Melbourne Institute Policy Briefs Series
Policy Brief No. 4/15
Beyond Nudge:
The Potential of Behavioural Policy
Reuben Finighan
THE MELBOURNE INSTITUTE IS COMMITTED TO INFORMING THE DEBATE
Beyond Nudge:
The Potential of Behavioural Policy*
Reuben Finighan
Melbourne Institute of Applied Economic and Social Research
The University of Melbourne
Melbourne Institute Policy Brief No. 4/15
ISSN 2201-5477 (Print)
ISSN 2201-5485 (Online)
ISBN 978-0-7340-4385-6
July 2015
* I thank Deborah Cobb-Clark, Paul Jensen, Guyonne Kalb and Mike Pottenger for useful
comments and suggestions. For correspondence, email <reuben.finighan@unimelb.edu.au>.
Melbourne Institute of Applied Economic and Social Research
The University of Melbourne
Victoria 3010 Australia
Telephone (03) 8344 2100
Fax (03) 8344 2111
Email melb-inst@unimelb.edu.au
WWW Address http://www.melbourneinstitute.com
Melbourne Institute Policy Briefs Series
The Policy Briefs Series is a collection of research publications that
examines current policy issues and provides an independent platform
to examine pertinent issues in public debate.
Log onto melbourneinstitute.com or follow @MelbInstUOM for more
details on this and other series from Australia’s leading and longest
standing research institute in the field of economics and social policy.
Abstract
Policymakers often mistakenly see behavioural policy as synonymous with “nudging”. Yet
nudges are only one part of the value of the behavioural revolution—and not even the lion’s
share. This policy brief looks at the full potential of behavioural policymaking. Using
examples in cigarette regulation, retirement savings, and poverty, it makes three arguments:
1. Trade-offs between social welfare and individual freedom of choice affect both
nudges and conventional policy tools (like mandates, taxes and transfers). Nudges are
not special tools that avoid such trade-offs.
2. Nudges typically sacrifice social welfare in favour of individual choice, and for this
reason many well-known nudges will be unattractive in Australia. Some nudges can,
however, play a complementary role by fine-tuning conventional policy regimes.
3. Conventional policy tools will continue to be the most powerful tools for countering
behavioural biases, and have the most promise for driving major, behaviourallyinformed reforms in Australia.
Policymakers should not ask “can we nudge this?”, but should instead ask how behavioural
evidence changes the way they think about all the options in the policy toolkit.
JEL classification: D03, D60
Keywords: Behavioural economics, behavioural policy, public policy, nudges, regulation,
paternalism
2 Behaviouraleconomicsisbeinghotlydebatedinpolicymakingcirclesacrosstheworld.In
Australiatoo,ithasbeenwelcomedwiththeblendofenthusiasmandscepticismthatisrightly
duetoanyparadigmshift.Whilethefieldhaswonsubstantialacademiccredibility—with
economistswidelyacceptingthatbehaviouraleconomicscanimproveuponthestandard
analysisinimportantways—itfacesanewsetofcontentiousquestions.Whatarethefield’s
practicalpolicyimplications?Doesbehaviouralpolicymakinghavethepotentialtocontribute
bigideasandstimulateimportantreforms,orwillitmerelytinkerattheedges?
Intryingtoanswerthesequestions,manypublicservantssubstitutethemwithanother:“what
isthepotentialofnudges?”Thisisanerror,andapotentiallyseriousone.Nudgesarenot
synonymouswithbehaviouralpolicymaking,butrepresentjustonepartofitsvalue.Theclose
identificationofnudgeswithbehaviouraleconomicsislargelyahistoricalquirk:Sunsteinand
Thaler’sgroundbreakingNudge(2008,YaleUniversityPress)wasthefirstbooktopopularise
behaviouralpolicymaking,andhaspowerfullyframedthediscussion.
