Upload
birte-groeger
View
218
Download
0
Embed Size (px)
Citation preview
7/29/2019 Behavioral Economics - Script Summary
1/53
BehavioralEconomics
Lecture1IntroductionBehavioralEconomicsTheoreticalEconomics=bodyofnormativetheoriesaboutrationalbehaviorofanidealizedindividualRationality=maximization(orminimization)ofagoalfunctionundercertainrestrictions
Person/household:utilityfunction(e.g.happiness) Firm:profitorlossfunction
Socialtheoryofnormativeeconomics=GameTheoryandrationalitydefinedbyanyofseveralequilibrium(ausgleichende)conceptswhichhavetobeselected
Descriptivemodels+empiricalevidence:behaviorinconsistentwithcanonic(traditional)model=irrationalbehavior Decisionsdependonreferencepoint Systematicbiasorerrorduetoheuristics
Experimentaleconomics:observedbehaviordeviatesfrompredictionsincanonicmodel(notplayingNashequilibriumstrategies)BehavioralEconomics
Howhumandecisionsmaysystematicallydepartfromthosepredictedbystandardeconomictheory
Extensionstoclassicaltheoriesincludesocial,reciprocal,altruistic,environmental,cognitivefactors
Cross-disciplinary,blendingneoclassicaleconomics(rationalchoicetheory),socialsciences,puresciences(biology..)
Collectionofdescriptivetheoriesofhumanbehavior Closelyakintocognitivepsychology Evolutionarypsychologyfordecisionmaking,becausecognitiveabilitiesof
altruismandnotmodernmarketexchange
1. Coreassumptionsofeconomicshomooeconomicus(HO):sophisticatedHOcanbesocial,cansolveeverywell-definedproblemincludingthousandsofgoods
andtechnologies,fullyinformed,perfectstatistician2. Simplifyingassumptionshomooeconomicussimplicissimus(HOS)goal=maximizingwealth,orconsumptionofmarketgoods=egoistic,notinterestedinenvironment,canbeendowedwithpreferencesandinformation
3. Socialplanner:notmaximizenationalincome,takeintoaccountmoraldistributionalandenvironmentalissues
BoundedRationality:HOSwithplainrationalbehavior
Describinglessperfectindividuals,limitedabilitiestocompute,limitedmemories Userulesofthumb,simpleheuristicstoachievesolutionsclosetooptimal
solutions Often:dreadfulmistakes
7/29/2019 Behavioral Economics - Script Summary
2/53
HigherRationality:HOwithhigherrationalityderivedfromsophisticatedmodelsandboundedrationality
Modesofbehaviorareintelligent Perfectlyadapted,rational,efficienttocopewithspecialenvironment Evolutionnotperfect:functionsthroughincrementalimprovements,alsomakes
mistakes Irrationalbehaviormayhavestrategicvaluechangeinenvironment,orpoint
ofview,maytransformsuboptimalmodeofbehaviortoahighlyefficientstrategy modesofbehaviorcircumventingdemands9ofinfinitelysophisticated
optimizationDutch-bookingbyintransitivepreferences,theoretical,notinreality
A>B,B>C,C>A A=B+e B=C+e 3efornothinginreturn,moneypump C=A+e
Lecture2EconomicsAxiomsandBehavior
EconomicAssumptionsandAxiomsProbability(Kolmogorov)Axioms
Probability(P)ofanevent(E)isnon-negative,realnumber Probabilityofsomeelementaryeventoccursis1,needtopreciselydefinewhole
samplespace Additivityofevents
7/29/2019 Behavioral Economics - Script Summary
3/53
ConsequencesandDeterministicTransitivityCombiningeventprobabilitiesDeterministicTransitivity(strong): P(A)>P(B)&P(B)>P(C)P(A)>P(C)DeterministicTransitivity(weak): P(A)P(B)&P(B)P(C)P(A)P(C)
Experimentalintransitivityduetoassumingprobabilisticchoices:ProbabilisticTransitivity P(A>B)>12&P(B>C)>12P(A>C)>12BayesTheorem
Fordecisionmaking,incorporatenewinformationinspecificevent Conditionalprobabilities(likelihood)ofoneeventgivenanotherevent Priorprobabilitycanbetransferredintoposteriorprobabilitybyincorporating
furtherknowledge
P(A)isthepriorprobabilityofAmeaningitdoesnottakeintoaccountany
informationaboutB. P(A|B)istheposteriorprobabilityofAgivenB. P(B|A)istheconditionalprobabilityofBgivenA. P(B)=P(B|A)P(A)+P(B|A)P(A)isthemarginalprobabilityofB,actingasa
normalizingconstant.Example:BaseRateNeglectcitywith100terrorist,1.000.000non-terroristsImaginethatall1,000,100peoplepassinfrontofthecamera.About99ofthe100terroristswill
triggeraringandsowillabout10,000ofthemillionofnon-terrorists.Therefore10,099peoplewillberungat,andonly99ofthemareterrorists.So,theprobabilitythatapersonwhotriggersthealarmisactuallyaterroristis99in10,099.P(terrorist)=0.0001P(alarm|terrorist)=0.99P(alarm)=0.010098 [P(alarm|terrorist)P(terrorist)+P(alarm|non-terrorist)P(non-terrorist)=0.99x0.0001+0.01x0.9999]P(terrorist|alarm)=(0.99x0.0001)/0.010098=0.0098039Theprobabilitythatwhenthealarmistriggeredthepersonisaterroristis0.98%.In99.02%ofthecasesthepersonisnotaterrorist( falsepositive).
Whatistheprobabilityofthealarmnotbeingtriggered,butthepersonpassingbeingaterrorist(falsenegative)?PreferenceofChoice
Preferences=fundamentalsofdecisions,describingourgoals Impossible:completeandconsistentsetofpreferences(not:irrational) Requirements:existenceandtransitivity Optimalx*fromasetofx,wherenothingisbetter Sometimesdifferentpreferencesdependingonpointoftime Makeassumptions:1.Specifyit,2.Testit,toexplainobservedbehavior Preferences=presentedbyutilityfunctionU(x)x1>x2ifU(x1)>U(x2)
o Cardinal(grundlegend)instrumento Ordinaldescription(Rangliste)ofsatisfactionorhappiness
7/29/2019 Behavioral Economics - Script Summary
4/53
SubjectiveExpectedUtilityTheory(SEU)integratesthesubjectiveexpectation(probability)andtheindividualutilityintoavaluetodescribepreferencesortopredictchoiceschoseaction(alternative,option)withhighestSEU
Multi-AttributeUtility(MAU)integratesthedifferentfeaturesofanoption.Itrepresentstheoverallutilityofthisoption.ProbabilitiesofSEUreplacedwithweightsrepresentingindividual,situationalpreferences
DominanceRuleisastrongerassumptionthanMAU.Itpredictsthatalwaystheoptionischosenwhichisbetterononeattributeandatleastasgoodonallotherattributes.AxiomsofExpectedUtility
Comparability=twooptionscanbecompared Transitivity Independence=therelationbetweejn2optionsisindependentoftheexistenceof
furtheroptionsXY[X,p;Z][Y,p;Z] Continuity=mustalwaysbeapthatmiddleoptionisindifferenttobothother
optionsforallXYZexistsapthatY~[X,p;Z,1-p]
RevealedPreferences=directtestsofutilitymodelsofpreferences
http://de.wikipedia.org/wiki/Revealed_Preference Utilityfunctionrationalizestheobservedbehavior(pt,xt)fort=1,...,Tifu(xt)
u(x)forallxsuchthatpt*xtpt*x.(u(x)rationalizestheobservedbehaviorifitachievesitsmaximumvalueonthebudgetsetatthechosenbundles)
Utilitymaximizationxtisdirectlyrevealedpreferredtox(xtRDx)o xtRDximpliesu(xt)u(x)o Localnosatiation:pt*xt>pt*ximpliesu(xt)>u(t)o Pt*xtpt*xxtisstrictlydirectlyrevealedpreferredtoxwritextPDx
SequenceofrevealedpreferencesxtPDxj,xjPDxk,...xnPDxxtisrevealedpreferredtox(xtRx) R=transitiveclosuretotherelationPDxtRximpliesu(xt)u(x)
GeneralizedAxiomofRevealedPreferences(GARP)
Ifxtisrevealedpreferredtoxs,thenxscannotbestrictlydirectlyrevealedpreferredtoxt(thatisxtisneverstrictlyinthebudgetsetwhenxsischosen)
xtRxsimpliesnotxsPDxtxtRxsimpliesps*xsps*xt.(Wenn ,danngiltnicht ).
WeakAxiomofRevealedPreferences(WARP)
Ifxtisdirectlyrevealedpreferredtoxs,thenxscannotbedirectlyrevealedpreferredtoxt(Wenn ,danngiltnicht )
7/29/2019 Behavioral Economics - Script Summary
5/53
ExamplesSocialExample:TheCommonsDilemma=herdsmandecisionaboutaddinganotheranimalwith2utilities
Positive:saleofadditionalanimal,utility=+1 Negative:overgrazing,butsharedbetweenallherdsman,utility=onlyafraction
of-1 Increasehisherdwithoutlimit-inaworldthatislimitedruin
LogicalExample:WasonSelectionTask=logicpuzzlewhichcardstoturntotesttruthofaspecificproposition
Demonstratingtheinconsistencyofapplyingthelogicalrulesbythepeoplewhentheproblemissetintwodifferentcontextsbutwithverysimilarconnectionbetweenthefacts
Ifacardshowsanevennumberononeface,thenitsoppositefaceshowsaprimarycolor
Choosecard8andbrowntotesttherule Aresponsewhichidentifiesacardwhichneednotbeinverted,oraresponsewhichfailstoidentifyacardwhichneedstobeinverted
arebothincorrect
BehavioralObservations(contradictingexpectedvalueorSEU)StPetersburgParadoxYouenteredinthePetersburgcasino.Yourentrancefeeis20dollarsineachgame.Inonegame,acoinisthrownrepetitivelyuntilitshowsnomorehead.Youwin2(^n)-20Dollars,ifntimesHeadappeared,thebankgetsineachgame20-2^ndollars.Youwillexperiencethatwiththisentrancefee,youareloosingmoney.Inreality,itisamathematicalfactthatyouareexpectedto
win:youwinwithprobability2^-ntheamountof2^nDollars.Summingoverallngivesaninfiniteexpectationforyourwin.Whateverentrancefeeyouwouldpay,youwouldwin.Theparadoxisthatthisdisagreeswithyourexperiencetoloose.(http://www.mathematik.com/Petersburg/Petersburg.html)
Gamewithcoins,entrancefee,winning2^(k-1)dollarsifthecoinistossedktimesuntilfirsttailappears
Expectedpayoutvalueis ,infiniteamountofmoney Rarelywinmuchbutwhenyouwinalotinthelongrunthanrepayhowever
muchmoneytheyhavealreadypaidtoplay
Accordingtotheusualtreatmentofdecidingwhenitisadvantageousandthereforerationaltoplay,youshouldthereforeplaythegameatanypriceifofferedtheopportunity,butwronginreality
ViolationsofIndependence
Decisionpeoplemakedependingonsetofchoicespeoplehave Alsodependoncommonconsequencesofthechoices
EllsbergParadox
Urncontaining30redballsandthesum60ofblackandyellowballs Choice1:GambleA($100drawredball)andGambleB($100drawblackball) Choice2:GambleC($100drawredoryellow)andGambleD($100drawblackor
yellow)http://de.wikipedia.org/wiki/Ellsberg-Paradoxon Theory:Ifdrawingaredballismorelikelythandrawingablackball,then
7/29/2019 Behavioral Economics - Script Summary
6/53
drawingaredoryellowballisalsomorelikelythandrawingablackoryellowball(preferGambleAtoGambleB,itfollowspreferringGambleCtoGambleD
Survey:violationoftheaxiomsofexpectedutility,mostpeoplestrictlypreferAtob,andDtoC
AllaisParadox
Reality:mostpeoplechose1Aand2B Theory:peopleshouldchose1Aand2Aor1Band2B Difficultieswithexpectedutilitytheoryalternatives
o Prospecttheoryo Rank-dependentexpectedutilityo Framingtheory(Identicalitemswillresultindifferentchoicesif
presentedtoagentsdifferently(i.e.asurgerywitha70%survivalratevs.a30%chanceofdeath))
Independenceaxiomofexpectedutilitymaybenotnecessary, Independenceaxiom:2identicaloutcomeswithinagambleshouldbetreatedas
irrelevanttotheanalysisofthegambleasawhole Butinthisgame:complementarities(choiceinonepartofagamblemaydepend
onthepossibleoutcomeintheotherpartofthegamble)disappointmentwhengettingnothinginGamble1B
Wedon'tactirrationallywhenchoosing1Aand2B,ratherexpectedutilitytheoryisnotrobustenoughtocapturesuch"boundedrationality"choicesthatinthiscasearisebecauseofcomplementarities
PapervonFlo:ProspectTheoryReasonBasedDM
Explainobservedviolationsbyextremenessaversionandtradeoffcontrast Describecompromiseandattractioneffects(attractivenessofanoptionenhances
ifitisanintermediateoptioninthechoicesetandisdiminishedifitisan
extremeoption) Oftenthecompositionofthealternativesmatters
7/29/2019 Behavioral Economics - Script Summary
7/53
PreferenceReversals
CertaintyEquivalent(CE)leadstootherresultsthanchoicebetweenprobableoutcomes(AllaisParadox)
Viloationstoutilitytheoryalsobypresentationformatandformoftask WillingnesstoPay(WTP)contradictsWillingnesstoAccept(WTA)
o Willingnesstoaccept(WTA)istheamountthatpersoniswillingtoaccepttoabandonagoodortoputupwithsomethingnegative,suchaspollution
o Willingnesstopay(WTP)isthemaximumamountapersonwouldbewillingtopay,sacrificeorexchangeinordertoreceiveagoodortoavoidsomethingundesired,suchaspollution
Framing(Tversky,Thaler)AlsFraming-Effekt(deutschetwa:Einrahmungseffekt)wirddasPhnomenbezeichnet,dassunterschiedlicheFormulierungeneinerBotschaftbeigleichemInhaltdasVerhaltendesEmpfngersunterschiedlichbeeinflussen.BeiManahmenzurGesundheitsvorsorgewirdoftaufdieschdlichenFolgenetwavonRauchen,ungeschtztemSex,bergewichtusw.hingewiesen(vgl.Furchtappell).IndiesemFallsprichtmanvonVerlust-Framing(engl.lossframe).MehrErfolghabenjedochBotschaften,dieineinenGewinnrahmeneingebettetsind(engl.gainframe),alsodiepositivenFolgendergewnschtenVerhaltensnderunghervorheben.Willmanjedocherreichen,dassbereitsvorhandeneKrankheitenoderRisiken(Raucher,bergewichtige)mehrbeachtetwerden,sindBotschaftenimVerlust-Rahmeneffizienter.AuchbeiBefragungen,etwaanhandvonFragebgen,kanndieserEffektauftreten.DeshalbsolltederFormulierungeinerFragebesondereBeachtunggeschenktwerden,umdieErgebnissederBefragungnichtzuverzerren.Eswurdefestgestellt,dassbeiBefragungenwesentlichrisikoreicherentschiedenwurde,wennnurvonVerlustendieRedewar.WurdedagegenvonGewinnengesprochen,entschiedensichmehrBefragtefreinerisikormereLsung.
