Upload
raymundo-gama
View
221
Download
0
Embed Size (px)
Citation preview
7/28/2019 Second Order Decisions
1/44
CHICAGOJOHNM.OLINLAW&ECONOMICSWORKINGPAPERNO.57
(2DSERIES)
Second-OrderDecisions
CassR. Sunstein and EdnaUllmann-Margalit
THELAW SCHOOLTHEUNIVERSITYOFCHICAGO
7/28/2019 Second Order Decisions
2/44
Second-OrderDecisions
CassR. Sunstein*andEdnaUllmann-Margalit**
I. Introduction
Inmaking decisions, people sometimes calculate the costs andbenefitsofalternativecoursesofaction, andchoosetheoption thatmaximizes net benefits, however these may be described andunderstood. Certainly this is a conventional picture of practicalreasoninbothprivateandpubliclife.1When decidingwhat tobuy,
where totravel, whether tosupportlegislation, orhow tovotein adisputeoverconstitutional rights, peoplemight seem to proceed inthisway. Acommon ideaaboutdecision-making isthat agents aretypicallyinthebusinessofmaximizingoroptimizing. Atthe timeofchoice, thisistheirbasicmethodandtheirbasicgoal.
Amomentsintrospectionshowsthatthispictureisinaccurateoratleasttoosimple. Thecostofdeliberationisoftenhigh. When andbecause the stakes are extremely low, people may use simplifyingstrategies;when and because the stakes are extremely high, peoplemay seek approaches that relieve them of the burden of or theresponsibilityforchoice. Sometimescalculationofcostsandbenefitsofalternativecoursesofaction isexceedinglydifficult, orinanycasetedious and not worthwhile. Every bureaucrat knows that cost-benefitanalysismayfail cost-benefit analysis, and almost everyone,bureaucratornot, issometimeswillingtodoagreatdealinordertoreduce or to eliminate the burdens of decision. Often agents verymuchwantnottomake(particular)decisions. Oftentheyknow thattheywillwantnottomakedecisionsevenbeforetheyundertaketheparticularcalculationsinvolvedin thoseparticulardecisions. Part ofwhat it means to optimize is to try to reduce the burdens ofjudgment, afactthatcanleadpeoplenot tocalculateatall, ortodoso in a sharply truncated fashion. Noncalculative or truncated
*
KarlN. LlewellynDistinguishedServiceProfessor, UniversityofChicago.**Center for Rationality and Interactive Decision Theory, The HebrewUniversityof Jerusalem. We aregrateful toSaulLevmore andRichardPosnerforvaluablecommentsonapreviousdraft.1See, e.g,, GaryBecker, AccountingforTastes(1996).
7/28/2019 Second Order Decisions
3/44
CHICAGOWORKINGPAPERINLAWANDECONOMICS 2
decisions can in turn have substantial individual and socialconsequences.Thisistruefordecisionsboth largeandsmall, andfordecisions
both by individuals and by institutions, political and otherwise.When peoplearedecidingwhat cereal to buy at the grocery store,whethertobuckletheirseatbeltsorlocktheircardoors, what routeto take to the movie theatre, or what to say in response to thequestion, howareyou,theymaywant, almostmorethananythingelse, asimplewayofproceed. Reductionoftheburdensof decisionand choice is valued in less routine settings as well. Consider thedecisionwhether to purchase ahouse, to getmarried, tomove to
another
city,
to
have
a
child,
or,
in
the political
realm,
to
create
anewright, toreducespendingonwelfareprograms, ortomakewaror peace;here people often find themselves in a poor position tocalculate ultimate consequences, and they seek to produce simplerstrategiesforchoice.The point very much bears on ethics, politics, law, and
institutional design. To be sure, people often believe that it isimportant to face the responsibility for decision, even thoughanotherstrategywouldproduceabetteroutcome. Sometimeselectedofficialssimply because of the democratic legitimacy that comesfromtheirelectionrefusetorelinquishresponsibilitytothosewithsuperior knowledge; democratic considerationsmay force them to
makedecisionson theirown. Butwhen officialsdecidewhether tosignacivilrightslaw, ortosupportaffirmativeaction, theymaynotbeinapositiontocalculatenetbenefits, andhence theymaychoosesomeotherdecision-makingstrategyforpoliticalorstrictlycognitivereasons. Public institutions generally operate on the basis of thisunderstanding; some institutions, like the EnvironmentalProtectionAgency and the Food and Drug Administration, owetheirexistencepartlytothe legislaturesdesiretoreducetheburdensofjudgment. Orconsideradjudication. When decidingcases, judgesareconstantlyinapositiontodecidehowmuch(andtosomeextentwhether) todecide, anddifferentjudges, with different assessmentof how to weigh the cognitive burdens, often split on just thisquestion.Ourparticularinteresthereisin second-orderdecisions.The term
requiressomeclarification. In thecase of second-orderdesires, onedealswithdesires-about-desires;inthecase of second-order beliefs,
7/28/2019 Second Order Decisions
4/44
3 Second-OrderDecisions
one deals with beliefs-about-beliefs. In the case of second-orderdecisions, however, onedoesnotexactlydealwith decisions-about-decisions. Rather, one deals with the decision about appropriatestrategies for avoidingdecisions or for reducing their costs. Moreparticularly, ourconcern iswithstrategies that peopleuse in order toavoid getting into anordinarydecision-making situation in the firstinstance.Thus people (and institutions)might be said to make asecond-orderdecisionwhen they choose one from among severalpossiblesecond-orderstrategiesforminimizing theburdens of, andrisk of error in, first-orderdecisions. Second-order decisions aboutsecond-orderstrategiesarethusourbasictopic.
Theprocedure
of
choice
might,
for
example,
be
a
delegation
tosomeotherpersonorinstitution(IllaskwhatmywisefriendJohnthinks; hell know how to handle that question or theEnvironmental Protection Agency will decide how to solve theproblemofgroundwaterpollution). Or the chosen proceduremayinvolveajudgment, beforeultimatedecision-makingsituationsarise,in favor of proceeding via some rule settled in advance (I alwaysbuckle up my kids in the back seat, even on short rides, orwheneverthebilltobepaidislessthan$50, I shallpaycash,oriftheopposingpartyproposesataxincrease, donotsupportit). Asweunderstand it here, the term second-order decisions refers tostrategieschosenbeforesituationsof first-orderdecisioninorder to
eliminatethe need for ordinary choice or to reduce the calculativedemandsofchoice.