Thispolicybriefaimstotakeamorecompletelookatthepotentialofbehaviouralpolicymaking,
payingspecialattentiontopolicytoolsbeyondthenudge.Nudges,itargues,largelytinkeratthe
edges.Majorreforms,especiallyinAustralia,aremorelikelytocomethroughusingbehavioural
evidencetoinformtheuseofconventionaltools,liketaxes,transfersandmandates.Behavioural
policymakingdoesnotmeanasking,“canwenudgethis?”Muchlikeconventionalpolicymaking,
itinvolvesweighingeachoftheoptionsinthepolicytoolkittofindthetoolthatbestbalances
thevalueofindividualchoicewiththemagnitudeofpotentialwelfaregains.Whatisnewabout
behaviouralpolicymakingisthatittakestherichbodyofbehaviouralevidenceintoaccount.
Thebriefbuildsthiscaseuponthreekeyinsights:
1. Trade‐offsbetweenagencyandwelfareaffectbothnudgesandconventionaltools.
Nudgesarenotspecialtoolsthatcanachievemajorwelfaregainswithoutinterfering
withchoice.Inpractice,theyareaffectedbythesametrade‐offsasconventionaltools.
2. Nudges,whichSunsteinandThalerpointouthavebeeninuselongbeforeNudge,
havegenerallybeenbestsuitedtodrivingmodestchangesinbehaviouratlow
cost.Theytendtoplayacomplementaryrole,extendingtheeffectofconventional
policyregimes.Onlyunderidealconditionsaretheypowerfulbythemselves.
3. Conventionalpolicytoolslikemandates,taxesandtransfershavebeen,andwill
continuetobe,themostpowerfultoolsforcounteringbehaviouralbiases.The
major,prosperity‐enhancingreformsthatbehaviouraleconomicswilldriveinthe
comingdecadeswill,forthemostpart,employconventionalpolicytools.
3 Theoreticalfoundations:Behaviouralbiasesandmarketfailure
Thebehaviouraleconomicsliteratureexplainsobservedbehaviourastheinteractionoftwo
factors:normativepreferences(whattheindividualactuallywantstoachieve,orwhatwould
actuallymaximisetheirwelfare),whicharethendistortedbybehaviouralbiasestoform
revealedpreferences(theactualchoicesobserved,whichoftenfailtomaximisewelfare).The
literaturedescribeshundredsofdifferentbehaviouralbiases,andamultitudeofthemcould
applytoagivenpolicyproblem.Forexample,thetypicalindividualholdsanormative
preferencetosaveenoughtohaveacomfortableretirement.Intheabsenceofasupportive
policyregimelikesuperannuation,thispreferencemaybedistortedbypresentbias,analytic
errors,forgetfulness,inattention,andmanybiasesbeyondthese.Foracomplex,intertemporal
choicelikesavings,thislistofbiasescouldgoonandon.
Biasesareusuallyorganisedintothreecategories:first,imperfectoptimisation,comprisingthe
biasesthatpreventusfromcalculatingtheconsequencesofourchoices,suchasthebiastoward
forecastingexponentialgrowthasthoughitwerelinear;second,boundedself‐control,including
psychologicalphenomenalikeprocrastination,emotionsandimpulses,whichpreventusfrom
implementingournormativepreferences(andguaranteethefailureofourNewYear’s
resolutions);andthird,non‐standardpreferences,wherewesimplywantthingsthat
conventionaleconomistsdonotexpectustowant,suchastoavoidexperiencingpainfullosses
evenifthismeanswinninglessoverall.Notethislastcategoryincludessomephenomenathat
canbegenuinedesires,likeconformitywithsocialnorms,althoughthesetooareregularly
exploited—forexample,bymarketerspayinghandsomeactorstosmokecigarettes.
Whenbehaviouralbiasesdistortnormativepreferences,thedecisionswemakemaybefarfrom
welfare‐maximising.ScholarslikeBrigitteMadrianarguethatbiasesshouldthereforebe
regardedasanewclassofmarketfailure.1Behaviouralmarketfailures,likeotherkindsof
marketfailure,openuppotentialroomforgovernmentintervention.Wherepolicyiswell‐
designed,itmayimprovetheefficiencywithwhichthemarketproducessocialwelfare.