Example:2Programsreducingnumberofpeoplekilledintrafficaccident(from600) ProgramA:Reductionto570,costs$12Mio ProgramB:Reductionto500,costs$55Mio
Differentresponseswhengivingalternatives,orchangepeoplekilledinpeoplesaved$-versusp-betsExperimentsondecisionsunderrisk:Whichlotteryispreferable,HorL?H:8/9chanceofwinning$4,otherwise0.L:1/9chanceofwinning$40,otherwise0.MostsubjectspreferHtoL(71%).Askedwhattheywouldpayforparticipatinginoneofthegambles(WTP),mostsubjectsattachedahigherpricetoL(67%).
Priceofgamblesaremainlydeterminedbypayoffs($-bet) Choicebetweengamblesaremanlydeterminedbyprobabilitiesofwinning
(p-bet)
Payoffsareweighedmoreheavilyinpricesthaninchoice(scalecompatibility)
Lecture3BoundedRationalityandHeuristics
BoundedRationalitywithSatisficing
Boundedrationality=explainingbehavioraldeviationsfromthenorm Conceptofsatisficingdecisions:decisionshavetobesatisfyingandsufficient MostpowerfulprincipleofBR:imitationofsuccessfulbehavior
7/29/2019 Behavioral Economics - Script Summary
8/53
GeneralProblemSolver(computerprogramtobuilduniversalsolvermachine) Solvinganyformalizedsymbolicproblem Separationbetweenitsknowledgeofproblemsandstrategytosolveproblem Cognitiveprocessesassolvingprocesses Generatingheuristicsbymeans-endsanalysis(Zweckmittelanalyse) SolvingproblemsliketheTowersofHanoi(sufficientlyformalized) Performedpoorlyinanyreal-worldproblems
BoundedRationalityandProblemSolving
Thetwocentralproblemsofdecision-makingare: Limitedinformation Limitedresources(time,informationprocessingability)
Consequences: Maximizationisusuallyimpossible!Alternatives? Optimallearningisusuallyimpossible!Alternatives? Alternativesimplybiases,orcleverness
CostBenefitTrade-Off=economicprincipledescribingusageofinformationWeighingadvantages(profits)ofmoreinformationagainstcostsofdevelopingsensesandintellectualcapacity(e.g.inherithouse,collectinginformationonthequestionwhethertolivethere,sellit,lentit,leaveit)
SatisficingConcept
Satisficing=contrarytomaximizing
Limitedinfoprocessingabilities
preventiondeterminationofmaxima(prefersmalllossestoinfinitesearch) Goal:reachingasufficientlevelofutility,profit=aspirationlevel
7/29/2019 Behavioral Economics - Script Summary
9/53
Aspirationlevelsareoftenreadjustedaccordingtotheexperienceswhiletryingtoreachthem
Decidewhethertoworkfurtheronacertainproblem,giveup,orregardthecurrentsolutionassatisfactory
Satisfactory=dependingonenvironmentandknowledgeaboutvaluefunction Unknownprofitfunction,experiencemayguesshowflatthemaximumis,
whetherismuchtogaino HuntersProblem:Howgoodtoshoot?o FarmersProblem:Howmuchseed?o MonopolistsProblem:Whichprice?
Companiesuseseveralaspirations,correlated:profit,marketshare,shareprice,creditratingratinganddecisionsbasedonimportance,changes,dependence,partialrealizations
Goaldiversityexistsforindividuals,teams,governments
Example:10-weekjobsearchYouhavelostyourjob.In10weeksfromnowonyouhavenojobandnopayment(simplified),exceptyoufindanewjob.Everyweekyougetanewjobofferandtilltheendoftheweekyouhavetodecidewhethertoacceptit.Youareonlyinterestedinmoney,youareriskneutral,andyouknowthatthehighestofferofaweekisuniformlydistributedontheinterval[10,20].Inthelastweekyouwillacceptanyoffer.
Othervariantsoftheproblem:
Noobvioustimehorizoninfinite?,setatimehorizon(aspirationlevel)? Noknowledgeaboutdistributionofjoboffers(e.g.SecretaryProblemtofind
optimalstoppingsolutions) Searchcosts,infoquality,evaluationofjoboffers
7/29/2019 Behavioral Economics - Script Summary
10/53
HeuristicsandBiases Findingviolationsofutilitytheorydeterminingexplanatoryconcepts
describingsystematicbiases Trialanderrormethodstoapproximateasolution Methodtodirectattentionfruitfullyinlearning,decision-making(rules,
suggestions,guides,techniques) Basedonexperienceandcommonsense,improveitsownperformance Ruleofthumb,helpbutnotguaranteedsolution
ExplanatoryConcepts
Representativeness:WhenpeopleareaskedtoratetheprobabilityofAbelongingtoaclassB,theyoftenrefertotheprototypeoftheclass.Standardprobabilisticpointofview,thisisfalse:conjunctionfallacy(occurswhenitisassumedthatspecificconditionsaremoreprobablethanasinglegeneraone),neglectingpriorprobabilities,insensitivitytosamplesize,misconceptionof
chance(gamblersfallacy),insensitivitytopredictability,illusionofvalidity(confidencewhenredundantdataalthoughredundancydecreasesaccuracy),misconceptionofregression
Availability:Inmanysituationspeoplehavetoratethefrequencyofaclassortheprobabilityofanevent.Hereitisoftenrelieduponinstancesoroccurrences,whichcanbeeasilybroadtomind.Instancesoflargeclassesarerecalledbetterandfasterthaninstancesoflessfrequentclasses.Oneinterestingcorollaryfindingtothisheuristicisthatpeopleaskedtoimagineanoutcometendtoimmediatelyviewitasmorelikelythanpeoplethatwerenotaskedtoimaginethespecificoutcomebiases:duetoretrievabilityofinstances(e.g.morefamousmen,falseratethatmoremennameswereinthelist),duetoeffectivenessofasearchset(letterrfirstorthird),biasofimaginability(easilyimagineddisastersnotreflectlikelihood),illusorycorrelation(of2events)
AnchoringandAdjustment:Whenmakingestimations,peopleoftenstartfromaninitialvaluetoderivetheanswer,therebythementalstartingpointoftendeterminestheoutcomeoftheprocess.Thiscanbethecase,eveniftheinitialstartingpointisinnowayconnectedtothetask/outcome.InsufficientAdjustment,overestimateprobabilityofconjunctiveevents,underestimateprob.Offailureincomplexsystems
SomeBiases
Overconfidence(optimism):peopleseemtosystematicallyoverestimatetheirabilitieswhenconfrontedwithgeneralknowledgequestionandaskedfortheirjudgmentconfidence.(cardrivers,chanceofsurviving,winnerscurse=beoptimistic,gotlicense,makelosses=thewinningbidderisoverpaying)illusionofvalidityinconsistencyandextremetypersistalthoughdifferentexperiences(lackofsearchforanduseofdisconfirmingevidence,lackofawarenessofenvironmentaleffectsonoutcomes,useofunaidedmemoryforcoding,storing,retrievinginformation)ContextDependency:evaluationofobjectsnaturallytakesplacerelativetocomparableobjectsandabsolutejudgmentsaredifficultorcanstronglybeinfluencedbyits
environment(i.e.,framing,decisionsetdependency,informationsampling)SunkCosts:pastexpenditures(investment,up-frontpayment,etc.),whicharenot
7/29/2019 Behavioral Economics - Script Summary
11/53
recoverable.Economictheory(aswellasconventionalwisdom)tellsusthatsunkcostsshouldnotberelevantfordecisions.Cannotquitbecausefeelingstrongregret(highexpendituresfornothing).Higherrationality:so.doesnotwanttodestroyhisreputationasasuccessfulbusinessman,thenhisdecisiontostayinthebusinessmayhaveanadditionalvalueandneednot
beirrational,ourtendencytohonorsunkcostsbeneficialinsomecases(instrumentofpre-commitment)Sunkcostsandquasi-rents:Honoringsunkcostsmayhelptopreventexpropriation(Endeignung)evenwithouthightransactioncosts.LimitedDepthofReasoning:peopleinprincipleplanonlywithlimitedforesight(e.g.gameofchess).Itisimportanttoguesswhatothersdo,butiftheydothesame(guesswhatwedo)wefindourselvesininfiniteregressatypicalprobleminGameTheory
SimpleHeuristicsthatMakeUsSmart
HeuristicsfortheEvaluationofAlternatives/BinaryChoiceRecognitionHeuristic:existenceversusnon-existenceofknowledgecanbeinformationitself(e.g.whichcityhasthelargerpopulation?SanDiegoorSanAntonio)Takethebest(TTB):Inordertodecideforoneoftwoalternatives:
1. Lookforrelevantattributesofthealternatives2. Ranktheattributesaccordingtotheirimportance3. Comparethealternativeswithrespecttothemostimportantattribute4. Ifonealternativeisbetterwithrespecttothisattribute,decideforit5. Otherwisecomparethealternativeswithrespecttothenextmostimportant
attributeandasin4.
Minimalistic:step2=randomorder,orfirstcheckdimensionwhichdiscriminatedinthepreviouschoiceConcern:learningofcuevalidations,decisionwhichkindofheuristicinanenvironmentDecisionsunderUncertainty:AlternativetoUtilityModels?
ThePriorityheuristic:approachforintegratingvaluesandprobabilitiesingambles Fundamentaldiscussionsaboutmakingchoiceswithouttrade-offs Incorporatessmallsetifassumptions,enablingpredictionofbehaviorinmost
gambles Consistingof3rules
(a)PriorityRule:Gothroughreasonsintheorder:minimumgain,probabilityofminimumgain,maximumgain,probabilityofmaximumgain(b)StoppingRule:Stopexaminationifthegainsdifferby1/10(ormore)ofthemaximumgain;otherwise,stopexaminationifprobabilitiesdifferby1/10(ormore)oftheprobabilityscale.(c)DecisionRule:Choosethegamblewiththemoreattractivegain(probability)
FurtherHeuristicsandBiasesofThinkingTheGazeHeuristic:HowtoCatchaBall?