Second-orderdecisionsareapervasivepartofordinarylifeandamajoraspectofethics, politics, andbothprivateandpubliclaw. Buttheyhavenotbeenstudiedsystematically. In thisessay, we attempttomakesomeprogresson the topic. Oneofourstrategies is to seewhat might be learned by exploring analogies and disanalogiesbetween the cases of individuals and institutions. Each of us livesaccordingtoawiderangeofsecond-orderdecisions, manyof themsoreflexiveandsothoroughly internalizedthat theydonot seemtobedecisionsatall. Politicaland legalinstitutions often confront orembodysecond-orderdecisions. Indeed, oneofthemost importanttasks for aConstitution itself is to make a series of second-orderdecisions. EvenafteraConstitutionisinplace, politicalactorsfacearangeof furthersecond-orderdecisions, andwell-organized private
7/28/2019 Second Order Decisions
5/44
CHICAGOWORKINGPAPERINLAWANDECONOMICS 4
groups tend to know this, at leastwhere the stakes are especiallyhigh.
Partofourpurposeissimplytoorganizethistopicbyprovidingataxonomy of strategiesthe first step toward understanding theadoption, at the individual and social levels, of one or anothersecond-order decision. We also try to provide some guidance onpositiveandnormative issues. Both individual people and collectivebodies may face an interesting meta-decision: Which of thepossibilitieson themenu isthe suitableonewith regard to a givenkindofcases?Should an agent or an institution adopt a rule or apresumption? When is it best to take small steps, or instead to
delegate
to
anotherparty?
Second-order strategies differ in the extent to which theyproduce mistakes and in the extent to which they imposeinformationalandotherburdenson theagentandonotherseitherbefore the process of ultimate decision or during the process ofultimatedecision. Thusasecond-orderdecisionmightwellbebasedon ajudgment about how best to (1) reduce the overall costs ofdecision and (2) regulate the number, magnitude, and quality ofmistakes. There are three interesting kinds of cases. First, somesecond-orderstrategiesimposelittleinthewayofdecisionalburdenseither before or during the ultimate decision. This is a greatadvantage, andamajorquestioniswhether the strategyinquestion
(consideradecisiontoflipacoin)producestoomuchunfairness ortoo many mistakes. Second, some second-order strategies greatlyreducedecisional burdens at the time of ultimate choice, but theyrequireconsiderablethinking inadvance(consider, forexample, thecreationofrulestogovernemissionsfromcoal-firedpowerplants, ortogovernmisconductbyoneschildren). Decisionsofthiskindmaybeburdensometomakeinadvance;buttheburdensmay beworthincurring ifthey remain farlessthan theaggregate burdens of on-the-spotdecisions. Here toothere isaquestion of how to regulatethe number, magnitude, and quality of mistakes. Third, somesecond-orderstrategiesimposelittlein thewayofdecisionalburdeninadvance, butmayimposehighburdensonotherswhomustmakethe first-order decision; a delegation of power to some trustedassociate, ortoanauthority, isthemostobviouscase.We attempt to understand these different kinds of cases by
drawingonactualpractices, individualandinstitutional. Theresultis
7/28/2019 Second Order Decisions
6/44
7/28/2019 Second Order Decisions
7/44
CHICAGOWORKINGPAPERINLAWANDECONOMICS 6
will produce some extremely bad outcomes (imagine an unusuallysafedriverrushinghisfriendtothehospital);sotoo formandatoryimprisonmentfor three-time felons;sotoowith abanonanydogsin restaurants(suppose a police officerneeds help from his bomb-sniffingGermanShepherd). Goodfriendsandgooddoctorstend tohaveaflexibleattitudetowardrules.3
Itisbecauseoftheirgeneralitythat rulesareoften criticizedasapathologyofunnecessarilyrigidpeople and (stillworse) ofmodernbureaucraticgovernment4;buttheymight bedefended as away ofminimizing the burdens of decision while producing good resultsoverall. The rigidity of rules can also produce serious interpretive
difficulties,
as
when
a
rule
confronts
an unanticipated
case
andproduces, in that case, a transparently absurd outcome; here thequestion is whether the rule should operate as something like apresumption. Agooddealof interpretivedispute in law is focussedon such problems, which iswhy rules often arenearly full, ratherthanfull, exantespecificationsofoutcomes.5
2.Presumptions
Oftenordinarypeopleandpublicinstitutions relynot on a rulebut instead on a presumption, which can be rebutted. Apresumptionistypicallyrebuttableonlyon thebasisofashowing ofacertainkindandweight.6Peoplemight say, forexample, that they
will
presume
against
disclosing
a
confidence;
or
they
might
say
thatthey will violate the speed-limit only in compelling, unusualcircumstances (like saving a life); or that the government maydiscriminateon thebasisofraceonly ifthere isanespeciallystrongreasonfordoingso. Inordertoobtaingreateraccuracy, rulesmaybesoftenedinthedirectionofpresumptions. Theresult, itishoped, istomakefewermistakeswhile at the same time limiting decisionalburdens.
Itisimportantheretodistinguishbetween apresumptionandarule-with-exceptions, though the distinction is subtle. A rulewith
3Ondoctorsandrules, seeKathrynHunter, DoctorsStories(1993).
4SeeEugeneBardachandRobertKagan, GoingBytheBook:TheProblem ofRegulatoryUnreasonableness (1982); seealso thepopular treatment in PhilipHoward, TheDeathofCommonSense(1992).5SeeHLAHart, TheConceptofLaw127-30(2ded. 1996).6SeeEdnaUllmann-Margalit, OnPresumptions, 80J. Phil. 143(1983).
7/28/2019 Second Order Decisions
8/44
7 Second-OrderDecisions
exceptionstendstohavethe following structure:DoXexcept incircumstancesA, in which casedonon-X (or, in which case youmay be exempt from doing X). Thus, for example, observe thespeedlimitexceptwhenyou'redrivingapolicecaroranambulanceinanemergency, inwhich casesyoumayexceedit.By contrast, atypical presumption says something like: Act on the assumptionthatPunlessanduntilcircumstancesA(areshown to)obtain, inwhich case, stop (or reconsider or do something else). The twoamount tothesamethingwhen theagent knows whether ornotcircumstancesAobtain. Thetwoarequitedifferentwhen theagentlacks that information. With a presumption, you can proceed
without
the
information;
with
a
rule-with-exceptions,
you
cannotproceed, thatis, youarejustifiedneitherindoingXnorinnotdoingX. Thus presumptions function asdefault rules; they free up theagent, who has a set course of action without knowing whethertherearerebuttingcircumstances.