Conventionalpolicytoolstypicallyaddressmarketfailures,orindeedachieveotherpolicygoals,
byshiftingmonetaryincentivesorrestrictingwhatpeoplecanchoose.Theyincludetaxesand
subsidies,mandatesorregulations,andvariouskindsofmaterialtransfers.Nudges,onthe
otherhand,eschewmonetaryincentivesorstrongrestrictionsonchoice,andinsteaddepend
uponsmallchangesin“choicearchitecture”thatgentlyguidechoicesinthedesireddirection.
Nudgesareonlyeffectivebecauseofthewaysinwhichreal‐worldpeoplearedifferenttothe
4 rational,self‐interestedagent—inRichardThaler’sterms,goodnudgeswillaffectthereal‐world
“Humans”ofpsychology,butnottheperfectlyrational“Econs”ofeconomics.Notethat
conventionaltoolscanbestrengthenedbehaviourally,suchthatanEconwouldnotnoticethe
change(e.g.reframeanincentiveasaloss,morepainfulforHumanssubjecttolossaversion).
Importantly,abehaviouralpolicyproblemdoesnotnecessarilycallforanudgesolution.Aswe
shallsee,nudgesandconventionalpolicyinterventionscanbeusedtoresolveeither
behaviouralorconventionalpolicyproblems.Table1illustratesthiswithamatrixoffour
possibilities—althoughnotethatpolicyproblemsareoftenamixoftheconventionalandthe
behavioural,andinterventionsareoftenamixofnudgesandconventionalapproaches.
Table1:Therangeofapproachesavailabletothebehaviourally‐informedpolicymakertoresolve
bothconventionalandbehaviouralpolicychallenges
Conventionalpolicyproblem
Behaviouralpolicyproblem
Conventional
intervention
E.g.alignteacherinterestswith
studentlearningbyprovidingapay
bonusforbetterstudent
performance.
E.g.toovercomesavingsmyopia,use
amandate(e.g.superannuation)ora
financialincentivetoincrease
savings.
Nudge
intervention
E.g.alignteacherinterestswith
E.g.toovercomesavingsmyopia,use
studentlearningbyprovidingsocial
anudge‐likeinterventionsuchaspre‐
rewards,likerecognitionandthanks. commitmenttosaving.
InsightOne:Bothnudgesandconventionaltoolsaresubjecttotrade‐offs
betweenagencyandwelfare
Nudge‐likepoliciesarenotnew.Policymakershavelongintuitedthatpeoplearenotperfectly
rational—thatframingisimportant,thatwefollowsocialnorms,thatwetendtostickwith
defaults,andsoon.Whatrecentadvancesinbehaviouraleconomicshavedoneisdramatically
enrichedourunderstandingofhownudgesworkandhowtheyshouldbedesigned—and,
notably,increasedtheirsalience.Asimportantly,orperhapsevenmoreso,behavioural
economicshasimprovedourunderstandingofwhenconventionaltoolsshouldbeusedand
howtheyshouldbedesigned.Why,then,hasalltheemphasisbeenonthenudge?
Theanswerispartlypolitical.SunsteinandThalerwroteNudgewithintheuniquepolitical
contextoftheUnitedStates,wherethelibertarianismofpoliticalentitiesliketheTeaPartyisa
powerfulforcetobereckonedwith.Theypromotethepoliticalphilosophyof“libertarian
5 paternalism”asawayofresolvingthetensionbetweenlibertariansandprogressives—onthe
onehand,thedesiretopreserveindividuals’agencyinbeingabletochoosewhattheyprefer,
andontheother,thegrowingrecognitionthatindividuals’choicesarestronglyshapedby
cognitivelimitationsandcontextualfactors,withsometimessevereconsequencesforwelfare.2
Nudgesaresupposedtoinfluencechoicesinwaysthatincreasewelfarebutthatavoidcoercing
choices,lettingpolicymakerssteerasafecoursebetweenthesewarringtribes.