Difficulttocomputetrajectory(Flugbahn)ofaballbecauseofdistance,velocity,projectionangle,speed,directionofwind
7/29/2019 Behavioral Economics - Script Summary
12/53
Gazeheuristic:ballhighupinair,fixgazeonball,startrunning,adjustspeedsothatangleofgazeremainsconstant
Nocalculationwheretheballwillland,heuristicleadstolandingpointCognitiveBiases
=DistortioninthewayhumansperceiverealityDecision-MakingandBehavioralBiases
Bandwagoneffectthetendencytodo(orbelieve)thingsbecausemanyotherpeopledo(orbelieve)thesame
Choice-supportivebiasthetendencytorememberone'schoicesasbetterthantheyactuallywere
Mereexposureeffectthetendencyforpeopletoexpressunduelikingforthingsmerelybecausetheyarefamiliarwiththem
BiasesinProbabilityandBelief Ambiguityeffecttheavoidanceofoptionsforwhichmissinginformation
makestheprobabilityseem"unknown" Optimismbiasthesystematictendencytobeover-optimisticaboutthe
outcomeofplannedactions Hawthorneeffectreferstoaphenomenonwhichisthoughttooccurwhen
peopleobservedduringaresearchstudytemporarilychangetheirbehaviororperformance
SocialBiases Haloeffectthetendencyforaperson'spositiveornegativetraitsto"spillover"
fromoneareaoftheirpersonalitytoanotherinothers'perceptionsofthem Herdinstinctcommontendencytoadoptopinionsandfollowthebehaviorsof
themajoritytofeelsaferandtoavoidconflict Projectionbiasthetendencytounconsciouslyassumethatotherssharethe
sameorsimilarthoughts,beliefs,values,orpositions Self-fulfillingprophecythetendencytoengageinbehaviorsthatelicitresults
whichwill(consciouslyornot)confirmourbeliefsMemoryBiasesandErrors
Beneffectance:perceivingoneselfasresponsiblefordesirableoutcomesbutnotresponsibleforundesirableones
Hindsightbias:filteringmemoryofpasteventsthroughpresentknowledge,sothatthoseeventslookmorepredictablethantheyactuallywere;alsoknownasthe'I-knew-it-all-alongeffect'
Selectivememory
Lecture4GameTheoryandExperimentalEconomics
GameTheoryandEquilibriaGameTheory:mathematicallycapturebehaviorinstrategicsituations(successinmakingchoicesdependsonchoicesofothers)
NormalformgamesOR(versus)Extensiveformgames Cooperativeversusnon-cooperativetheory Staticversusdynamictheory Gameswithcompleteversusincompleteinformation
Equilibrium:setsofstrategiesinwhichindividualsareunlikelytochangetheirbehavior
7/29/2019 Behavioral Economics - Script Summary
13/53
EquilibriumConcepts(suchasNashEquilibrium):makepredictionsofoptimalbehaviorParlorGames:representationofsocialproblemsNormalFormGames
Representingasocialproblem,describetwoplayers,eachhavingthesametwostrategiesofwhichtheyhavetochooseone
Smallnumberofdecisionmakers(notone,notmany) Utility(profit,costs)ofadecisionmakerdependsnotonlyonownactionsbut
alsoontheactionsofothers Everyplayerihasa(finiteorinfinite)setSiofstrategies(=completelydescribed
planofbehavior) Allplayerschoosestrategiesatthesametime
2x2normalformgameswithpurestrategies
PrisonersDilemma
7/29/2019 Behavioral Economics - Script Summary
14/53
Rationality:goal=utilitymaximizationSocialconcept:Ifastrategysikofplayeriimpliesastrictlylowerutilityunderallstrategycombinationsofotherplayersthansij,thensikiscalledastrictlydominatedstrategy.sikiscalledweaklydominatedifitimpliesanequalorstrictlylowerutilitythansij.
Noplayerchoosesastrictlydominatedstrategy,theplayerchoosesthedominantstrategy(sij)PD:bothplayerschoosedefect((4,0)istschlechterals(5,2))
Player2:strictlydominantstrategyDPlayer1:strategyisNOTstrictlydominatedPlayer1:eliminatestrategyCChooseD(Cisstrictlydominatedafterthiselimination)Successiveeliminationofstrictlydominatedstrategies
NOstrictlydominatedstrategiesSuccessiveeliminationmaycauseproblemswithdifferentresultsAnothersolutionconceptBestReply(RationalExpectationsaboutstrategiesofotherplayers)
Arrowdiagram:ifplayer1playsDthenplayer2s
bestreplyisDWeimmediatelyseethat(C,C)and(D,D)areNashequilibria.Cooperationgame
Example:thebestreplyofplayer1toCofplayer2isC
NashEquilibrium(NEinpurestrategies):pairofstrategies,whicharebestrepliestooneanother
TofindNE:checkallstrategycombination Ina2x2game:usearrowdiagram
BattleofSexesCouplewithdifferentpreferencesbutdominatingdesiretobetogetherCoordinationinterestswithdifferentandcommoninterests
O=OperaF=Football
7/29/2019 Behavioral Economics - Script Summary
15/53
GameofChicken(Hawk-Dove-Game)Carsoncollisioncourse,whodeviatesloosesSymmetric(BattleofSexes)vsasymmetric(Chickengame)consequencesforout-of-equilibriumbehavior(fightfordominance)
P=peaceful(deviate)A=aggressive(neverdeviate)
MatchingPenniesPlayer1choosesheadortail,player2guesses,if2isrighthereceivesapennyfrom1andviceversaNONashequilibriuminpurestrategies(newproblemH=headT=tail
2x2normalformgameswithmixedstrategiesMixedstrategy:thepurestrategieswillbechosenwithprobabilityp(player1).
p=1/2,q=1/2
7/29/2019 Behavioral Economics - Script Summary
16/53
Counterintuitiveattributesofsuchmixedstrategyequilibria:
Aplayersrewarddoesnotdependonhisownstrategychoice. Aplayersequilibriumstrategydoesnotdependonhisownvalues
ProblemswithNashequilibriainmixedstrategies: Beingindifferent,whyshouldplayerschoosetheirequilibriumstrategies? Howcanweimaginetherealizationofamixedstrategy?
Generalizations
1. Finitenumberofplayersandfiniteumberifstrategies:gamehasatleastoneequilibriuminmixedstrategies
2. Finitenumberofplayersandinfinitenumberofstrategies:a. Cournotoligopolywithhomogenousgoods:nsuppliers,xiquantitiesofa
homogenousgood,pmarketpriceprofitG=xi(p-ci),NE:allxibestrepliestooneanotherminimaxfunction
b. Cournotoligopolywithinhomogeneousgoodsc. Bertrandoligopolywithinhomogeneousgoods
CounterEquilibriumarguments
BertrandDuopolywithhomogeneousgoodsanddifferentconstantmarginalcostsnobestreply
JewishPoker:everyonethinksofanumber,thehighernumberwinsProblemswithNashequilibriumasasolutionconceptSometimesnon-existenceofequilibria(notsevere)Coordinationofchoices
- Ifthereisoneequilibriuminpurestrategies(notsevere)- Ifthereismorethanoneequilibrium(equilibriumselection)- Inmixedstrategyequilibria
ExtensiveFormGames
Gametreeindifferentvariations,orderofchoices(moves)DetermineinitialnodereachingeveryothernodeonauniquepathStrategy:completeplanofbehavior=determinedecisionineverynode(moves)
7/29/2019 Behavioral Economics - Script Summary
17/53
ArgentinaS1:oinnodeA;S2:finnodeAGreatBritain:S1:kinnodeGB,Subgame:isthatpartofthegametree,whichstartsfromacertainnodewhichisnotaterminalnode.LetI-'beasubgameofI-.TheneverystrategysofI-inducesastrategys'ofI-'accordingtotherulethatallmovesofswhichdefinebehaviorinanodeofI-'arealsomovesofs.
Nashequilibrium=subgameperfect,iftheequilibriumstrategiesinduceaNashequilibriumoneverysubgameMethod:BackwardInduction(DemnachwirdimLaufedesVerfahrensrckwrts,alsoinRichtungdeserstenEntscheidungsknotens,derjenigePfadhervorgehoben,welcher
frdenAkteurdiemaximaleAuszahlunggenerierensoll.DadieserPfadeinNash-GleichgewichtinjedemTeilspielinduziert,istdasresultierendeGleichgewichtauchteilspielperfekt)
7/29/2019 Behavioral Economics - Script Summary
18/53
DieRckwrtsinduktionsschritteinExtensivform:
Schaubild1:DasletzteTeilspiel(GRN)wirdbetrachtet:
Schaubild2:DaszweiteTeilspielwurdevereinfacht(ROTmitrationalerAuszahlungausGRN):
Schaubild3:DaskomplettinduzierteSpiel(BLAUmitrationalenAuszahlungenausROTundGRN):
http://de.wikipedia.org/wiki/R%C3%BCckw%C3%A4rtsinduktionTheUltimatumGame:Proposersplitsamountofmoney,respondercanacceptorreject
TheGameofTrust TheCentipedeGame
Generalization1:infinitelymanystrategies(InvestmentGame,GiftExchangeGame)Generalization2:infinitelymanyperiodsGeneralization3:incompleteinformation(esp.signalinggames)
7/29/2019 Behavioral Economics - Script Summary
19/53
ExperimentalResultsGenerally,thereismorecooperativebehaviorinTrust,Investment,andCentipedeGamesthanyouwouldexpectunderegoisticpreferences.2x2NormalFormGames
PrisonersDilemmacooperationduetodifferentmonetarypayoffsandpreplaycommunication(noeffectintheory),nocooperationwhenrematching(repeatedgameswithrandomstrangers)Matchinggamewith2equilibria(ResultsofMatchingGameswithCommonInterest),selectionmaybefocalorphysicalprominent,differencebetweenmatchingandpicking(salience)Gameswithdifferentinterests:BOScommoncoordinationfailure(likepredictedinmixednash
equilibrium),solution:one-waycommunicationbutnottwo-waycommunication,effectoftiming(lowermismatchwhenknowingthatrowplayedfirstbutnotknowingwhatrowdid)ChickenGame:AggressiveStrategychosenmoreoftenthanpeacefulstrategybutIthinknotsignificantly(39:33)CournotExperiments
Furtherresultsofameta-studybyEngel(2007)emphasizethattherearemanyinteractioneffects,i.e.oftenitisnotpossibletolinearlydisentangletheeffectsofdifferentvariables:
competitionwith(in-)homogenousgoodsand(a-)symmetricplayersyieldlowerpricesandprofits(-)orhigherpricesandprofits(+)thantheNashequilibrium,accordingto:
competitionwith(in-)homogenousgoodsandplayerswhoarere-matchedeveryround(partners)orgetnewco-playersineveryround(strangers)yieldlowerprices
andprofits(-)orhigherpricesandprofits(+)thantheNashequilibrium,accordingto:
Furtherinteractioneffects:2competitors=partnerscolludemoresuccessfullythanstrangers;morethan2:strangerscolludemorethanpartners.Thisresultchangesiffeedbackinformation(aboutbehaviorinpreviousrounds)isgivenalsoforstrangers.UltimatumExperiment
Proposeroffers30to50percent,offersbelow20percentarerejectedhalfofthetimeoftenoptimum:give50%
7/29/2019 Behavioral Economics - Script Summary
20/53
Vendettas(Rachefeldzug)
=lesscooperationthantheorypredicts 2playerswithcertainprobabilitytowinaprize,probabilitytostealfromone
anotherwithanefficiencylosssumofprobabilitiesdecreases SubgameperfectNashequilibrium:nostealing Experiment:60-80%ofthegamesendedwithprobabilitiessmallerthan10%
Lecture5TemporalEffectsandSelf-Control
DiscountinganTemporalDistortionIntertemporalConsumption
Jahr1:1000Jahr2:1000*(1+r);r=interestrateJahr1:1000/(1+r)Jahr2:1000;r=discountrate(presentvalueof1000tomorrow)1/(1+r)=discountfactordr=(1-d)/dIfdiscountfactorishigh(closetoone,ordiscountrateisverylow)whathappensinthefuturemattersagreatdeal,youdonotdiscountfutureverymuch=patience.Ifdiscountfactorislow(closetozero,ordiscountrateisveryhigh)whathappensinthefuturematterslittle,youdiscountyourfutureheavily=impatience.