In law, the distinction between rules-with-exceptions andpresumptions is sometimes conceived as a distinction between exante specification and ex post specification of rebuttingcircumstances. Thusaspeedlimitlawmayhavespecifiedexceptions(policeofficersandambulancedriversmay violate it); a prohibitionon killing does not apply in cases of self-defense. With apresumption, the rebutting circumstances are not identified in
advance;itisunderstoodthat lifemayturnup problems that couldnothavebeenanticipated. Here the ideaofa presumption overlapswiththeideaofastandard,tobetakenuppresently.Many importantpresumptionsresult from the suggestion that
inthecaseofuncertaintyorlackof information, an individual, oragovernment, shoulderrononesideratherthan another. Consider,for example, the presumption of innocence and the notion ofprevention as the strategy of choice in environmental law. Folkwisdom iscapturedin the notion, better safe than sorry, an ideathatoftenhasanethicaldimensionandthathasanaloguesinmanyareas of law and politics. There are also presumptions in favor ofliberty and equality. Daily decisions are permeated and muchsimplified by presumptionsin favor of particular grocery stores,routes to the downtown area, lunch plans. Often there is also animplicitbutwidelysharedunderstandingofthekindsofreasonsthatwill rebut the relevant presumptions. Thus Ronald Dworkins
7/28/2019 Second Order Decisions
9/44
CHICAGOWORKINGPAPERINLAWANDECONOMICS 8
influential claim that rights are trumps can be understood as adescription of rights as strong presumptions, rebutted only by ademonstrationofaparticularkind.7
Presumptionsplayan important rolein the lawofcontractandstatutory interpretation.8 Described as default rules, much ofcontract law isfoundedonan understanding of whatmost partieswoulddomostofthetime;thepartiescanrebutthepresumptionbyspeaking clearly. These market-mimicking default rules producecontinuingdebatesabouttheextenttowhich courtsshouldattempttoask, not whatmost partieswould do, butwhat these particularpartieswouldhavedoneiftheyhadmadeprovisiononthepoint;the
more
specific
inquiry
increases
the
burdens
of
decision
but promises
to increase accuracy. Sometimes contract rule presumptions areinformation-eliciting,thatis, theyattempttoimposeon thepartyinthebetterpositiontoclarifycontractualtermstheobligationtodoprecisely that, on pain of losing the case. In the law of statutoryinterpretation, there is a similar set of presumptions, designed todiscernwhat Congresswould have done or instead to impose thedutytoobtainaclearstatement fromCongresson thepartyin thebestpositiontodoso. It ispossible to see disputes over liberty andequality as rooted in presumptions, more or less crude, aboutappropriatesocialstates, presumptionsthatcanberebuttedbyspecialcircumstances.
3. Standards
Rulesareoftencontrastedwith standards.9Abanon excessivespeedsonthehighwayisastandard;soisarequirementthatpilotsofairplanesbecompetent,orthat studentbehaviorin the classroombereasonable.Thesemightbecomparedwithrulesspecifyinga55mphspeedlimit, orabanonpilotswho areovertheageof70, orarequirement that students sit in assigned seats. In daily life, you
7SeeRonaldDworkin, Taking Rights Seriously (1975); this is an effort toreadDworkin through the lens provided by the discussion of exclusionaryreasonsinJosephRaz, PracticalReasonandNorms37-45(2ded. 1990).
8SeeIanAyresandRobertGertner, FillingGaps inIncompleteContracts, 99YaleLJ87(1989).9See, e.g., Kaplow, Rules and Standards: An Economic Analysis, 42DukeL.J. 189(1992);Sullivan, Foreword:The JusticesofRules andStandards, 105HarvL. Rev. 22(1993).
7/28/2019 Second Order Decisions
10/44
9 Second-OrderDecisions
mightadoptastandardinfavorofdrivingslowlyonasnowyday, orof being especially generous to friends in distress. The degree ofvaguenessisofcoursehighlyvariableamongstandards.The central difference between a rule and a standard is that a
rulesettlesfarmoreinadvanceandallowson-the-spotjudgments tobe quitemechanical. Standards can structure first-order decisions,moreor lessdepending on their content, butwithout eliminatingthe need to continue to deliberate. In law, the contrast betweenrules and standards identifies the fact that with some legalprovisions, interpretershavetodo a greatdeal ofwork in order togivelawrealcontent. Themeaningofastandarddependsonwhat
happens
when
it
is
applied.
Of
course
thenature
of
the provision
cannot be read off its text, and everything will depend oninterpretivepractices. Oncewedefinethe term excessive,wemaywellendupwith arule. Perhapsofficialswilldecidethataspeed isexcessivewheneveritisover60milesperhour.
An important illustrationhere comes from standards of proofandinparticularfromthenotionsofclearandconvincingevidenceand beyond a reasonable doubt. Judges have refused to assignnumberstothese ideas. Thus the legal systemhas standards ratherthan rules. Why shouldthe reasonabledoubtstandardnotbesaidtocallfor, say, 97%certaintyofguilt?Partoftheanswer liesin thefact that this standard must be applied to many different
contextsdifferent crimes, different police behavior, differentdefendants, and so forthand across those contexts, a uniformformula may well be senseless. The reasonable doubt standardallowsadegreeofadaptationto individualcircumstances, andthis ispart of its advantage over any single number. This is also itsdisadvantage, for itimposessubstantialburdenson thosewhomustmaketheultimatedecision.