Iftheideaofavoidingpoliticallychargedtrade‐offsbetweenwelfareandagencysoundstoo
goodtobetrue,unfortunatelythisisbecauseitistoogoodtobetrue.Thedistinctionbetween
nudgesandconventionaltoolsturnsouttobeexaggerated,withnudgessubjecttomuchthe
sametrade‐offs.Taketwoexamples:first,theSaveMoreTomorrownudgescheme,whichhas
beentoonon‐coercivetobeeffective;andsecond,effortstoreducesmoking,anexampleof
nudgesthatareeffectivepreciselybecausetheyarecoercive.
Savingsbehaviourisstronglyaffectedbyarangeofbiases,andthecelebratedSaveMore
Tomorrowschemeusesasetofcleverlydesignednudgestocounterthem.Thescheme’smost
importantinnovationisadefaultautomaticescalationinsavingsovertime,soindividualswill
savemoreinfutureyearswithoutexperiencingalossinincometoday.Inanearlytrial,average
savingsreachedalmost14%.Yetactualimplementationhasbeenlessimpressive:amajorityof
employersofferedautomaticescalationby2011,butonly11%ofemployeeschosetoenrolin
theplan.Theescalationdefaultwas3%over3years,withnofurtherincreasesthereafter.The
US‐widesavingsrateincreasedbyafeeble0.33%,3atinyfractionoftheeffectachievedby
schemeslikeAustralia’ssuperannuation—withenormousimplicationsforwelfare.
Theproblemwithagency‐preservingnudgeslikeSaveMoreTomorrowisthattheycontinueto
relyuponthediscreditedrationalmodel.Peoplemustchoosetoparticipateandmustchoose
therightsettings,butthesearechoicesthatthebehaviouralevidenceclearlyshowswecannot
reliablymake.YetifanudgeschemelikeSaveMoreTomorrowwasstrengthenedenoughtobe
effective—say,allemployersarerequiredtoplaceemployeesinopt‐outsavingsescalation
schemes,withahighdefaultsavingsrate,andpenaltiesforthosewhodoopt‐out—thenit
effectivelybecomesacoercivemandatewithacostlyescapeclause.Evidencefromthe
behaviouralsciencessuggeststhateventhisschemecouldbewelfarereducing,becausepeople
whochoosetoopt‐outoftendosoirrationally,withsometimesenormouswelfarecosts.
Anoppositecaseisillustratedbycigarettesmokingnudges,whicharesubstantiallymore
coercivethantheSaveMoreTomorrowscheme.Plainpackagingrulesmakecigarettes
6 unattractive;theQuitcampaignprovidesservicesthatpromptsmokerstoplanaheadandmake
commitments;removingcigarettesfromviewandsellingthemataseparatecountermakes
themhardertoaccess;healthwarningsplacedatthepointofsalemakeforatimelyand
emotionallypowerfulreminder;andsocialstigmapitssmokersagainstthepowerfulforcesof
socialconformity.Peoplearestillfreetoexercisethechoicetosmoke,buteverynudgeinthe
bookispointingintheotherdirection.Thoughthisisanudgeregime,itisclearlyhighly
coercive—anditispreciselythehighcostsimposedonsmokersthatmakestheregimeeffective.
InsightTwo:Nudgesarebestsuitedtodrivingmodestchangesinbehaviour
atlowcost,andcanenhancetheeffectsofconventionaltools
Nudgesarenotspecialtoolsthatevadeagency‐welfaretrade‐offs.Theyare,nonetheless,very
usefulundertherightconditions.Thispolicybriefwillnotattempttoimproveuponthemany
excellentoverviewsthatalreadyexist.Amoreimportantquestionforthebriefisthis:given
nudgessometimescompletelyshiftaggregatebehaviour,andonotheroccasionsareentirely
ineffective,whenshouldpolicymakersbeparticularlyenthusiasticaboutthepotentialof
nudges?