Exponentialdiscounting Hyperbolicdiscounting
AgentsbehavioristimeconsistentE.g.Youpreferatoballtimesorbtoaalltimesoryouareindifferent
asaparametergoverningthedegreeofdiscountingAgentsbehavioristimeinconsistentverycommon
ChoicebetweensmallerearlierawardandlaterbiggerawardimpulsivitySourcesofinformationaboutdiscountratesare:
Fieldstudies/observationsofrealdecisions(revealedpreference) Questionnairestudies(hypotheticaldecisions) Experiments(realdecisions)
Often:Preferenceforpresentpleasures(lowdiscountfacto,highdiscountrate)Seefuturepleasuresonadiminishedscale
7/29/2019 Behavioral Economics - Script Summary
21/53
ResearchingDiscountRates
Implicitdiscountratesarelarger(futurematterslittle)forsmallthanforlargesumsofmoney.(MagnitudeEffect)explainedbyperceptionsofabsoluteamountsandmentalaccounting
Discountratesarelargerforgainsthanforlossesofequalmagnitude.(SignEffect)explainedbydebtorlossaversion
Discountratesdeclinewithtimehorizon.(Non-ExponentialorQuasi-HyperbolicDiscounting)dynamicinconsistenciesinchoicesovertime
7/29/2019 Behavioral Economics - Script Summary
22/53
Hyperbolicdiscounting:profitabilityofinvestmentseemstobedifferentdependingonthetimewhenyouevaluateiti.ediscountingwith1/kinperiodkExamples
Weaknessofwill Myopia Timeinconsistencies
PlanningandSelf-Control=self-controlmechanismtosticktopreviouslyset(planned)goalsovertimeObservingChoicesovertime
1.EndowmentEffectWillingnesstopay(WTP)vs.Willingnesstoaccept(WTA)
Economictheory:WTP=WTA
Experiments:WTP
7/29/2019 Behavioral Economics - Script Summary
23/53
Reasonsforarationalendowmenteffect Ownerspenttimeormoneyforthegood(honoringsunkcosts) Asymmetricinformationriskaversionofferless(lowerWTP) Differentqualitiesandasymmetricinformation:betterqualityissoldlessoften
ornotatall
2.StatusQuoBias=Donothingormaintainyourcurrentorpreviousposition
Example:Notquittingonesjob,notchangingonesfieldofstudies,ornotmakingadivorce,notdroppingoutofbusiness(maybestrategic)
Reason(mayberational):incompleteinformationaboutalternativesbutcompleteinformationaboutstatusquo,hightransactioncosts
3.PreferenceoverProfiles
Notonlyprefernowoverlaterbutalsoutilitystreamsovertime
Preferred:utilityspreadandincreasingutilityprofiles(e.g.cleaningnowandgoingtothemovieslaterisbetter)negativetimepreferences
Peakendrulespecifiesordersofeventshighestutilitytooccurlast
MeansofSelf-control(3strategies)
1. Possiblephysicalormentalrestrictionofthedecisionspace:actsofpre-commitmentorpersonalrules(e.gbudgeting)
2. Manipulationofenvironmentalstructurea. Addingsituationalcomponentswhichsupportdesiredactivityb. Removingdeviationevokingstimulic. Perceptionofeventscostsandbenefits(elaboration,grouping,temporal
factors)3. Changethecontingency(Zufall)structure:sidebetswithpenaltiesandrewards
PrecommitmentIndividualhaschoseninthepresenttodo(ornottodo)somethinginordertocontrolthewayinwhichitwillrespondtoparticularsituationsinthefuture.Socialsettings:Signalingprecommitmentiscrucial,e.gcuttingcreditcard
MentalaccountingStructuringcreditsanddebits,distinguishingdifferenteventcategoriessourceofmoneyaffectshowitisspentMentalaccountsconstrainoptimization,restrictfungibility(bertragbarkeit)
Savingforahousewith2,5interestrateonbankaccountandatthesametimebuyingacarwithaloanat5%
Getmoneyforbirthdayanddon'tspenditforpayingelectricbillProcrastinationorchoosingnottochooseNaifs(hyperbolicdiscounters)notawareofanyself-controlproblems
Sophisticatedhyperbolicdiscouners:areawarebutoftenovercompensateorpreproperate(doingsthearlywhichwouldbebetterdoinglater)
7/29/2019 Behavioral Economics - Script Summary
24/53
Lecture6IndividualDifferencesandEmotions
IndividualCharacterizations
Psychology:manyindividualdifferences Economy:focusonriskypreferences Learning=oneprocessleadingtoindividualpreferences,+corresponding
strategyselection=fundamentalcriterionforoptimaldecisionsBehaviorunderRiskandUncertainty(keyfeaturesofmost(complicated)decisions)AssumptionsofSEU(Subjectiveexpectedutility)
Risk=nottoknowtruestateoftheworld,butprobabilities Uncertainty=notknowtheprobabilitiessubjectiveprobabilities Maximizeexpectedutility
Composelotteries:allowingx,yalsotobelotteries=allowingtodealwithmorethantwostatesoftheworld
Neumann/Morgenstern/Savage-Axiomsfordecisionunderrisk:assumptionsof
behaviorwhichguaranteeexistenceofutilityfunctionusedmaximizeEURiskSeekingandLossAversionApersonisgiventhechoicebetweentwoscenarios,onewithaguaranteedpayoffandonewithout.Intheguaranteedscenario,thepersonreceives$50.Intheuncertainscenario,acoinisflippedtodecidewhetherthepersonreceives$100ornothing.Theexpectedpayoffforbothscenariosis$50,meaningthatanindividualwhowasinsensitivetoriskwouldnotcarewhethertheytooktheguaranteedpaymentorthegamble.However,individualsmayhavedifferentriskattitudes.Apersonissaidtobe:
risk-averse(orrisk-avoiding)-ifheorshewouldacceptacertainpayment(certaintyequivalent)oflessthan$50(forexample,$40),ratherthantakingthegambleandpossiblyreceivingnothing.
risk-neutral-ifheorsheisindifferentbetweenthebetandacertain$50payment. risk-loving(orrisk-seeking)-iftheguaranteedpaymentmustbemorethan$50(forexample,
$60)toinducehimorhertotaketheguaranteedoption,ratherthantakingthegambleandpossiblywinning$100.
Theaveragepayoffofthegamble,knownasitsexpectedvalue,is$50.Thedollaramountthatthe
individualwouldacceptinsteadofthebetiscalledthecertaintyequivalent,andthedifferencebetweentheexpectedvalueandthecertaintyequivalentiscalledtheriskpremium.Forrisk-averseindividuals,itbecomespositive,forrisk-neutralpersonsitiszero,andforrisk-lovingindividualstheirriskpremiumbecomesnegative.
7/29/2019 Behavioral Economics - Script Summary
25/53
ThedifferencebetweentheutilityoftheexpectationvalueandexpectedutilitydependsonthecurvatureofU(x).(ErwarteterNutzenausSpielE(U)istgreralsNutzenE(EW)deseigenenErwartungswertdesSpielsE(W)zb50riskseeking)Wedonotwantthemeasuretodependonlineartransformationsofx(seeabove).
Arrow-Pratt-Measure(ofabsolute/relativerisk-aversion)Themoststraightforwardimplicationsofincreasingordecreasingabsoluteorrelativeriskaversion,andtheonesthatmotivateafocusontheseconcepts,occurinthecontextofformingaportfoliowithoneriskyassetandonerisk-freeasset.Ifthepersonexperiencesanincreaseinwealth,he/shewillchoosetoincrease(orkeepunchanged,ordecrease)thenumberofdollarsoftheriskyassetheldintheportfolioifabsoluteriskaversionisdecreasing(orconstant,orincreasing).Thuseconomistsavoidusingutilityfunctions,suchasthequadratic,whichexhibitincreasingabsoluteriskaversion,becausetheyhavean
7/29/2019 Behavioral Economics - Script Summary
26/53
unrealisticbehavioralimplication.Similarly,ifthepersonexperiencesanincreaseinwealth,he/shewillchoosetoincrease(orkeepunchanged,ordecrease)thefractionoftheportfolioheldintheriskyassetifrelativeriskaversionisdecreasing(orconstant,orincreasing).
r(u)=independentofabsolutemagnitudeofutilityandunitofutilityr:locallydefined=measurelocalriskaversionThestrongertheconcave(convex)curvatureofU(x)thehigherisriskaversion(riskseeking)Utilityonitsownisnotabletocapturedifferencesingeneralriskaversion.
ProblemswithscaleMustpeopleberiskneutralwithrespecttosmallbets?(Rabin2000)
Whatwouldbetheconsequenceifsomeonealways,i.e.foreverywealthx,rejectsthefollowinglottery(L)againstthealternativewin/lose0Euro?L=win11withprobability0.5;loose10withprobability0.5.Therejectionofthelotteryimplies:0.5U(x10)+0.5U(x+11)
7/29/2019 Behavioral Economics - Script Summary
27/53
ProspectTheory
Kahneman/Tverskydifferenceinchoicesunderuncertaintyavoidanceoflossesorseekingofrisk
ImportantElements:o Referencepointfromwhich
gainsandlossesaredetermined(makethebestoutofsituation,adapttocomfortableoruncomfortablesituations)
o Valuefunctionovergainsandlosses(S-shaped,risk-averseforgains,riskseekingforlosses,atRP:steeperforlossesthanfor
gains)o Transformationofobjectiveprobabilities(lowprobabilitiesareoverweighted(ziehenesmehrindieEntscheidungsfindungein),highpareunder-weighted)
Differentutilitycurvesforgains,losses,probabilities,varyingreferencepoints Alwaystransformsituationinprospecttheorylanguagedescribedecisions
andreferencepoint
EndowmentEffect
FramingshiftofreferencepointCase1:Scenariosdescribedthatlivescanbesavedriskaverse(gains)Case2:Scenariosdescribedthatlivesarelost riskseeking(losses)Advantages DisadvantagesProspectTheorycanexplainanumberofphenomena,oftencontradictingstandard
economictheory.
Thereissomearbitrarinessinexplainingthesituation,inparticularwithrespectto
thereferencepoint.Itistoosimple:realevaluationsare(sometimes)morecomplex.
7/29/2019 Behavioral Economics - Script Summary
28/53
DecisionswithEmotionsDefiningEmotions
1. Emotionsarenecessarytoeducatestatesanddecisions;theyprovideuswithourgoalfunction.
2. Emotionshelptoselectandstoreimportantinformation.3. Emotionsprovideuswithtoolsofself-commitment,whichhelptocopewith
othersandwithourselves. Classifiedinacontinuumorasdistinct(basic)states
RobertPlutchiksclassificationin8primaryemotionsPrimaryemotionsblendtogethertoformfullspectrumofhumanemotionalexperience.
Relatingeachbehaviorwithsurvivalvalue,partofbiologicalheritage.Ekmanslist(1)anger(2)fear(3)sadness(4)happiness(5)disgust
(Experiments)
EmotionsandEvaluation(Lazarus) TaxonomyofCoreRelationalThemesforvariousemotions,helpdefinefunction
andelicitingconditions Offenseagainstmeandmineforanger Immediate,concrete,overwhelmingphysicaldangerforfear Irrevocablelossforsadness Takinginorbeingtoclosetoanindigestibleobjectfordisgust Reasonableprogresstowardrealizationofagoalforhappiness Example:marriagepartner,chosefood,crossingstreet
EmotionsandExperience
Emotionscarryexperienceswisdom Speciesandindividualexperiences,(e.gfoodtaste,landscapepreferences,fearof
certainanimals,prejudicecopingwithpeople) Prejudice:tooyoungandoldpartners,ugly=illnessadvantageous Betterlearningwheninformationaccompaniedbystrongemotionsselect
appropriatereactionincertainsituation(i.e.self-commitment,Herdplatte) Iftheothersteals,theangrypersonisworseoffthantherationalpersonbutthe
otherwontstealifyouareanangrytype.
7/29/2019 Behavioral Economics - Script Summary
29/53
RegretTheory
Feelregretwhenyoumakedecisionandfindoutitwaswrongdecision. Takeanticipatedregretintoaccountwhentheydecidelossaverseness Powerfuldissuaderorcreatespecificmotivationtodoonethingtoavoid
somethingelse Modifiedutility=contextdependent
andavailablealternativescaninfluencethechoicebyanticipatedregret
LearningandAdaptionStrategySelectionandLearning
Learning=fundamentaldriverofoptimization Learningcurvesindifferentdomainsexpectsteepcurveatbeginning,
flatteningoutoverrepetitions Learningofstrategiesindifferentenvironmentsoptimizationstrategy
selection(adaptiontodifferentenvironmentaltaskcharacteristics) 2differentstrategiesrelyonlearningperformancesticktostrategyor
switchtotheotherone Example:TakeTheBest(TTB)vs.WeightedAdditive(WADD)=formoflinear
informationintegrationo TTB:inanon-compensatoryenvironmentfromprob.=0,5to0,7after
severaltrialblockso WADD:inacompensatoryenvironmentfromprob.0,5to0,75after
severaltrialblocks
7/29/2019 Behavioral Economics - Script Summary
30/53
ImitationandAdaption
Copyingandimprovementofexistingstructuresdevelopmentofnewdesignforsolvingproblems
Catchinganaircraftwithtailhookandarrestingwires,plusradiocommunication,lightssignalsindifferentcolorsfromdifferentangles
Bionics(biology+electronics):applicationofmethodsandsystemsfoundinnaturetothestudyanddesignofengineeringsystemsandmoderntechnology
Imitationdesirableevolutionpressureforcesnaturalsystemstobecomeoptimizedandefficient
E.g.lotuseffect,sonar,radar,hullsofboatsfromdolphins,artificialneuralnetworks
Implementingfunctionfoundinnatureratherthanjustimitatingbiologicalstructureformofappliedcase-basedreasoning(natureasdatabaseofworkingsolutions)
Modelingtechnologybasedon3biologicallevelo Mimickingnaturalmethodsofmanufactureo Imitatingmechanismsfoundinnature(velcro)o Studyingorganizationalprinciplesfromsocialbehavioroforganisms,such
asflockingbehaviorofbirdsoremergentbehaviorofbeesandants E.g.Velcro,catseyereflectors,morphingaircraftwings,pineconeclothing,
architecturebasedonstructureofhumanbonesorplantsExamplesandboundedrationalityareconnectedontwolevels:
1. Natureinventedallthesecleverdesignswithoutanyplanning:trialanderror,i.e.variationbythemixingofgenesormutationandevaluationbydeathorsurvival.