4.Routines
Sometimes a reasonable way to deal with a large decisionalburdenistoadoptortocontinuearoutine. Bythis term wemeansomething similar to habits, but more voluntary, more self-
conscious, andwithout the pejorative connotations of somehabits(considerthehabitof chewing ones fingernails). Thus a forgetfulpersonmightadoptaroutineoflockinghisdooreverytimehe leaveshisoffice, eventhough sometimeshewillreturn ina fewminutes;
7/28/2019 Second Order Decisions
11/44
CHICAGOWORKINGPAPERINLAWANDECONOMICS 10
thusacommutermight adoptaparticularrouteandfollow it everyday, eventhoughon somedaysanother routewouldbebetter. Theadvantageofaroutineisthatitreducestheburdensofdecisionevenifitproducesoccasionalerror. Theadoptionofroutinesisofcourseacommon phenomenon in daily life, as people act withoutthinking.Thesearethe standard operating procedures bywhichpeoplenegotiatetheirdailyaffairs.We have said that routines are related to habits; they are also
relatedto rules. Often they are the concrete specifications of howprecisely a rule is to be followed. If, say, the rule is that in asnowstorm, whendrivingconditionsarehazardous, schoolsaretobe
called
off,
then
the
routines,
taken
as
standard
operatingprocedures
,
willspecifyexactlyhow the responsibilitiesforcarrying this out areto be allocated: what key features in the weather report shouldtrigger the cancellation, who should notify whom (local radiostations, localTVstations, possiblysomeparticularparents), inwhatordertheschoolbusesaretogoout, andsoforth. Something similarhappenswhen visitingdignitaries come to a nation; the rules ofprotocolsaywhowillreceivespecialtreatment (the redcarpet)andthe routines specifywhat stepswill be taken, who doeswhat andwhen. In this way routines work like manuals; their point is tominimize the discretion allowed to the accidental people whohappentobetherewhentheeventoccursallthe thinking isdone
inadvance.Institutionalpracticesentrench routinesaswell. Anyparliament
isrun in large part by routines, many of them unwritten. To theextentthatalegalsystemreliesonprecedent, itfollowsapracticeofthis general kind. In fact respect for precedent can be seen anespecially important kind of routine. Judges follow precedents notbecausetheybelievethatpastdecisionsare correctthey usuallydonotevenaskwhethertheyarebutbecausedoingsoisaroutine. Ifanaccountistobeoffered, itis(roughly)thatalegalsystemwillbebetter ifjudges follow precedent, because adherence to precedentpromotes planning, decreases the burdens of decisions, andaccomplishesbothofthesegoalswithout, onbalance, creatingmoremistakesthanwouldbecreatedwithoutrelianceonprecedent. Thusfollowingprecedent isakindofenablingconstraintaconstraint
7/28/2019 Second Order Decisions
12/44
11 Second-OrderDecisions
on the power of choice that helps to simplify and to facilitatechoice.10
5. Smallsteps
Apossiblewayofsimplifyingadifficultsituationatthe time ofchoice istoattempttomake a small, incremental decision, and toleave other, larger questions for another day. When a personaldecision involves imponderable and apparently incommensurableelements, peopleoftentakesmall, reversiblestepsfirst.11
For example, Janemay decide to live with Robert before shedecideswhether shewants tomarryhim;Marilynmaygoto nightsch
ooltoseeifsheisreallyinterestedin law; the
governmentmi
ght
experimentwith certain subsidies to independent movie producersbefore committing itself to a full-scale program. A similar smallstepsapproachisthehallmarkofAnglo-Americancommon law.12
(Ifitappearsatthispointthatthecommonlawcanrunafouloftheruleoflaw ideal, theappearancecapturesreality, oratleastsomanypeoplenowurge.13)Judgestypicallymakenarrowdecisions, resolvinglittle beyond the individual case; at least this is their preferredmethodofoperationwhen they arenot quite confident about thelargerissues. Itissometimessuggestedthatbecauseofthe likelihoodof unintended bad consequences, government do best, in certaindomains, if their steps are small and incremental.14The notion of
pilot
programs
is
based
on
this
idea.
In
the
psychological
literature,
the small steps approach has been identified with both steady,reliablesuccess(smallwins15)andrecurrenterror.16
10
SeeStephenHolmes, PassionsandConstraint(1996).11
SeeEdnaUllmann-Margalit, Opting:TheCase of Big Decisions, in The1985YearbookoftheWissenschaftkelegZuBerlin.12
SeeEdwardLevi, AnIntroductiontoLegalReasoning(1948).13
Thetensionbetweentheruleoflawandthecommonlawmethodis thebasicthemeofAntoninScalia, AMatterofInterpretation(1997).14
SeeJamesScott, SeeingLikeAState(1998).15
SeeKarlWeick, SmallWins, 39Am. Psych. 40 (1984). Keickurges, [I]tseemsusefultoconsiderthepossibility that social problems seldom get solved,
becausepeopledefinetheseproblemsinwaysthatoverwhelmtheir ability todoanythingabout them. . . . Callingasituation amereproblem thatnecessities asmallwin . . . improvesdiagnosis, preservesgains, and encourages innovation.Callingasituationaseriousproblemthatnecessitiesa largerwinmaybewhentheproblemstarts.Id. at48.
7/28/2019 Second Order Decisions
13/44
CHICAGOWORKINGPAPERINLAWANDECONOMICS 12
6.Picking
Sometimesthedifficultyofdecisionpushespeopletodecideonarandombasis. Theymight, forexample, flip a coin, ormake someapparentlyirrelevantfactordecisive(itsasunnyday, soIlltakethatjobinFlorida). Ortheymight pickrather than choose (takingthe latterterm tomeanbasingadecision on reasons).17Sometimesthishappenswhenthestakesareverylow. In the supermarket, busyshoppersoftenpick;iftheywere tochoose(among, say, toothpastesorpainrelieversorcereals)theymight find themselvesshoppingforan intolerablylong time. Therearemanypublic analogues. A legalsystemmight, forexample, usealottery, andindeedlotteriesareused
inmany domainswhere the burdens of individualized choice arehigh, andwhen there issome particular problemwith deliberationabout the grounds of choice, usually because of underlyingasymmetriesamongthealternatives.While people sometimes pick because the stakes are low, they
may pick in the extreme opposite case too:When the differencesbetween the alternatives are enormous, too big and confusing tocontemplate, orin some respect incommensurable. Theymay pickbecause they donot know where to begin (so to speak). Or theconsequencesfordecisionmaybesolargethatpeopledonotwanttotakeresponsibilityformakingthedecision;hencetheypick(considerSophiesChoice). Heredelegationmightbeanalternativetopickingasthesecond-orderstrategy.
7.Delegation
Afamiliarwayofhandlingdecisionalburdensistodelegatethedecisiontosomeoneelse. Peoplemight, forexample, relyonaspouseorafriend, orchoosean institutionalarrangementbywhich certaindecisionsaremadebyotherauthoritiesestablishedatthetimeorwellinadvance. Inactualpractice, sucharrangementscanbemoreor lessformal; they involve diverse mechanisms of control, or entirelyrelinquishedcontrol, bythepersonorpeopleforwhosebenefit theyhavebeencreated.