Awell‐designednudgemaybeaparticularlypowerfulpolicyoptionwhere:
1. individualunderlyingpreferencesarerelativelyhomogeneousandarealreadyaligned,
oronlyweaklymisaligned,withthedesiredbehaviour;
2. thecognitivebiasesthatdistortunderlyingpreferencescanbeeasilyovercome,orused
tothepolicymaker’sadvantage;and
3. eitherindividualchoicecannotbeavoided,oragencyisvaluableandthebenefitsof
preservingagencyarenotoutweighedbywelfarecosts—ineitherofthesecases,the
useofstrongerpolicyoptionsisnotpossible.
Organdonationnudgesaretheposterchildforsuchsuccess:(1)alargemajorityofpeoplehold
normativepreferencesthatarealignedwiththesociallyoptimaldecisiontodonate;and(2)the
decisiontodonateisnotfixedbypowerfulbiases,butislargelytheresultofadefaultbiasthat
policymakerscaneasilyharness.Theseconditionsallowanudge—aswitchfromanopt‐intoan
opt‐outdefault—tocompletelytransformoutcomes,suchthatamajorityratherthanaminority
choosetodonate.Onpoint(3),anudgeissuperiortoamandateherebecauseagencyiscrucial:
peoplecannotreasonablybeforcedtodonatetheirorgans,andthebenefitsofpreservingchoice
arejudgedtooutweighthecosts.
7 Amoremodestexampleisaprogramaimedatimprovingdrugprescriptionchartstoreduce
medicalerrorsinhospitals:(1)doctorsandnursesholdanormativepreferencetoenter,and
read,therightdrugandprescribeddosagefromtheprescriptionpaperwork;(2)theproblemof
illegiblehandwritingcaneasilybeovercomewithbetterdesignedforms;and(3)thisisacase
wherethebehaviourisalreadymandatedtothehilt—medicalpractitionersfacestrong
incentivestodeliverthecorrectdrugsanddosages.Nudgescanhelpstaffcomplywiththe
mandate(and,formostpractitioners,theirownpreferences).Indeed,thenewchartsvirtually
eliminatederrorscausedbyconfusionbetweenmilligrams,or“mg”,andmicrograms,or“mcg”.4
Inmanycasestheseconditionsarenotmet,andnudgeswillproducerelativelysmallgainsat
themargin—albeitusuallyatverylowcost.Theywillbeauseful,butlimited,partofawider
policymix.
InsightThree:Conventionalpolicytoolswillcontinuetobethemost
importanttoolsforcounteringmanybehaviouralbiases
Policymakersdonottypicallythinkofconventionalpolicyinstrumentsascentraltobehavioural
policymaking,yettheyshould.Thissub‐sectionprovidesexampleswhereabehaviouralanalysis
ofapolicyproblemfavourstheuseofaconventionalpolicytool,withexamplesprovidedunder
mandates,standardincentives,andtransfers.
Mandatesandregulation
Mandateshavebeenthetoolofchoiceformanyimportantwelfareoptimisationproblems—
seatbelts,bicyclehelmets,cigaretteadvertising,rentalpropertystandards,fairmortgage
contracts,waterfluoridation,milkpasteurisation,superannuation,minimumwages,andmuch
beyond.Regulationsarethemostpowerfulinstrumentsforshapingbehaviour,withthelevelof
interferenceincreasingastheymovefromrequirementsorrestrictionsimposedonmarketing
(e.g.paydayloanadvertising),toproductattributes(e.g.unfairinterestrates),andfinally
behavioursorwholeproducts(e.g.paydaylending).
Theirstrengthgivesthemhighpotentialforbothbenefitandharm,soitisunsurprisingthat
theyareoftencontroversial.Benefitismostlikelywherenormativepreferencesarerelatively
homogeneoussothataone‐size‐fits‐allmandatecanproducebenefitsformostpeople;where
choiceerrorsarecommonandcostlyintheabsenceofamandate;andwhereagencyisnottoo
8 highlyvalued.Harmislikelywherenormativepreferencesareheterogeneous,andindividuals
areblockedfromundertakingbehavioursthatgenuinelyenhancetheirwelfare.