2. Menimitatenature,eitherbyadoptingitssolutionsorbyadoptingitsmethods(geneticalgorithms,forexample).
Noteagainthatimitation(ofsuccessfulbehavior)isoneofthemostpowerfulprinciplesofboundedrationality.
Lecture7SocialBehavior
SocialMotives
Peopletakesocialconsequencesintoaccount Careaboutownwell-beingandalsoaboutthatofotherpersonsFairness
Concerningdivisionofresourcesallrecipientsreceivefairamount ProportionalDivision:Eachplayergetsatleasthis/hershare. ExactDivision:Everyplayerassumesthateveryonereceivedexactly(neither
morenorless)theirfairshare. EquitableDivision:Thereceivedproportion,bytheirownevaluation,isthesame
foreveryplayer. Envy-FreeDivision:Noonewantssomebodyelsessharemorethantheirown. EfficientDivision(Pareto-Optimal):Nootherallocationwouldmakesomeone
betteroffwithoutmakingsomeoneelseworseoff.
7/29/2019 Behavioral Economics - Script Summary
31/53
TheContestedGarment:Twoholdagarment;oneclaimsitall,theotherclaimshalf.Thentheoneisawarded34,theother14.Theprincipleisclear.Thelesserclaimantconcedeshalfthegarmenttothegreaterone.Itisonlytheremaininghalfthatisatissue;thisremaininghalfisthereforedividedequally.Notethatthisisquitedifferentfromproportionaldivision
Spite:Phenomenonoccurringwhenaplayer'svalueofanallocationdecreaseswhenoneormoreotherplayers'valuationincreasesplayerwithspiteprefersallocationinwhichotherplayerreceiveslessthanmore.(30pfelvs2Bananen(toll))Avoidanceofinequality:specificformoffairness,differencesnegativeutilityforplayerAltruism
Explainingaltruisticbehavior:KinSelectionandMemes Example:beegivinglifewhenstinging,human:religion(common>individual)
DictatorGame(Dictatorendowedwithmoney,cangiveasmuchashelikestobeneficiary)1. PureAltruism(mostfrequentlyuseintheory):dictatorsutilityisinfluencedby
thebeneficiarysincomea. UDictator=fD(xDictator,xBeneficary),xi=incomeofi
2. BeckerianAltruism(Becker):dictatorsutilitydependsonthebeneficiarysutilityUndercompleteinformationitcanbesolvedtogetPurealtruism
a. UDictator=fD(xDictator,UBeneficary)b. UBeneficary=fB(xBeneficary,UDictator)
3. WarmGlow:actofgivingmakesthedictatorhappya. UDictator=f(xDictator)+g(YDB),b. Y(DB)=transferfromDictatortoBeneficiary,c. IfxBeneficiary=YDB,then(3)isaspecialcaseof(1).
4. ReciprocalAltruism:dictatorsaltruismdependsonthebeneficiarysaltruismtowardher
a. UDictator=f(xDictator,xBeneficary,aBD)b. aBDmeasuresthebeneficiarysaltruismtowardthedictator.
5. AltruismasInequityAversion:wherethedictatorsutilityisinfluencedbythedifferencebetweenhisownandthebeneficiarysincome(thesearespecialcasesofpurealtruism
o Standardgame:hardlyanydifferencesbetweenapproaches Restrictiontotransferhalftheendowment(hardlygivenmorethanhalf)dueto
altruismasinequityaversion Morecomplicatedsituationsmaybegovernedbyalternativeoradditionalsocial
motives,suchasconditionalcooperationorreciprocity
Cooperation ColemandescribesCalculativetrust.Valuetrust>valuemistrust
7/29/2019 Behavioral Economics - Script Summary
32/53
ColemanHypothesis:rationalexpectationsabouttrustworthinessincreasingfrequencyoftrustwithaveragenetprofitabilityoftrust
Mostcases:reciprocity(wecooperatewhenothercooperate)Reciprocity
Respondingtopositive(negative)actionwithanotherpositive(negative)action Positivereciprocalactionsaltruisticactions Involvesimplicitexpectationoffuturepositiveactionsbybeneficiary Explainsmaintenanceofsocialnorms(maintainnorminabsenceofformal
sanctions)punishingaction(negativewordssocialostracism) Cheaterdetectionandpunishmentinfluencedbythecontextdefiningsocialnorm
CheaterDetectionWasonSelectiontask:Ifsomeonestaysovernightinthecabin,thenthatpersonmustbringalongabundleofwoodfromthevalley.
A. Cheatingversion:cabininthealps,rumorsthatnotallhikeswhostayovernightcarryownshareoffood89%selectedrightcardstosolvelogicpuzzleB. Nocheatingversion:hikersbringwoodormembersofalpineclub,samecards
only53werecorrect(alternativeversiondecreasecorrectness)IfcuedbytherulesofonepartyA(i.e.employees)ortheotherpartyB(i.e.employers)theresultscanchangeconsiderably.Ifanemployeeworksontheweekend,thenthatpersongetsadayoffduringtheweek.
Amainlyturnedthefollowingcards:workedontheweekendanddidnotgetadayoff
Bmainlytheothercards:didnotworkontheweekendanddidgetadayoffTitforTatStrategy(=Exampleforreciprocity,heuristic)
1. Unlessprovoked,theagentwillalwayscooperate.2. Ifprovoked,theagentwillretaliate.(Rchtsich)3. Theagentisquicktoforgive.4. Theagentmusthaveagoodchanceofcompetingagainsttheopponentmorethan
once.Despitethefactthatthetit-for-tatsneverwonamatchandthedefectorsneverlostamatch,thetit-for-tatsstillcameoutahead,becausethefinalscoreisnotdeterminedby
thewinnerofmatches,butthescorerofpoints.Thetit-for-tatgainedmorepointstyingwitheachotherthantheylosttothedefectors.Themoretit-for-tatsthereareinthegame,themoreadvantageitistobeone.Reciprocityandconditionalcooperationisastrategyagainstfreeriders,whencheaterscannotbepunishedotherwise.Butnotethatoftenalsothelevelofdefectorshastoreachaspecificleveltobeadvantageousforthereciprocalagent.Furthercriticismisthealternationproblemandfinitegames.Example:2malebaboon,onedistractingalphamale,theothercopulatewithfemalereversedrolesforpaybackConditionalCooperation
Peoplewouldliketocooperateifotherscooperate(providingapublicgood Notrelevantfordictatorgame(asymmetricpower)butforPublicGoodsand
Solidaritygame
7/29/2019 Behavioral Economics - Script Summary
33/53
ConditionalCooperationwithPublicGoodsPotentialforreciprocalactionsbyplayersincreasestherateofcontributiontothepublicgood,providingevidencefortheimportanceofreciprocityinsocialsituationsPublicGoodGameasgeneralizationofPrisonersDilemma(>2player)
AllnparticipantsendowedwithincomeE InvestpartxiofEresultinginatotalinvestmentx Allparticipantget(with1/n
7/29/2019 Behavioral Economics - Script Summary
34/53
Lecture8ExperimentalEconomics
ExperimentalDesign
Rapidgrowthoflaboratoryexperimentsconfidence Carefullycontrolledexperimentsdrawbetterinferencesthanavailablenon-
experimentaldata Questions:howandwhatdetailtoreportexperimentalprocedures,whatdata,
howtoaggregateforreportingpurposes,howtoanalyze Pitfalls:misinterpretthesignificanceandrobustnessofreportedresults Divergentpracticesamongexperimentaleconomistsaboutwhatunitofdatais
called"anexperiment."(amatterofterminology,sometimesmasksdifferencesinreportingpracticesambiguityaboutwhatisbeingreportedandpublished
StrictGuidelinesInducedvaluetheory:3necessaryexperimentalconditionstoinduceanagentscharacteristic.(Alwaysbesatisfiedtoensureeconomicallyvalidbehavior)
Monotonicity(Monotonie)-prefermorerewardtolessandnotbecomesatiated Salience(Hervorspringen)-rewardmustdependonactions Dominance(berlegenheit)-subjectsutilitycomespredominantlyfromthe
rewardmediumandotherinfluencesarenegligible +Procedureastransparentaspossible +Experimentreplicable
FactorialDesign
Differentinfluencesareseparatedintoseveralfactors,investigatedincombination
Usuallyfullfactorialdesignextensiveexperiments Alllevelsofeachfactorarecombinedwithalllevelsoftheotherfactors Example:3factors(2with3levels,1with2levels)3x3x2(18trials) Allpossibleinteractioneffectsarecaptured Keyparadigms:
o Independenttrialsasseparatesessionso Realismandcontrolledeconomicenvironmentso Parallelism(transferabilitytorealworlddecisions)o Understandresearchquestionsandpitfalls(instructions,payments,learning,setup,samplesize..)
Simplifications:crossoverdesign,dualtrial,fractionalfactorialdesign..VariablesofInterest
Dependent(focusvariables,hopetofindaneffect)andindependentvariables(varysystematically)
Decisionexperiments:Process(decisionprocessitself)andoutcomevariables(describeresultofprocess)investigatedincombinationtofosteratheory
E.g.TTB:predictionsaboutchoicebetween2alternativesandassumptionsabouttheprocessleadingtoresult Betweensubjectsdesign:effectondependentvariablesiscomparedfordifferentsubjects(oversubjects)
7/29/2019 Behavioral Economics - Script Summary
35/53
Withinsubjectsdesign:effectofdifferentlevelsofvariablesforonesubjectAdvantage:additionalnoise(individualpreferences)doesnotinfluencetheeffectofdifferentmeasures
Withindesignismorepowerful,butpronetodemandeffects(esp.learning) Both:randomizationindirectcontroloveruncontrollablevariableslike
demographicsorlearning
ExperimentalStudies Experiment=goldenpathofscientificdiscovery Strongvariationsinwaysofcontrolledinvestigation,strongerrestrictionsin
experimentaleconomics Mainfocusonexperimentalgames,individualdifferencesorunderlyingcognitive
processes Laboratoryvs.fieldstudies
Methods1. Classroomexperiments:simplyruninlectures2. Laboratorysetting:standard,bestcontrolofvariousvariablesofinterest(ViaLab
atEUV)3. Fieldstudies:increaseecologicalvalidity
a. Reducedrandomizationinfavoroforiginalityofbehavior(subjectpool,infolevel,commodity,norms)innaturalenvironment
b. Artificial:decisionsituationsimulatedinthelaboratorywithspecificsubjectpool
c. Framed:specificcharacteristicsaretakenintothelaboratoryd. Natural:takeplaceweredecisionsareusuallymade,participantsnot
informedaboutmonitoringe. Quasineutral:describesituationsinnaturalenvironmentwhichare
slightlymanipulatedtotesteffectofspecificchange4. Meta-analysis:investigateresultsfromdifferentstudies
a. Aim:identifyunderlyingpatterns,relationofdifferentobservations(e.gthroughcommonmeasureofeffectsize)
b. Identifycharacteristicsexplainingdifferencesinresultsc. Oftenimportantcomponentsofasystematicreviewprocedure
Measurements
Weber-Law +otherhypothesizedinterindividualregularities=resultofearlyresearchofuniversalcharacteristicsofperception
Otherapproaches:interestedininterindividualdifferences(questionnairessurveys,testbatteries)
Economics:regularitiesandvariationsinregardtobehavioralbiases,motives,attitudesimprovemodelsofsocialbehavior(e.g.includedifferenttypesofpeoplewithsocialmotives)
Strategymethod:evaluatingfulldecisionspaceofapersonRiskMeasure
Bestresearchedindividualcharacteristic:riskpreferences Attitudestowardsriskscanbeinferredbasedonindividualchoicesbetween
lotteries
7/29/2019 Behavioral Economics - Script Summary
36/53
Systematicallyuseproposedsetofparedlotterychoices(AandB)classificationofriskaversion(loving)
Majorityofsubjectisriskaverse Riskaversionincreaseswithstakesbutnotwithhypotheticalpayment
RelativeOverconfidence Welldocumentedandeasytoresearchexperimentally Example:Viadrinastudents:mathematicalabilitiesrangeofothers,ownrange Averagestudentsrateshisownabilitieshigherthanthoseoftheaverage
comparablestudent =Optimismbias
Intrinsicversusextrinsicmotivation
Introductionofpayment:canreducesupplyofvoluntaryservices,crowdingoutmightresult
Manypeoplelikework,strongintrinsicincentivestodelivergoodresults Introductionofadditionalextrinsicincentivesmayreduceworkeffort Reasonsforcrowdingout(Verdrngungseffekt):
o Feelingofdiminishedself-determinationoro Compensationforsocialvalueofanactivitymaybedisappointinglylow
Thedestructionofintrinsicmotivationisusuallyirreversible,i.e.theabandoningofextrinsicmotivationdoesnotre-establishintrinsicmotivation
Crowding-in:introductionofextrinsicrewardsfosteranactivityifthepriceforasocialactivityishigherthanexpected(signalingtheory)
Negativepricestopunishunsocialbehavior:notworking
Relationshipbetweenpositive/negativeincentivesandsocialbehaviorillustratedbyW-Effect(Adecreaseinproductivityisobservedforsmall,butnotforlargeincentive)
Motivationofothersisimportantfortheownevaluation
StiftungWarentestoderBIldZeitung?