16SeeDanielKahnemanandDonLovallo, TimidChoices andBoldForecasts:ACognitivePerspectiveonRiskTaking, 39Mgmt. Sci. 17(1993).17
See Edna Ullmann-Margalit and Sidney Morgenbesser, Picking andChoosing, 44SocialResearch757(1977).
7/28/2019 Second Order Decisions
14/44
13 Second-OrderDecisions
Sometimes the principal grants full authority to the agent towhom powerhasbeendelegated;trusteesoften have authority ofthissort. Sometimestheprincipalretainsultimatepowerofdecision.Thus, forexample, inasystemofseparatedanddividedpowers, war-makingdecisionsaretypicallydelegatedtospecifiedofficials, subjecttovarioussafeguards. In theprivatesphere, peoplemay rely on thewisdom of those in whom they have great confidence, and herethere isacontinuum frommereconsultation toadelegationoffullauthorityovertheoutcome.
8.Heuristics
People
oftenuseheuristicde
vices,
ormental sh
ort-cuts, asaway
ofbypassingtheneedforindividualizedchoice. Forexample, it canbeoverwhelming tofigureoutforwhom tovotein local elections;peoplemay therefore use the heuristic of party affiliation. Whenmeeting someone new, your behaviormay be a result of heuristicdevicesspecifyinghowtobehavewithapersonfallingin thegeneralcategory in which the new person seems to fall. The relevantcategorymaybeage, gender, education, race, religion, demeanor, orsomethingelse. Whatisimportant isthatdecisionsareaproductofheuristicdevicesthat simplifyacomplexsituationandthat can alsoleadtoerror.
Agreatdealofattentionhasbeengiventoheuristicdevicessaid
to
produce
departures
from
rationality,
understood
as
a
result
ofdecisions based on full information.18 And sometimes heuristicdevicesdo lead to errors, even systematic ones. But often heuristicdevices are fully rational, if understood as away of produce prettygoodoutcomeswhileatthe same time reducing cognitive overloadorotherdecisionalburdens.
III. DecisionsandMistakes
A.CostsofDecisionsandCostsofErrors
Under what circumstances will, or should, an agent orinstitutionchooseoneoranothersecond-orderstrategy?Beginwith
a somewhat crude generalization: Rational people attempt to
18
See, e.g., JohnConlisk, Why BoundedRationality?, 34 J. Econ. Lit. 669(1996).
7/28/2019 Second Order Decisions
15/44
CHICAGOWORKINGPAPERINLAWANDECONOMICS 14
minimizethesumofthecostsofmakingdecisionsandthecostsoferror, wherethecostsofmakingdecisionsarethecostsofcoming toclosureonsomeactionorsetofactions, andwherethecostsoferrorareassessedbyexaminingthenumber, themagnitude, andthekindsoferrors. Weunderstanderrorsassuboptimaloutcomes, whateverthecriteriafordecidingoptimality;thus both rules and delegationscanproduceerrors. If thecostsofproducingoptimaldecisionswerezero, itwouldbebesttomakeindividualcalculationsineachcase, forthis approach would produce correct judgments withoutcompromisingaccuracyoranyotherimportantvalue. Thiswouldbetrueforindividualagentsandalsoforinstitutions.
Two qualifications
arenecessary
.
The
first
is
that people
maywanttorelievethemselvesofresponsibilityforcertaindecisions, evenif those people would make those decisions correctly. This is animportant reason for delegation (and hence for institutionalarrangementsofvariouskinds, including the separation of powers).Asecondqualificationcomesfromthefactthat specialproblemsarecreatedbymulti-partysituations:publicinstitutions seektopromoteplanningby setting down rules and presumptions in advance, andthe need for planning can argue strongly against on-the-spotdecisionseveniftheywouldbebothcorrectandcostlesstoachieve.Wewilltakeupthesequalificationsbelow.The chief motivation for second-order decisions is that most
peopleknow two important facts: their own (first-order)decisionsmaybewrong, andarrivingattherightdecisioncanbeverydifficult,orhave high costs. For any agent these costs are of qualitativelydiverse kinds: time, money, unpopularity, anxiety, boredom,agitation, anticipated ex post regret or remorse, feelings ofresponsibility for harm done to self or others, injury to self-conception, guilt, orshame. Thingsbecomedifferentlycomplicatedformultimemberinstitutions, whereinterest-grouppressuresmaybeimportant, andwhere there is the special problem of reaching adegree of consensus. A legislature, for example, might find itespeciallydifficulttospecifytheappropriateapproach to affirmativeaction, giventheproblemsposedbydisagreement, varying intensityof preference, and aggregation problems; for similar reasons amultimember court may have a hard time agreeing on how tohandleanassertedrighttophysician-assistedsuicide. The resultmay
7/28/2019 Second Order Decisions
16/44
7/28/2019 Second Order Decisions
17/44
CHICAGOWORKINGPAPERINLAWANDECONOMICS 16
thiswillpreventthePresidentfrombeingcriticizedfornecessarybutunpopulardecisions(suchasrefusingtoincreasethesupplyofmoneywhen unemployment seems too high); the fact that the FederalReserveBoardisunelectedisanadvantagehere. Thereareanaloguesinbusiness, inworkplaces, andevenin families, where amother orfathermay be given the responsibility formaking certain choices,partlyinorderto relieve the other of responsibility. Of course thisapproachcancausepredictableproblems.
B.RestrictingOptionsandReducingKnowledge
These various points are closely related to two important
phen
omen
a:
wantin
gn
ottohave
o
ptions
and
wantin
gn
ottohaveknowledge. Through restricting options and reducing knowledge,
peoplecansimplifydecisions, andhence theyoften adoptasecond-orderstrategytoaccomplishthesegoals.
Itissometimessuggestedthatpeoplewouldalwaysprefertohavemorechoices rather than fewer, and on conventional assumptionsabouthow peoplemaximize,the suggestionmakesagreat deal ofsense. There is a familiar exception: options consisting of threatsdisguised as offers. But the exception does not come close toexhausting the field. Even ifwe put threats to one side, we canreadily see that often people would like fewer rather than moreoptions, andtheywouldmuchliketobeinapositiontotakecertain
possibilities
off
the
agenda.