Mandatescanimprovewelfareincasesofboundedself‐controlorimperfectoptimisation.
Beshearsetal.(2014)provideanexampleofhowmandatescanaddressboundedself‐control:
theystudyamodelofretirementsavingswhereindividualsaresubjecttodifferentdegreesof
presentbias,andthereforedifferentlevelsoflikelihoodforwithdrawingfromtheirsavingstoo
early.Whatlevelofpenaltyonearlywithdrawals,todiscouragethebehaviour,wouldmaximise
welfare?Theiranalysisfindsthewelfare‐maximisingpenaltyis100%—thatis,withdrawing
shouldbeimpossible.Alowerpenaltydoesallowasmallnumberofpeopletowithdrawearlyin
arationalway,buttheirwelfaregainsarecompletelyoverwhelmedbythewelfarelosses
experiencedbythosewhoirrationallywithdrawduetopresentbias.5
Thisisoneweaknessofnudge‐basedsavingschemesthatallowopt‐outs,likeSaveMore
Tomorrow:thepeoplewhoopt‐outofsuchschemesmaybelievetheycandobetterwithoutit,
butthisistypicallynotthecase.Onaverage,theyexperiencesignificantwelfarelosses.Indeed,
theUSInternalRevenueServiceimposesa10%penaltyonearlysavingswithdrawalfrom
401(k)plans,butevenso,forevery$1thatissaved,$0.40iswithdrawnprematurelyat
substantialcost.
Australia’ssuperannuationpolicyusefullyhighlightsanotherpropertyofasuccessfulmandate:
popularconsent.Becausemandatesareopentochargesofpaternalism,theydependheavily
upondemocraticlegitimacy.Mandatesaremostlikelytosucceedwhentheyareunderstoodto
havebeenimposedbyacommunityuponitself,withoutmarginalisingunderrepresentedsub‐
groups;andwhentheyenactmostindividuals’normativepreferences,withouteliminatingany
cherishedchoices.Inthecaseofsuperannuation,mostpeopledowishtosaveadequatelyand
over80%ofAustraliansareinfavourofthescheme.
Standardincentives
Incentivesaremostusefulwhenchoicesareshapedbyself‐controlbiases.Theliteratureisthus
focusedheavilyuponpolicyproblemslikethepromotionofhealthybehaviours,suchasquitting
smokingorgoingtothegym;educationaloutcomes,likestudentperformance;andlabour
marketoutcomes,likeincentivesforfindingajoborforworkingharder.Somepeoplerecognise
thattheylackself‐control,andwillwillinglyimposecostsuponthemselvesinwaysthatwould
bebizarreforarationalagent.Forexample,studieshavefoundthatdataentryworkers
9 willinglysubjectthemselvestopaymentschemeswithhighpotentialdownside,butnoobvious
upside—forexample,theythreatenthemselveswithsubstantialmonetarylossiftheyfailto
reachadailydesiredtarget,say,1,000entriesinaday.Peoplechoosesuchschemesas
commitmentdevices,andtendtoworkharderundersuchself‐imposedthreats.
Letusreturntocigaretteaddictionasaprimeexampleofindividualsstrugglingtoimplement
theirnormativepreferences.Around70%ofsmokerswouldliketoquitatanygivenmoment,
butonly3%succeedinagivenyear.Researchsuggeststhatsmokingimposescostsonone’s
futureselfofaroundUS$35perpack,afigurethatisvastlyhigherthanwhatpeopleeverpayfor
apack.Eventhemostcoercivenudgescannotprovideasignalthatreflectsthiscost.Cigarette
taxes,ontheotherhand,canprovideamotivationalcountertoaddictionthatstudiessuggestis
oneofthemoreeffectivepartsofapolicymixforreducingsmoking.6
Whileataximposeshighercostsonsmokersandhasbeencriticisedasregressive,behavioural
evidencesuggeststhatitmayreducesmokingsufficientlythatitslong‐runeffectonmost
individualsisnetpositive.Acigarettetaxcanthereforebeprogressiveratherthanregressive.