Lecture9DecisionofandWithinGroups
GroupDecisions Mostdecisionsmadebygroupofpeople(family,firmsetc.) Decisionprocessdependingoninteractionswithothermember
TendenciesofGroups
Socialperception:processofacquiring,interpreting,selectingandorganizingsensoryinformationinfluencedby2factors
7/29/2019 Behavioral Economics - Script Summary
37/53
Structuralfactors:drivingsolelyfromthenatureofthephysicalstimuliandthenaturaleffectstheyevokeinthenervoussystemoftheindividual
Functionalfactorsofperceptualorganizationderiveprimarilyfromtheneeds,moods,pastexperience,andmemoryoftheindividual(hungry)
Allfunctionalfactorsinperceptionaresocialintheusualsenseoftheterm Differentperceptionsofthehungryandnot-hungrystudentscouldnotbedueto
structuralfactors,sincethesamepictureswerepresentedtobothgroupsbutcouldbedueonlytothedifferencesinneedormotivationofthemembersofthetwogroups
Social(functional)factorsreflectingthedemand(Bedrfnisse)ofthegroupcanstronglyguidegroupdecisionsandleadtogroupthink
BehavioraltendenciesresultingfromsharedvaluesandreciprocalreinforcementGroupPolarization
Observedtendency:riskyshift=lessconservativeandmoreriskydecisionsthanthosebyindividuals
Groupactionmoreextremeversionofeachindividualspreferredaction Universalconfirmationbias=tendencytosearchfororinterpretinformation
thatconfirmspreconceptions,avoidancewhichcontradictpriorbeliefs Anotherexplanation:supporttheory
SupportTheory(Tversky)
Explainjudgmentsbytheaggregationlevelofinformation P(packed)P(unpacked)P(constituent) Judgingprobabilityofdescriptiondependingon
o Whatmemberspeoplebringtomindgivenadescriptiono Howsupportgetsaggregatedacrossthesemembers(supportvalue)
Groupdecision:unpacksupportiveinformationandpackedcounter-supportiveinformation
=Explanationfortheconjunctionfallacyandinvolvedheuristics(representativeness,availability)
AdaptionofGroups
Groupsadapttogether=specificstrategiesprevail Evolutionarilystablestrategy(ESS):adoptedbyapopulationofplayers,cannot
beinvadedbyanyalternativestrategythatisinitiallyrare =EquilibriumrefinementoftheNashequilibriumthatisevolutionarilystable
oncefixedinapopulation,naturalselectionaloneissufficienttopreventalternative(mutant)strategiesfromsuccessfullyinvading
E(S,S)>E(T,S)orE(S,S)=E(T,S)andE(S,T)>E(T,T)(ThisdefinitionisthedefinitionofweakESS(oranevolutionarilysteadystate)
ThestrongversionofESSis:E(S,S)E(T,S)andE(S,T)>E(T,T) Inmostsimplegames,ESSandNashequilibriumcoincideperfectly
Strategy ESS
Both(A,A)and(B,B)areaNashequilibrium.OnlyBisanESS.Bcan
neutrallyinvadeapopulationofAstrategists,whereuponitwillcometopredominatesinceBscoreshigheragainstAthanAdoesagainstB.E(A,A)=E(B,A),butitisnotthecasethatE(A,B)>E(B,B).
7/29/2019 Behavioral Economics - Script Summary
38/53
WhileDstrategistsmaytemporarilyinvadeapopulationofCstrategistsbyscoringequallywellagainstC,theypayapricewhentheybegintoplayagainsteachother:CscoresbetteragainstDthandoesD.SoherealthoughE(C,C)=E(D,C),itisalsothecasethatE(C,D)>E(D,D).AsaresultCisanESS.
Definition:xundysindzweiStrategienE(x,y)istderReproduktionserfolgeinesIndividuumsmitStrategiexineinerUmwelt,inderalleanderendieStrategieyausfhren.xistnurdanneineESS,wennfrallegilt:E(x,x)>E(y,x),oderE(x,x)=E(y,x)undE(x,y)>E(y,y)DerersteTeilderDefinitionistquivalentzumsogenanntenstriktenNash-Gleichgewicht,d.h.:xistbesteAntwortaufsichselbst.KeineandereStrategieschneidetgegenberxbesserab,alsxselbst.DerzweiteTeilsiehtvor,dass,wennStrategieyinreinerx-Umgebungnichteliminiertwird,dasiegleicherfolgreichist,xinreinery-Umgebungsichgegenberydurchsetzenkann-somitkanndiePopulationnichtvonxnachykippen.BeispieleIneinemPokerclubspielenalleSpielerdieStrategieX.NunkommteinfremderSpielerzuBesuch.WennseineandereStrategieYnichterfolgreichistunderkeinGeldgewinnt,dannisteswahrscheinlich,dasserzurStrategieXwechselt(ESSistwirksam).WennseineY-Strategieabergewinnbringendist,dannwechselnalleMitgliedervonStrategieXaufY.YistalsodieESS;einzweitesNeumitgliedmitStrategieYwrdenichtbesserfahrenalsdiejetzigenClubmitglieder.Angenommen,inEuropagbeeseineVogelart,diejedenHerbstnachSdenzieht.WennnuneinIndividuumauftritt,dassichinmildenWinterndazuentscheidet,inEuropazubleiben,dannhatesdenVorteil,dassesimFrhlingschondie(Nahrungs-)Ressourcenanzapfenkann,bevordieanderenVgelzurckkehren.MitbesserenFettreservenimKrperausgestattetistdasIndividuumnunfhig,besseralsalleanderenNachkommengrozuziehen.JeneVgel,dieimmernachSdenziehen,sterbenlangfristigaus,dasiesichschlechterfortpflanzen.EbensoistesfreineVogelartvonNachteil,immerinEuropazubleiben;derWinterknnteeinmalsokaltsein,dassvielevonihnensterbenunddanndievomSdenzurckkommendenVgeleinevielbessereAusgangslagebesitzen.DieESSderVogelartbestehtalsodarin,
jenachMildedesWintersinEuropazubleiben.DasLehrbuch-BeispielderESSdasHabicht-Taube-Spiel:IndividuendergleichenPopulationwerdenalsHabichte(aggressiv,stark)undTauben(friedlich,ausweichend)eingeteilt.StteineTaubezueinerreinenTauben-Population,verndertsichnichts.Dasgleichegilt,wennsicheinHabichtzuanderenHabichtengesellt.EsgibtabervierbesondereFlle:EineTaubesttzuHabichten:DadieTaubedenKonfliktenzumBeispielumsFutterausweichtundsoKraftundKrperverletzungeneinspart,fhrtsieeineerfolgreicheStrategie.DazukannsiemittelsDrohgebrdenAggressivittvortuschenundkrftesparendHabichtenRessourcenabluchsen.EinHabichtsttzuTauben:DieTaubenmachendemNeuankmmlingPlatzundberlassenihmkampflosalleRessourcen.DerHabichtisterfolgreich.EineTaubeodereinHabichtsttzueinergemischtenPopulation,inderTaubenundHabichteimkorrektenZahlenverhltnis(entsprichtderESS!)vorkommen.FrdenNeuankmmlingspieltesnuneineRolle,oberhufigeraufeineTaubeoderaufeinenHabichttreffenwird.HatsichdiePopulationaufdie
geeigneteMischungeingependelt,kannesihregalsein,obsichderEindringlingalsTaubeoderalsHabichtverhlt.DiesogenannteBrger-Strategie(engl./franz.bourgeois)entwickeltsichzurESSundsieistweitgehendimmungegenbereinerunausgewogenenHabicht-Taube-ZusammensetzungderPopulation.AlsBrgergilt:Wennmansichverteidigt,istmaneinHabicht;greiftmanjemandenan,verhltmansichwieeineTaube.ImwiederholtenGefangenendilemmaisteinePopulationausTit-for-tat-Strategiennichtevolutionrstabil.WhrendsieresistentgegenberDefektiere-Immer-Strategienist,kannsievonKooperiere-Immer-Strategienunterwandertwerden,wieeinfachanhandderDefinitionmitx=Tit-For-Tatundy=KooperiereImmernachgeprftwerdenkann:TFTgegenKooperiereImmerfhrtzugleichenAuszahlungenfrbeideSpieler,dahergiltBedingung1(E(x,x)>E(y,x))nicht.JedochgiltauchTeil2derBedingung2(E(x,y)>E(y,y))nicht,daauchindiesemFallbeideStrategienzugleichhoherAuszahlungfrbeideSpielerfhren.EinePopulationausKooperiere-Immer-StrategienkannwiederumvonDefektiereImmerunterwandertwerden.
7/29/2019 Behavioral Economics - Script Summary
39/53
Determineevolutionarystablesets:gamesplayedagainandagainwithdifferentpartnersinapopulation
Strategiesnotchosenbyforesightandcalculationbutbyevolutionaryselectionbasedonfitnessandreproduction
DecisioninGroups Socialaspects=largerole,decisiondependingonotherdecisionsresults
dependonbehaviorofthegroupasawhole Searchforoptimalresultsingames:matchingmarket,publicgood,beauty
contest,penaltygameMatchingMarkets
Classicalproblem:matchspecificgoodsandpeople,esp.inseparablegoods Marriageproblem,allocationsofsitestofirms,employerandemployeeallocation Important:sharedfairnessperception Optimalallocationwouldyieldthehighesttotalresultforthegroup Mainlyresearched:optimalmatchinginstitution(centralizedorauctiondesign)
PublicGoodsGame
Inorderforcontributionstobeprivately"irrational"thetokensinthepotmustbemultipliedbyanamountsmallerthanthenumberofplayersandgreaterthanone.Otherthanthis,thelevelofmultiplicationhaslittlebearingonstrategy,buthigherfactors
producehigherproportionsofcontributors.ParticipationratesdifferoverculturesandoversocialrewardsandpunishmentsBeautyContestGame
nplayerschooseanumberbetween0and100.Thatplayerwhoisclosestto2/3oftheaveragewinsacertainamountofmoney.Inthecaseofdrawsthemoneyisdistributedrandomlytothewinners.Nash:Choose0.Stepsofreasoning:LeadingtoNash.
Level0,whichwouldbethepreliminarystepofanyreasoning,translatesintonumbersthatarearbitrarilydistributedovertheinterval,resultinginexpectedvalueof50(50.5ifnumbersarefrom1to100)
Level1reasoningis(2/3)50=33.33 Level2reasoningis(2/3)33.33=22.22 Level3reasoningis(2/3)22.22=14.8(...)
7/29/2019 Behavioral Economics - Script Summary
40/53
ThePenaltyGame
Mostdifficulttoactrandomly Example:jumpdirectionofgoalkeeperandkickdirectionofthescorer Randomizedkicksbutgoalkeepernotoptimizing
Reciprocity Keysocialperceptionfactors:trust,reciprocityandpunishment Reciprocity=responsiblefordevelopmentandstabilityofhumansocieties Expressionsinalllanguages:totfortat(wiedumir,soichdir),onehandwashes
theother(positive),AugeumAuge(negative) Needadynamicenvironment,peopleactsequentially,repeatedly(contrastto
altruism)SequentialGames
Extensiveform,decisionsmadeoneaftertheotherwhileplayersmakemorethanonedecision(e.gcentipedegame)
Choicedependsonsomeoneelseschoice,firstdecisiononlaterdecisions
Reciprocalbehaviorasformofcooperationorsomeofaltruismmodels(inequityaversion)
UltimatumGame:utilitiesderivedfromthematerialoutcomeswithinequityaversion(formula)describesactualobservation
Gameoftrust:nash:Awouldnottrust,butwithinequityaversionformula
trustandcooperation,butspecificlevelie.a.necessary(smallissometimesnotenough)
Atrustbecauseheknowsbwillcooperate Atutwasgutesweilbauchwasgutestut.