21
Indeed,
they
may
be
willing
to
do
or
topay a lot to reduce the option set. Sometimes this is because theaddition of options increases the burdens of decision withoutincreasing, muchoratall, the likelihood of a gooddecision. Thus1000 television channels, or 500 selections on the menu of yourfavorite restaurant, might well increase decision costs withoutimprovingoutcomes, in such away as to produce a net loss. As asecond-order decision, people familiarly truncate the universe ofoptions:Iwantshoes, Iwanttoshoparoundforanoptimalbuy, butIdecideinadvancetolimitmyhuntingtoallthe shoestoresinoneparticularmall. OrIwant togotograduateschool, butImightbeoverloadedwith toomany choices, so I apply to only five schools
21
A good discussion, highly relevant to second-order decisions, is GeraldDworkin, IsMoreChoice BetterThan Less, inGeraldDworkin, Autonomy(1991).
7/28/2019 Second Order Decisions
18/44
17 Second-OrderDecisions
(knowing that someofthose to which I have not appliedmay bebetterthanthebestof the fivetowhich Iamadmitted). Thereareother examples of enabling constraints; consider legislativeprocedures, or rules of order and relevance, designed to reduce thenumber of issues that can be considered at any one time. Whensomethingisconsideredoutoforder,byinformalorformalrule, itis because this limitation, embodying a second-order decision,simplifiesjudgmentbyreducingoptions.
Sometimes both people and institutions want not to haveoptionsforaquitedifferentreason:theysufferfromweaknessofwillandfeartemptation. Theyknow that ifcigarettesorchocolates are
available,
they
may
succumb,
and
they
therefore
attempt
to
closeoff the universe of possibilities. Legal systems are frequentlyresponsivetothisproblem. Considermandatorycoolingoffperiodsfor certain purchases, or mandatory payments to a social securitysystem. In circumstances of temptation, second-order decisionsusuallytaketheformofrulesembodyingprecommitmentstrategies.
It isreasonable to think thatmore knowledge is usually betterthan less, butboth individuals and institutions often seek to be orremain ignorant.22 Whether or not ignorance is bliss, no onesearchesforall available information. Sometimes this is because ofthe sheerdifficulty of obtaining all relevant facts. But people takepositive stepsare willing to incur substantial coststo prevent
themselvesfromfindingthingsout. Thismaybebecauseknowledgecreatesstrategicproblemsorbiasesdecisionsin thewrongdirection.ThegoddessJusticeisblindfolded;theblindfoldsymbolizesakindofimpartiality. Thelawofevidenceisbasedlargelyonajudgment thatcertaininformationwillprejudicethejuryandshouldnotbeheard,evenifitismaterialtothatdecision.
Similarly, peoplemayhave a second-order reason for denyingthemselvesknowledgethatwillmakethemchoosewrongly, imposeon them unwanted feelings of responsibility (as when anacquaintanceconfidesadeepsecret), orotherwise produceharm toselforothers. Thenotionofplausibledeniability,madefamousintheWatergateera, canbetakenasametaphor fordecisionsnot to
22
See Edna Ullmann-Margalit, Not Wanting to Know, in ReasoningPractically (Edna Ullmann-Margalit ed., forthcoming, Oxford UniversityPress1998).
7/28/2019 Second Order Decisions
19/44
CHICAGOWORKINGPAPERINLAWANDECONOMICS 18
obtain information that may compromise the person who hasbecome informed. Presidents and Supreme Court justices preventthemselvesfromknowingmany relevantfacts. Thusmany second-order decisions consist of a failure to secure more information,especiallybutnotonly if it is costless to do so. People fail to seekoptionsorinformation, ortakeaffirmativestepsnot togeteitherofthese, inordertominimizetheburdensofdecisionandthenumberand magnitude of errors, or to reduce actual or perceivedresponsibility.
C. BurdensExAnteandBurdensExPost
Thus
far
wehave
offered
ataxo
n
omy
of
second-
order
strategiesand suggested some general grounds on which someone might
pursueone or another approach. It will be useful to organize thediscussion by observing that several of them require substantialthought before the fact of choice, but little thought during theprocessofultimatechoice, whereasothersrequirelittlethought bothbeforeand during the process of choice. Thus there is a temporaldifference in the imposition of decision costs, which we describewith the terms High-Low and Low-Low. To fill out thepossibilities, we add Low-High and High-High aswell. Notethatbythe termsdecisioncostswe refertotheoverallcosts, whichmaybebornebydifferentpeopleoragencies:theworkdonebefore
the
fact
of
choice
maynot
be
carried
out
by
the
same
actors
who
willhave to do the thinking during the process of ultimate choice.ConsiderTable1:
Table1:BurdensExAnteandBurdensExPost
littleexante thinking substantialexante thinking
littleexpostthinking
Low-Low:picking; small steps;various heuristics;somestandards(1)
High-Low:rules;presumptions; somestandards;routines(2)
substantial
ex
post
thinkingLow
-High:delegation(3)
High-H
igh:Hamlet;characters inHenry
Jamesnovels;dysfunctionalgovernments(4)
7/28/2019 Second Order Decisions
20/44
19 Second-OrderDecisions
Cell 1 captures strategies that seem to minimize the overallburdens of decision (whether or not they promote good overalldecisions). Thesearecasesinwhichagentsdonotinvestagreatdealofthought eitherbeforeor at the time of decision. Picking is themost obvious case; consider the analogous possibility of flipping acoin. Small steps are somewhatmore demanding, since the agentdoeshavetomakesome decisions, but because the steps are small,there need be comparatively little thought before or during thedecision. Aswehavenoted, cell1isthe typicalprocedureofAnglo-American common law; we shall soon investigate thismethod in
more
detail.
The
most
sharply
contrasting
set
of
cases
is
High-High,
Cell(4). Asthiscellcapturesstrategiesthatmaximizeoveralldecisioncosts, it ought for our purposes to remain empty. Fortunately itseemstoberepresentedonlybyasmallminorityofpeopleinactuallife (considerHamlet or certain characters in Henry James novelsandtheirreal-worldanalogues, andalsoincompetentbureaucracies).Cell (2) captures a common aspiration for national legislatures
and for ordinary agents who prefer their lives to be rule-bound.Someinstitutionsandagentsspendagreatdealof timedecidingonthe appropriate rules; but once the rules are in place, decisionsbecomeextremely simple, rigid, evenmechanical. Everyone knowspeople of this sort; they can seem both noble and frustratingprecisely because they follow rules to the letter. Legalformalismthe commitment to setting out clear rules in advanceand mechanical decision afterwardsis associated with cell (2);indeed, theidealoftheruleoflawitselfseemstoentailanaspirationtocell(2).23
Whenalargenumberofdecisionsmustbemade, cell(2)isoftenthe best approach, as the twentieth-century movement towardbureaucracy and simple ruleshelps to confirm. Individual cases ofunfairnessmay be tolerable if the overall result is to prevent thesystemfrombeingoverwhelmedbydecisionaldemands. Cell (2) isalsolikelytobethebestapproachwhen alargenumberofpeopleisinvolvedanditisknowninadvancethatthepeoplewhowillhavetocarry out on-the-spot decisions constantly change. Considerinstitutionswithmanyemployeesandalargeturnoverofemployees23
SeeAntoninScalia, AMatterofInterpretation(1996).