BehaviouralstudiesofUSandCanadiansmokershavedemonstratedthatincreasingtaxeson
cigarettescanactuallyincreasethewell‐beingofcigarettesmokers,byassistingthemtoexert
self‐control.7
Incentivesarealsousedtoencouragepro‐socialbehaviour,andbehaviouralevidenceturnsout
tobecrucialforsuccessfuldesign.Incentivesaremessages.Themerepresenceofanincentive,
whichattachesapositiveornegativesignaltoachoice,canbemoreimportantthanthe
magnitudeoftheincentive.Individualsinterpretthismessagewithinaparticularsocialcontext,
andincentivesthusinteractwithsocialpreferencesinimportantways.Carelessuseof
incentivescanbackfirebydisplacingand“crowding‐out”beneficialsocialpreferences,while
well‐designedincentiveswillinstead“crowd‐in”andstrengthensocialpreferences.
Whethercrowding‐outorcrowding‐inpredominatesisdeterminedbytheinterpretationofthe
incentives.Incentivesaremorelikelytobackfirewhentheyareseenasmanipulativeor
imposedfromabove;whentheyarearbitrary,difficulttoactupon,orunfair;whenthey
encourageselfishmaximisationbehaviour;orwhentheysuggestthatanactivityisunpleasant
andthereforemustbecompensated.Individualsaremorelikelytoseeanincentiveas
legitimateiftheyunderstandandendorseitspurpose,andiftheyperceivetheincentiveas
imposedbytheirpeersforthegoodofthegroupasawhole.Pro‐socialpreferencescanbe
strengthenedwhensmallincentivesareusedtoreiterategroupnorms.
10 Finally,notethatincentivestendtobelesseffectiveforaddressingimperfectoptimisation
biases.Anexcellentexampleisthedesignofanaircraftcockpit:earlydesignswereextremely
complex,withhundredsofbuttons,leversandinstruments.Clearlypilotshaveeverypossible
incentivetoflywell,giventheirverysurvivaldependsuponit—andyetthecomplexityofthe
environmentledtocriticalerrors.Insuchacase,furtherincentivesareclearlyfutile.
Optimisationbiasesarebettercounteredbynudges(suchasmakingthechoiceproblemeasier,
ortheincentiveseasiertounderstand),aswellasmandatesandinsomecasestransfers,
althoughincentivesmaybeusedtosuggestausefulstrategyforsolvingacomplexproblem(e.g.
smallincentivesforinvestorstodiversifytheirportfolio).
Transfers
Therationalmodelholdsthattransfers,suchasthoseprovidedasunemploymentbenefits,
shouldcomewithasignificanthasslecostinordertodisincentivisetheiruseandpreventmoral
hazard.
Acognitivescarcitymodelsuggestssomethingquitedifferent.8First,transfersshouldbe
evaluatednotonlyfortheirwell‐beingbenefitsbutalsotheirimpactoncognitiveresources.
CashinfusionsthatliberateIndianfarmersfrompoverty,forexample,produceincreasesinIQ
ofnearlyafullstandarddeviation—adramaticeffectthatwouldbetheenvyofanyeducational
system.Second,thehasslecostofmeetingtherequirementsofwelfareservicescanperversely
imposea“cognitivetax”thatonlyaddstothecognitivechallengesofpoverty.Policymakers
shouldbecarefultodesignschemesthatdonotposeexcessiveburdensonalreadycomplicated
lives.9
Wherebehaviouralevidenceandtherationalmodelagreeisonhowindividualsspendsuch
transfers:studiessuggestthatspendingisrelativelyrational.Thenumberofvicegoods
purchasedasaproportionoftotalspendingtendstodecreasewhentransfersareprovided,
whilespendingonpolicymakers’preferredgoods,likenutritiousfoods,tendstoincrease.10
Utah’shomelessnessstrategymakesforaninterestingcasestudy.In2005,“no‐stringsattached”
housingwasfirstprovidedtothechronichomeless,withoutrequiringdrugtestsandwith
minimalrent.Therewerefearsthattheprojectwould“incentivisemooching”,asperthe
rationalmodel.Utah’schronicallyhomelesspopulationhassincefallen90%,andbytheendof
2015“maybevirtuallygone”.11Mostrecipientssuccessfullymakerentpayments,andcostsare
11 muchlowerthanthatofprovidingservicesforthechronicallyhomeless.Whilenobehavioural
analysishasyetbeenperformed,theprojectmayhavepowerfulcognitiveeffects:providing
housingsimplifiesthelivesofthehomeless,providesabufferagainstshocks,andfrees
cognitiveresourcesforchallengingtaskslikequittingdrugsandfindingemployment.