DassiehtauswiereziprozittRepeatedGames
Describefrequentsocialbehavior,oftensamesituationoverandover(e.gcooperationattheworkplace,helpneighborsprisonersdilemma)
GametheoreticequilibriumifPDplayed1000times? Nocheatsinrepeatedgame,alwaysdenynocooperationfromtheotherplayer Bothdecidetocooperatebutonlyiftheothercooperates Titfortat:mostsimpleandusefulstrategystartwithcooperationandproceed
bycopyingotherplayerschoicefromthepreviousperiodAltruisticpunishment
Prosperifallcontributetopublicgood(e.g.obeyrules,keepcommonapartmentclean,donotsteal,donothurtsomeone)
Nocontributionpunishment(verbal,material,beating,exclusion Punishmentatacostforpunisher,believablethreatmayfostercontributions(=altruisticpunishment)Membersofthegroupcontributemoreinpublicgoods Butcounterpunishmentpossiblelessfirstpunishmentlesscontributions
7/29/2019 Behavioral Economics - Script Summary
41/53
EfficiencyWagesunderTitforTat Firmspaymorethanminimumwagesmoreandbetterapplicantsandworkers
voluntarilychooseahighereffortlevelehigherprofitforfirmp(e) Chooseefficientwagewhichmaximizesprofit Highereffortlevelswithhighercostswhenpaidhigherwagesprovedbyhigh
stakeexperimentsinRussiastrongpositivecorrelationHowtoreciprocate?
Reciprocity=fundamentalbehaviorinhumansocietiesbutverycomplex! Basicruleissimple,applicationisnot Behaviorkindorhostile,howkind,howtomeasure,favorbychance,favoras
unintendedsideeffect,well-meantcontraryofwell-done? Strictdistinctionbetweenintendedandnon-intendedharmsinlawas
prototypicalsocialinstitution(helpandbreak) WhatdoesBbelieveaboutAsintentions?WhatdoesAbelieve,Bwillbelief
aboutAsintention?Thelatterdetermines,whatAisgoingtodo!(Beliefsoverbeliefs!!!!)
Notoftenpossibletofindoutintentionsandkindnessofaperson Repeatedinteractions:measureintentionsontheachievedresults Chancetotalkincommondilemmasituationsincreasedcooperationrate
Lecture10DecisionAnalysisandResearchTechniques
ExperimentalDataAnalysis Objective,aggregationinfair,neutral,clearmanner Documentsupportiveandopposingresultswithappropriatemeasures Datagathering:processvs.outcomedata,sortsofmeasures Correspondinganalysestodifferentsortsofdata
Typesofdata
Datacharacteristicsappropriatewayofdataanalysis 4Differentmeasurementscales(dataaggregationpossibilities)andstandard
numericalmeasures Statisticaltestssortofdistributionofthedependentvariable
DecisionLabs
Differentdevelopeddecisionlabs:measuringprocessvariables Prominent:Mouselabdecisionprocessiscapturedbytheactualbehaviorof
thedecisionmaker(e.g.eyetracker,chemicalmarkers) Choiceandprocess:compensatorystrategieslikeWADDassumemorecue-wise
processingandnoncompensatorystrategieslikeEBAorTTBassumemore
alternative-wiseprocessing Supportassumptionofanadaptivedecisionmaker Objection:onlysuccessfulimplementationsreported,nodescriptionoffailures
7/29/2019 Behavioral Economics - Script Summary
42/53
WaysofDataAnalysis
Testdifferentrelationsbetweenindependentanddependentvariables Linear,parametricandnonparametricmodels Linearrelations:simplecorrelationorregressionanalysis Nonparametric:usedforotherdistributionalrequirements Qualitymeasures:leastsquares,maximumlikelihoodestimation(evaluateand
comparefitinregardtoobservationaldata) Standardproblemsinmodelselection:dataoverfittingandflatmaxima
OccamsRazor
Universalscientificprincipleformodelfrugalityandsimilarsayings Pluralityisnottobepositedwithoutnecessity Possibleexplanationsbecomeneedlesslycomplex,toomanyspecifics Lesspredictivepower/universalityofthemodel Evaluation:AikakeInformationCriterionorBayesInformationCriterion
ForecastingwithBiases
Behaviorisbasedonexpectationsabouttheconsequencesofbehavior Rationalexpectations:makeunbiasedforecasts,ifothersarebetterinformed
askorbuyinformation/advice Distinguishbetweenexpectationsdevelopedbythedecisionmakersthemselves
andforecastsorprophecies(fromothers,inparticularfromthosenotaffectedbythedecisionsofthedecisionmaker)
Difficult:formationofexpectations(GameTheory:expectationsareendogenized) Often:Behaviorbasedonroutine,onlypartlyaccountedconsequences,sourceof
expectationsmaybeobscure,donotknowthatwereceivedinformationthatguidedourbehavior,impressionsaboutabilitiesandtrustworthiness Explicitforecastbyothersessentialroleforourdecisions(beliefinsuch
propheciesmaychangeourbehaviorinawaywhichmakestheprophecycometrueorpreventitbecomingtrue)
RationalExpectations(RE)useallavailableinformationefficiently?
1. Microeconomictheory:REinvolvingefficientinformationprocessingasfundamentalassumption,peoplehaveperfectforesightorknowdistributionoffuturevariablesidealtheories
2. Realdecisionmakers:noHO,haveanalteregoexperimenterREridiculousbuthowfaraway?Istherebetteralternative?
3. Macroeconomics:realworlddevelopmentsandnotindividualbehavioraggregatedexpectationsfromREtoEfficientMarkethypothesis(weakerrequirement)individualcanrelyonthemarketbecauseofprofessionalspeculatorsresponsibleforefficiencyofpricestructure(marginaltraders)HayekHypothesis;+societyasawholeusesallinformation(decentralizedinformationprocessing)
ImplicitTestsofREandotherHypothesesofExpectationFormation
1. TestREinmacroeconomics:mixedresultsbutstilldominanthypothesis2. Existenceofbubbles(TulipMania,crashes)asargumentagainstREbuttypicallyfoundinassetmarketswithinexperiencedsubjects
7/29/2019 Behavioral Economics - Script Summary
43/53
3. ArgumentforEMH:non-existenceofsuccessfulspeculationmethod,somestudiesrevealpossibilityofsmallprofitsusingappropriatespeculationrulebutdecreasingwithtransactioncost
DirectTestbyAnalyzingExpertForecasts
Noaccurateforecasts,onlyinregionofnearcertainty(weatherfortomorrow) Rule:Naturally,thelongertheforecastingperiodtheworsetheforecasts. Notaccuratenotrational,rationalitydependingoninformationavailableand
properlyandcompleteusage Tounderstandingprocessofexpectation,informationavailablehastobeknowin
advanceandrelevantcriteriafordecisionproblemandinfluencingawardstructure,lossfunctions,goalfunctionalltypicallyunknown
Differentinlaboratoryexperiments:subjectsgetexactlydefinedsetofinformationandwell-definedaward-structure
TheBoundsandLikelihoodHeuristic Modelingofaverageforecasts,basedontimeseriesgestaltcharacterisitics Twocharacteristicsoftimeseriesimportantfortheforecasts:averagevariation
andturningpoints Actuallypredictedchangedependsonlikelihoodthatxtisaturningpoint
o Upswingcasext>xt-1probabilitythatxtisalocalmaximum(Ifinperiodtalllocalmaximaabovext,lowprobabilitythatxtisturningpointIfalllocalmaximabelowxt,i.e.xtverylikelytobeaturningpoint)
o DownswingCase:xt
7/29/2019 Behavioral Economics - Script Summary
44/53
Self-Fulfillingprophecies Predictionthatdirectlyorindirectlycausesitselftobecometrue Falsedefinitionofthesituationnewbehaviororiginalfalseconceptiontrue Perpetuatesareignoferror,prophetseeseventsasproofthathewasrightfrom
thebeginning Thomastheorem:Ifmendefinesituationsasreal,theyarerealintheir
consequenceshowsituationsareperceivedandthemeaningassignedratherbysituationsthemselves
Example:CartwrightMillingvilleBankbreakcyclebyredefiningthepropositionsonwhichthefalseassumptionsareoriginallybased
Roleofself-fulfillingprophecies
Inevitabilityofwar(TheWilltoBelieve):forexampleOedipus Fakeittillyoumakeit(actasif):imitateconfidencesothatastheconfidence
producessuccess,itwillgeneraterealconfidence,methodofovercoming
depression(forcehappinesstobecomereal) MooresLaw:computerhardwaretransistorsincreaseexponentially,doubling
everytwoyearsdescribingdrivingforceoftechnologicalandsocialchange-industrystrugglestokeepupwithit(goalforentireindustry)
PygmalionEffect(Rosenthaleffect,teacher-expectancy-effect):studentsperformbetterbecausetheyareexpectedtoandreverse
CleverHansEffect:horseperformingintellectualtasksbywatchingreactionofhumanobserversrespondingtoinvoluntarycuesinthebodylanguagemakeexperimentsdouble-blind(neitherexperimenternorthesubjectknowswhatconditionthesubjectisinandresponses)orreplacingwithcomputer
Lecture11DecisionSupport
ProcessModels Processvariablescandistinguishbetweendifferentmodels(e.gcompensatoryvs
non-compensatory) Theoreticmodels:processholdsvaluableinformationforempiricalsupport Heuristics=descriptionsofprocesswithsearchrule,stoppingrule,decisionrule Oftenprocessheuristicsasexplanationforobservedbehavior,w/o
predictivevalueincontrasttoforexampleprospecttheory Gamblersfallacy:repeatedtossofacoinoutcomesarestatistically
independent(sameprobability)butmaybelieveafterarowofsuccessiveheadsthatthenexttossmustbeatail
Fallacy:Reasoningthatitismorelikelythatthenexttosswillbeatailthanaheadduetothepasttosses-thatarunofluckinthepastsomehowinfluencestheoddsinthefuturerepresentativenessheuristic
LawofAverages:assumptionthatinsmallsamplestheoutcomesofarandomeventshouldevenout(didntoccursometime,mustoccurnext)
Mostnaturallyobservedsequencesarenotindependent,probabilityofthefuturecanchangebasedontheoccurrenceofpastevents
Probabilitymatching=strategysuboptimaloutcomesservesastheoreticalconceptforbehavior(hastwasbeobachtetVersuchschwaris>rotaberin
7/29/2019 Behavioral Economics - Script Summary
45/53
echtgibtesmehrrotekugelninderurne,verlierstdeswegen)subjectsrespondtothelightsinthesameproportionastheyoccur
Animalsandhumansoftenmatchbaseratesintheirbehavior(Peoplecategorizeeventsaccordingtotheiroccurrenceinatrainingphase,althoughtheywouldachievebetterresultsiftheyalwaysgoforonecategoryifitsbaserateishigher)
ProcessModelswithassumptionsaboutprocessandresultingchoicepredictionsEliminationbyAspects
EliminationbyAspectsHeuristic(EBA)=decisionruleunderboundedrationality Recurringevaluationstilloneonalternativeleftforchoice Canbecombinedwithcompensatorystrategiesafterthechoicesetisinstructed EBA=lexicographicstrategyandnoncompensatory(likeTTB)quick
decisionswithoftensameaccuracyascomplexcompensatorystrategies TTBforbinarychoices(only2alternatives) Concerns:rankingofattributesandsettingoftresholds1. Eachalternativeisviewedasasetofaspects2. Ateachstageanaspectisselected(themostimportantattribute)3. Allalternativesthatdonotincludethataspect(minimumforthisattribute)are
eliminated4. Processstopswhenallbutonealternativeareeliminated
DecisionFieldTheory(DFT)
Computationalmodelofdecisionmaking,modelingdeliberation(berlegungen)asadiffusion(Streuung,Verbreitung)process
Describeshowaperson'spreferencesevolveacrosstimeuntiladecisionisreachedratherthanassumingafixedstateofpreference
Predictdecisionmakingunderuncertainty,resultingbehavior,alsoresponsetimesandcontexteffects
Explainviolationsofstochasticdominance,violationsofstrongstochastictransitivity,violationsofindependencebetweenalternatives,serialpositioneffectsonpreference,speedaccuracytradeoffeffects,inverserelationbetweenprobabilityanddecisiontime,changesindecisionsundertimepressure,aswellaspreferencereversals
Intuitively,ateachmomentintime,thedecisionmakerthinksaboutvariouspayoffsofeachprospect,whichproducesanaffectivereaction,orvalence,toeachprospect.Thesevalencesareintegratedacrosstimetoproducethepreference
stateateachmoment. Thestoppingruleforthisprocessiscontrolledbyathreshold Choiceprobabilityisdeterminedbythefirstoptiontowintheraceandcrossthe
upperthreshold,anddecisiontimeisequaltothedeliberationtimerequiredbyoneoftheprospectstoreachthisthreshold.