7/28/2019 Second Order Decisions
21/44
CHICAGOWORKINGPAPERINLAWANDECONOMICS 20
(thearmy, entrylevelsoflargecorporations, andsoforth). Theheadof an organizationmaynot want newly recruited, less-than-well-trainedpeopletomakedecisionsforthefirm:rulesshouldbeinplacesoastoinsurecontinuityanduniform levelofperformance. On theotherhand, thefactthat lifewillconfound the rulesoftenproducesargumentsforinstitutionalreformintheformofgrantingpowertoadministratorsoremployeestoexercisecommon sensein the faceofrules.24
An intermediate case can be found withmost standards. Thecreationofthe standardmayitselfrequire substantial thinking, butevenwhen the standard is in place, agentsmayhave to do some
deliberating
in
order
to
reach
closure.
Decisions
arenot
mechanical.
Of course there are many different kinds of standards, and it ispossibletoimaginestandardsthat requireagreatdealof thought exante and standards that require very little, just as it is possible toimagine standards that greatly simplify or standards that giverelativelylittleguidance.Cell(3)suggeststhat institutions and individuals sometimesdo
little thinking in advance but may or may not minimize theaggregatecostsofdecision. Aswehaveseen, delegationsmayrequirelittlethinkinginadvance, atleastonthesubstanceoftheissuestobedecided, andtheburdensofdecisionwillbefeltbytheobjectofthedelegation. Of course some people think long and hard about
whether andtowhom todelegate, andofcoursesomepeoplewhohave been delegated power will proceed by rules, presumptions,standards, small steps, picking, or even subdelegations. Note thatsmallstepsmightbeseenasanefforttoexportthecostsofdecisiontoonesfutureself.
Itisanimportantsocialfactthatmanypeoplearerelievedoftheburdens of decision through something other than their ownexplicitwishes. Considerprisoners, thementallyhandicapped, youngchildren, or(atsometimesandplaces)women; there is a range ofintriguing cases in which society or law makes a second-orderdecisionon someoneelses behalf, often without any indication ofthatpersonsowndesires. Theusurpationofanothers decisions, orsecond-orderdecisions, is often based on a belief that the relevantotherwillsystematicallyerr. Thisofcourserelatesto the notion of24
SeePhillipHoward, TheDeathofCommonSense(1996).
7/28/2019 Second Order Decisions
22/44
21 Second-OrderDecisions
paternalism, which can be seen as arising whenever there isdelegationwithoutconsent.
In some cases, second-orderdecisions produce something bestdescribedasMedium-Medium, with imaginable extensions towardModerately High-Moderately Low, and Moderately Low-ModeratelyHigh. As examples consider some standards, which, itwillberecalled, structurefirst-orderdecisionsbutrequireadegreeofworkon the spot, with thedegreedependingon thenature of theparticularstandard. Butafterunderstanding thepolarcases, analysisoftheseintermediatecasesisstraightforward, andhencewewillnotundertakethatanalysishere.
Wenow
turn
to
the
contexts
in
which
agents
and
institutionsfollowoneoranotherofthebasicsecond-orderstrategies.
IV. Low-High(withSpecialReference toDelegation)
A.InformalandFormalDelegations
Asafirstapproximation, adelegationis a second-order strategythat reduces the delegators costs both before and at the time ofmaking theultimatedecision, through exportingthosecosts to thedelegate. Informaldelegationsoccurallthetime. Thus, forexample,one spousemaydelegate to another the decision about what thefamilywilleatfordinner, what investments tochoose, orwhat cartobuy;adietingteenagermaydelegate to his older sibling or bestfriend the decision whether and when desert may be eaten; anauthormaydelegate to his coauthor the decision how to handleissues within the latters expertise. These delegations often occurbecausetheburdensofdecisionarehigh for the delegator but lowforthe delegate, whomay have specialized information, whomaylackrelevantbiasesormotivationalproblems, orwhomaynotmind(andwhomayevenenjoy)taking responsibility for the decision inquestion. (These casesmay then be more accurately captured asspecialcases ofLow-Low.) The intrinsic costs of having tomaketh
e d
ecision
ar
e
oft
en
count
erbala
n
ced b
y th
e b
en
efit
s
of h
avin
gbeenaskedtoassumeresponsibilityforit(though thesemaybecostsratherthanbenefitsinsomecases). Thus somedelegatesaregladtoassumetheirrole;this isimportant to, though itisnotdecisive for,
7/28/2019 Second Order Decisions
23/44
CHICAGOWORKINGPAPERINLAWANDECONOMICS 22
theethicalissuewhethertodelegate(considerthequestionofjusticewithin the family). And there is an uneasy line, raising knottyconceptual and empirical questions, between a delegation (with adelegator and a delegate) and division of labor (consider theallocationofhouseholdduties).
Publicinstitutions, mostprominently legislatures, oftendelegateauthoritytosomeotherentity. Therearemanypossiblefactorshere.A legislature may believe that it lacks information about, forexample, environmental problems or changes in thetelecommunications market; the result is an EnvironmentalProtection Agency or a Federal Communications Commission.
Alternatively,
the
legislature
may
have
the
information
but
find
itselfunabletoforgeaconsensusonunderlyingvaluesabout, forexample,the right approach to affirmative action or to age discrimination.The legislaturemaybeawarethat itsvulnerabilityto interest-grouppressureswill leaditinbaddirections, and itmayhope and believethattheobjectofthedelegationwillberelativelyimmune. Interest-group pressures may themselves produce a delegation, as wherepowerfulgroupsareunabletoachieveaclearvictoryin a legislaturebut are able to obtain a grant of authority to an administrativeagencyoverwhichtheywillhavepower. Orthe legislaturemaynotwant to assume responsibility for some hard choice, fearing thatdecisions will produce electoral reprisal. Self-interested
representativesmaywellfinditintheirelectoralself-interesttoenacta vague or vacant standard (the public interest, reasonableaccommodation of the disabled, reasonable regulation ofpesticides), andtodelegatethe taskofspecificationtosomeoneelse,secure in the knowledge that the delegate will be blamed forproblemsinimplementation.
