstantly,notonlyasavalidnormativestandardforpublicpolicies(e.g.,incost-benefitanalyses),butasadescriptivelyaccuraterepresentationofactualbehavior.AcentralassumptionoftheDUmodelisthatallofthedisparatemotivesunderlyingintertemporalchoicecanbecondensedintoasingleparameter—thediscountrate.InSection3,weexaminethis,andmanyotherassumptionsunderlyingtheDUmodel.Wedonotpresentanaxiomaticderivationofthemodel,butinsteadfocusonthosefeatureswhichhighlighttheimplicitpsychologicalassumptionsunderlyingthemodel.Samuelson'sreservationsaboutthedescriptivevalidityoftheDUmodelwerejustified.Section4reviewsthegrowinglistof“DUanomalies”—patternsofchoicethatareinconsistentwiththemodel’stheoreticalpredictions.VirtuallyeveryassumptionunderlyingtheDUmodelhasbeentestedandfoundtobedescriptivelyinvalidinatleastsomesituations.Moreover,aswediscussattheendofthesection,theseanomaliesarenotanomaliesinthesensethattheyareregardedaserrorsbythepeoplethatcommitthem.Unlikemanyofthebetter-knownexpected-utilityanomalies,theDUanomaliesdonotnecessarilyviolateanystandardorprinciplethatpeoplebelievetheyshoulduphold.3Theinsightsaboutintertemporalchoicegleanedfromthisempiricalresearchhaveledtotheproposalofnumerousalternativetheoreticalmodels,whichwereviewinSection5.Someofthesemodifythediscountfunction,permitting,forexample,decliningdiscountratesor“hyperbolicdiscounting.”Othersintroduceadditionalargumentsintotheutilityfunction,suchastheutilityofanticipation.StillothersdepartfromtheDUmodelmoreradically,byincluding,forinstance,systematicmispredictionsoffutureutility.ManyofthesenewtheoriesrevivepsychologicalconsiderationsdiscussedbyRaeandotherearlyeconomists,butwhichbecameextinguishedwiththeadoptionoftheDUmodelanditsexpressionofintertemporalpreferencesintermsofasingleparameter.InSection6,wereviewattemptstoestimatediscountrates.WhiletheDUmodelassumesthatpeoplearecharacterizedbyasinglediscountrate,thisliteraturerevealsspectacularvariationacross(andevenwithin)studies.Thefailureofthisresearchtoconvergetowardanyagreeduponaveragediscountratestemspartlyfromdifferencesinelicitationprocedures.Butitalsostemsfromthefaultyassumptionthatthevariedconsiderationsthatarerelevantinintertemporalchoicesapplyequallytodifferentchoicesand,thus,thattheycanallbesensiblyrepresentedbyasinglediscountrate.Throughoutthepaper,westresstheimportanceofdistinguishingamongthevariedconsiderationsthatunderlieintertemporalchoices.Wedistinguishtimediscountingfromtimepreference.Weusethetermtimediscountingbroadlytoencompassanyreasonforcaringlessaboutafutureconsequence,includingfactorsthatdiminishtheexpectedutilitygeneratedbyafutureconsequence,suchasuncertaintyorchangingtastes.Weusethe