Conclusion
Behaviouraleconomicshaspeeledawaytheaxiomsoftherationalmodeltorevealthelayersof
psychologicalcomplexitylurkingunderneath.Policymakerscanusethisricherunderstandingof
humanpsychologytoimprovethedesignofnudges,mandates,incentivesandtransfers.These
toolsareessentialpartsofthepolicymix,andinpracticeareoftencomplementary.
Conventionaltoolslikelyprovidethemostsubstantialpotentialforsignificantwelfare‐
enhancingreforms,whilenudgestendtobemosthelpfulforfine‐tuninginstitutionsinorderto
bringefficiencygainsatlowcost.Whichevertoolisemployedforagivenpolicyproblem,
policymakersshouldseektobalanceindividualagencywithsocialwelfare.
Unlockingthispotentialwillrequirenewpracticesofpolicydesignandevaluation,focusedon
generatingandapplyingnewevidence.Behaviourdepartsfromtherationalmodelinmyriad
ways,andthereisanequallylargearrayofoptionsforsettingbehaviourontherightcourse.
Suchawideandcomplexfieldofpossibilitymustbenavigatedwithaninterdisciplinarymixof
methodsfromtheappliedbehaviouralandeconomicsciences,toidentifythecausalrelations
thatunderliepolicyproblemsandthatmakeinterventionswork.Thisisbehavioural
policymaking—theweighingofallpolicyoptionsinlightofthefullbodyofevidence.
1Madrian,B(2014)“Applyinginsightsfrombehavioraleconomicstopolicydesign”,WorkingPaperno.
20318,NationalBureauofEconomicResearch.
2Sunstein,C&Thaler,R(2008)Nudge.YaleUniversityPress.
3Benartzi,S&Thaler,R(2013)“Behavioraleconomicsandtheretirementsavingscrisis”,Science,
339(6124),1152–1153.
King,D,Jabbar,A,Charani,E,Bicknell,C,Wu,Z,Miller,G,Gilchrist,M,Franklin,B&Darzi,A(2014)
“Redesigningthe‘choicearchitecture’ofhospitalprescriptioncharts”,Healthservicesresearch,4(12).
5Beshears,J,Choi,J,Clayton,C,Harris,C,Laibson,D&Madrian,B(2014)“Optimalilliquidity”,paperfor
theRetirementResearchConsortium,September26.
6E.g.seeLewit,E&Coate,D(1982)“Thepotentialforusingexcisetaxestoreducesmoking”,Journalof
HealthEconomics,1(2),121–145.
7Gruber,J&Mullainathan,S(2005)“Docigarettetaxesmakesmokershappier?”,TheBEJournalof
EconomicAnalysis&Policy,5(1).
8Mani,A,Mullainathan,S,Shafir,E&Zhao,J(2013)“Povertyimpedescognitivefunction”,Science,
341(6149),976‐980.
9Mullainathan,S&Shafir,E(2013)Scarcity:WhyHavingTooLittleMeansSoMuch.Macmillan.
10Evans,D&Popova,A(2014)“Cashtransfersandtemptationgoods:areviewofglobalevidence”,World
BankPolicyResearchWorkingPaper6886.
11Rascon,J&NBCNews(2015)“Utah'sstrategyforthehomeless:givethemhomes”,NBCNews,May3.
4
12