Konsequenzenwerdensequentiellberdachtundintegriert,KontinuierlichwerdenmomentanePrferenzstadiengebildet,Aufmerksamkeitschwankt,bersteigteinPrferenzstatuseinenSchwellenwert,wirdeineEntscheidunggetroffen
DFTnicelyexplainscontexteffects(i.e.,attraction,compromise,orsimilarityeffectscompareSessionII)byattentionshifts.
7/29/2019 Behavioral Economics - Script Summary
46/53
DecisionbySampling(DbS)
AsalternativedecisionprocessmodelwithnounderlyingpsychoeconomicscaleslikeinProspectTheory
Anattributessubjectivevalueisconstructedfromaseriesofbinary,ordinalcomparisonstoasampleofattributevaluesdrawnfrommemoryandisitsrankwithinthesample
Samplereflectsimmediatedistributionfromattributevaluesfromcurrentcontextandbackground,real-worlddistribution
Concaveutilityfunctions,losseslargerthangains,overestimatingofsmallprobabilities,underestimationoflargeprobabilities
Evaluationsvaryfromoccasiontooccasion(accordingtoavailableitemsandmemorizedsamples)
Alsoeventsoftimeappeartobeexponentiallydistributed,thus,explainingdiscountingeffectsbytherelativecomparisonsample
DecisionSupportSystems Provideusefulinformationwhenneededandallowquickandefficientdecision
making Questions:Howmuchinformationisavailable,howmuchshouldweuse DataCompression:Informationcanoftenbecompressedwithoutlossandis,
then,easiertoprocess Frogandtoad:trytoeatallkindoflittlemovingobjects,don'tneedmore
informationabouttheirprey,informationusedwhennecessaryandprocessedbysimplereinforcementlearning
FundamentalprobleminDS:questionregardingindividualdemandhelpwithoutdemand=paternalism
Well-being=multidimensionalandconditionsforhappinessdifficulttodefine Onorganizational/expertlevel:criteriaforoptimadecisioneasiertodefine
7/29/2019 Behavioral Economics - Script Summary
47/53
DecisionTrees
Usedinoperationsresearch,decisionanalysiswithgraphormodelofdecisionsandpossibleconsequences
IdentifystrategymostlikelytoreachagoalORtreeasdescriptivemeansforcalculatingconditionalprobabilities
Easytounderstandandtoimplement,simplifieddecisionproblems Example:teachingofstatisticsandusingfrequenciesinsteadofprobabilities Medicaldecisions:strategywithhighsensitivity(e.gsendmostpatientsactually
havingseriousheartproblemincareroom)andhighspecificity(e.gfewpatientsincareroomunnecessarily),fastdecision,transparencyofdecisionsystem
Decisiontreesarefast,transparent,easytoteach,andtheycanbesurprisinglyaccurate
CreditMonitoring:fastdecisionstoavoidlosses,transparent,easilycommunicatedelectroniccreditratingsystemsinloandecisionsinformofhierarchicaldecisiontreesoptimizeandformalizedecisionprocess
Decisiontreepathallowsresultstobescroredagainstcreditpolicy,peergroup Electronicmonitoring:employeeperformancetracking,companyperformance Datamining:decisiontreeaspredictivemodel=mappingobservationsaboutan
itemtoconclusionsabouttargetvalueclassificationtree,regressiontree(leavesrepresentclassificationsandbranchesrepresentconjunctionsoffeaturesthatleadtothoseclassifications)
ErrorsandLearning
Behavior=goal-orientedguidedbylearning(mostimportantmentalfunction) Errorlearninghypothesis:economicvariablescanbepredictedonthebasisof
previousvaluesanderrormargin(adaptiveexpectationsprincipleforecastandrevisionaccordingtonewandcurrentinformation)
Manydifferentformsoflearningascrucialroleforimprovingbehavior Errorsandcorrespondingcopingmechanisms:strengtheneconomicoutputs
Non-associativeLearning
Habituation:progressivediminution(Verminderung)ofbehavioralresponseprobabilitywithrepetitionofastimulus=integrationyoufirstreacttoastimulibutwhenitisneitherrewardingnorharmfulyoudon'trespondafterasecondstimuli(veryspecific)
Sensitization:progressiveamplificationofaresponsefollowsrepeatedadministrationsofstimulus(e.g.rubarmwarmsensationpain(reminderthatitisharmful))adaptiveandmaladaptivelearningprocesses
AssociativeLearning
Classicalconditioning:repeatedlypairingunconditionalstimulus(evokesparticularresponse)withpreviouslyneutralstimulusconditionedresponse
Operantconditioning(Lawofeffect):useofconsequences(reinforcementorpunishment)tomodifytheoccurrenceandformofbehaviordealswithmodificationofvoluntarybehavior
Exploration(play) Behaviorwithnoparticularendinitself,improvingperformanceinsimilarfuture
situations(e.g.catandmice)
7/29/2019 Behavioral Economics - Script Summary
48/53
Involvescost(foranimals):increasedvulnerability,riskofinjury,possibleinfection,energyconsuming
Oppositeofexploitation(Nutzung)learnedability Usedinbeginningofnewtasktofindpossiblesolutions,adaptiveinchanging
environments Importantforeconomicexperiments:allowforlearning,resultingbehaviorcould
substantiallyimproveObservationalLearning
Imitation:repetitionofobservedbehavior Humanscancopythreetypesofinformationsimultaneously:thedemonstrators
goals,actions,andenvironmentaloutcomestuneintosurroundingcultureavoidlosses
Imitation=mostpowerfulprinciplesofboundedrationality Learningbyimitationundervaluedinanimals(instinctswereassumedasdrivers
forbehavior) Neednottoknowwhymodeofbehaviorissuccessful,canstillachieveitanyhow Centralquestions:WhomshouldIimitate?HowshouldIimitate? Rulesofthumb:http://rulesofthumb.org/
Lecture12AppliedDecisionScience
BehavioralFinance Behavioraleconomicsandfinance=closelyrelatedfieldsapplyingscientific
researchonhumanandsocial,cognitiveandemotionalfactorstobetterunderstandeconomicdecisionsandhowtheyaffectmarketprices,returns
Behavioralanalystsconcernedwitheffectsofmarketdecisions,publicchoice(biasespromotingself-interest)
ConsumerBehavior
Variationsinindividualdecisionmakingfindmeansforimprovement Examplesofindividuallyguidingprinciples:self-controlandmentalaccounting Applicationstospecificchoicesituations
BehavioralLifeCycleHypothesis Incontrasttolife-cycletheory(stablelifestyle)enrichedbyself-control,mental
accountingandframing Keyassumption:householdstreatcomponentsoftheirwealthasnonfungible
(nichtaustauschbar,ersetzbar),evenintheabsenceofcreditrationing Wealthdividedincurrentincome(I),currentassets(A),futureincome(F) Temptationtospend:greatestforIandleastforF(behavioralobservation:
saving,bonustreateddifferentthanregularincome,retireddonotdissaveenough)
Plannerconcernedwithmaximizingfunctionoflifetimedoerutilities(U)withZ=pleasureandpain,W=willpower(psychiccosts) Zt=Ut+Wtnosingleutilityfunction(accountspecific),irrelevantfactors(i.e.formofpayment)canmatter,choicescanbestrictlyinthebudgetset
7/29/2019 Behavioral Economics - Script Summary
49/53
Naveruleinassetallocation:1/n-rulethemorechoicesareavailablethemoreassetsareselected,theamountisroughlythesameforall
allocations(VerteilungBetraginetwagleich)
Contradictiontoportfoliotheorywhereassetsshouldbeallocatedtoriskyandnon-riskyprospects
BehavioralPortfolioTheory:efficientdiversificationofinvestmentismeasureddependentonnumberofmentalaccounts
Differentimplicationsofsingleandmultipleaccounts
portfolioresembleslayeredpyramidswherelayersareassociatedwithdifferentaspirations
(Bestrebungen,Erwartungen)EffizienteverteilungvonInvestmentsaufverschiedeneSchichtenzbArmutvermeiden,undindemanderenMentalAccountaufReichtumerabzuzielen.
Anotherapplicationforconsumerdecisionmaking:DoubleEntryMentalAccountingSystemconsumptionexperienceisdecreasedbythepaymentexperienceovertimedifferenttimelydistributionsofpaymentsleadtodifferentlyexperiencedutility
Insomecircumstancescoupling(i.e.,fixedfeeforhealth)andinothersdecoupling
(i.e.,buyersusingcreditcards)increaseoverallexperiencedutility.Hence,hyperbolicmentaldiscountingisoftenexploitedbyofferswithdelayedpaymentswithhighpainsofthepayment
experienceanddepletedutilityofconsumptionexperiencelater.
7/29/2019 Behavioral Economics - Script Summary
50/53
PolicyImplications
Behavioralobservationswithstrongimplicationsforappliedpolicies(applicationpotentials)
ProminentapproachLibertarianPaternalism(formofeconomicguidance) Influencingbehaviorandalsorespectingfreedomofchoice Oftenpeople'spreferencesareunclearandill-formed,andtheirchoiceswill
inevitablybeinfluencedbydefaultrules,framingeffects,andstartingpointspaternalismcannotbeavoided
Steerchoicesinwelfare-promotingdirectionsw/oeliminatingfreedomofchoice(tools:boundedrationality,boundedself-control)
Itisalsopossibletoshowhowalibertarianpaternalistmightselectamongthepossibleoptionsandtoassesshowmuchchoicetooffer
Subjectsratedmeanportfolioequallytoownportfolioandmedianportfoliosignificantlymoreattractivepeoplenotgainmuchbychoosingportfoliosforthemselves
SafeMoreTomorrowPlan:influenceretirementsavings
1. Employeesareapproachedaboutincreasingtheircontributionratesaconsiderabletimebeforetheirscheduledpayincrease.Becauseofhyperbolicdiscounting,thelagbetweenthesign-upandthestart-updateshouldbeaslongasfeasible.(Automaticenrollmentlater)
2. Ifemployeesjoin,theircontributiontotheplanisincreasedbeginningwiththefirstpaycheckafteraraise.Thisfeaturemitigatestheperceivedlossaversionofacutintake-homepay.(Nocomparableloss)
3. Thecontributionratecontinuestoincreaseoneachscheduledraiseuntilthecontributionratereachesapresetmaximum.Inthisway,inertiaandstatusquo
biasworktowardkeepingpeopleintheplan(defaultrule).4. Theemployeecanoptoutoftheplanatanytime.
Mostpeople(78%)whowereofferedtheSMTplanelectedtouseit;virtuallyeveryone(98%)whojoinedtheplanremainedinitthroughtwopayraises,andthevastmajority(80%)remainedinitthroughthethirdpayraise;andtheaveragesavingratesforSMTplanparticipantsincreasedfrom3.5to11.6%overthecourseof28months.
Consider:willingnesstotakeprecautionsdecisiontobuyinsuranceisproductofrecenteventsandnotsystematicinquiryofcostsandbenefits(peopledonot
chooseoptimallyevenwhenthestakesarehigh) Defaultplansaffectoutcomes,existingarrangementisoftenrobustillustration
byautomaticenrollmentphenomenonincontrasttoopt-inmodelsthereisahigherenrollmentinopt-outmodels(onlyfewopt-out)
ActualbehaviorishighlysensitivetoplandesignfeaturesMinimalpaternalismistheformofpaternalismthatoccurswheneveraplanner(privateorpublic)constructsadefaultruleorstartingpointwiththegoalofinfluencingbehavior.Solongasitiscostlessornearlycostlesstodepartfromthedefaultplan,minimalpaternalismismaximallylibertarian.Thisistheformofpaternalismthatisdescribedasinevitable.
Unsureofwhatchoiceswillpromotewelfare,aplannermightrejectdefaultplansorstartingpointsentirelyandforcepeopletochooseexplicitly(requiredactivechoices).
7/29/2019 Behavioral Economics - Script Summary
51/53
MoreAggressiveformsofPaternalism Defaultplanaccompaniedbyproceduralconstraintsdesignedtoensurethatany
departureisvoluntaryandentirelyrationalitisnotcostlesstodepartfromthedefaultplan
Justificatio