B. WhentoDelegate
Obviouslyadelegationissometimesamistakeanabdicationofresponsibility, an act of unfairness, a recipe for more rather thanfewererrors. But when isdelegation the right option?Delegationdeserves to be considered whenever an appropriate delegate is
available and there is a sense in which it is inappropriate for theagent to bemaking the decision by himself. Before delegating,comparisonwithotherpossibleapproachesmaywellbeinorder. Ascomparedwith making all first-order decisions on an all-things-
7/28/2019 Second Order Decisions
24/44
23 Second-OrderDecisions
considered basis, a delegation promises to lower decision costs,certainlyfor the delegator and on certain assumptions on balance;thisdependson thecapacitiesofthedelegate (can he or shemakedecent decisions quickly?). If the delegate is trustworthy, thedelegationmaywellproducefewermistakes.Compared to a High-Low approach, a delegation will be
desirableifthe legislature, or the delegator, is unable to generate aworkableruleorpresumption(andifanythingitcouldcomeupwithwould be costly to produce) and if a delegate would therefore dobetteron themerits. Thismaybethecaseonamultimemberbodythat isunabletoreachagreement, orwhen an agent or institution
faces
a
cognitive
or
motivationalproblem
,
such
as
weakness
of
will
orsusceptibilitytooutsideinfluences. AdelegationwillalsobefavoredoverHigh-Low ifthedelegatorseekstoavoidresponsibilityforthedecisionforpolitical, social, or other reasons, though the effort toavoidresponsibilitymayalsocreateproblemsoflegitimacy, aswhen alegislator relies on experts to make value judgments aboutenvironmentalprotectionordisabilitydiscrimination.
As comparedwith small steps or picking, adelegationmay ormaynotproducehigher totaldecisioncosts(perhapsthedelegateissloworaprocrastinator). Evenifthedelegationdoesproducehighertotaldecisioncosts, itmayalsoleadtoahigherlevelofconfidence intheeventualdecisions, which, ifthedelegateisgood, willbesound.
It follows, unsurprisingly, that the case for delegationwill turn inlarge part on the availability of reliable delegates. In the UnitedStates, theFederalReserveBoardhasahighdegreeofpublicrespectand hence there is little pressure to eliminate or reduce thedelegation. But adelegatea friend, a spouse, an EnvironmentalProtection Agencymay prove likely to err, and a rule, apresumption, orsmallstepsmayemergeinstead.There is also the independent concern for fairness. In some
circumstances, itisunfair todelegateto, for example, a friend or aspouse the power of decision, especially but not only because thedelegate isnot a specialist. Issues of gender equality arise when ahusbanddelegatestohiswifeall decisions involving the householdand the children, even if both husband and wife agree on thedelegation. Entirelyapartfromthisissue, adelegationbyone spousetoanothermaywellseemunfairif(say)itinvolvesachildsproblemswith alcohol, becauseitis an abdication of responsibility, away of
7/28/2019 Second Order Decisions
25/44
CHICAGOWORKINGPAPERINLAWANDECONOMICS 24
transferringtheburdensofdecisiontosomeoneelsewho shouldnotbeforcedtobearthemalone.
In institutional settings, there is an analogous problem if thedelegate (usually an administrative agency) lacks politicalaccountability even if it has relevant expertise. The result is thecontinuing debate over the legitimacy of delegations toadministrative agencies.25 Such delegations can be troublesome ifthey shift the burden of judgment from a democratically electedbodytoonethatisinsulatedfrompoliticalcontrol. So too, there isapossiblyillegitimateabdicationof authoritywhen ajudgedelegatescertainpowerstolawclerks(asisoccasionallyallegedaboutSupreme
Court
justices)
or
to
special
masters
who
are
expert
in
complexquestionsoffactandlaw(asisallegedinconnectionwithaproposeddelegation in theMicrosoft litigation). Avoidance of responsibilitymaybeaseriousproblemhere.
C.Complications
Three important complications deserve comment. First, anydelegatemayitselfresorttomaking second-orderdecisions, and itisfamiliartofinddelegatesundertakingeachofthe strategiesthatwehavedescribed. Sometimes delegates prefer High-Low and hencegeneraterules;almosteveryoneknowsthatthisisthe typicalstrategyoftheInternalRevenueService, adelegateofCongressthat likesto
proceed
via
rule.
Many
spouses,
delegated
the power
of
decision
bytheir husbands orwives, operate in similar fashion. Presumptionsmaybefavoredoverrulesforthenow-familiar reasonthat theycanreduce ex ante costs and promote greater flow. Alternatively,delegatesmay use standards or proceed by small steps. This is thegeneral approach of the National Labor Relations Board, which(strikingly)avoidsruleswheneveritcan, andmuchpreferstoproceedcase-by-case. Or a delegate may undertake a subdelegation.Confrontedwithadelegationfromherhusband, awifemayconsultasiblingoraparent. AskedbyCongresstomakehard choices, thePresident may and frequently does subdelegate to some kind ofcommission, forsomeofthesamereasonsthat spurredCongressto
delegateinthefirstinstance. Ofcourseadelegatemayjustpick. She
25
Compare David Schoenbrod, Power Without Responsibility (1995) withJerryMashaw, Chaos, Greed, andGovernance (1997).
7/28/2019 Second Order Decisions
26/44
25 Second-OrderDecisions
may, forexample, choosetoflipacoin, or shemay decidewithoutdoingmuchthinkingaboutwhatdecisionisbest.The second complication is that the control of a delegate
presentsapotentiallyseriousprincipal-agentproblem. How can thepersonwho hasmade the delegation ensure that the delegatewillnotmakeseriousandnumerousmistakes, orinsteadfritter away itstime trying to decidehow to decide? There are multiple possiblemechanisms of control. Instead of giving final and irreversiblepowersofchoicetothedelegate, apersonorinstitutionmight turnthedelegateintoamereconsultantoradvice-giver. Awiderangeofintermediaterelationshipsispossible. In thegovernmental setting, a
legislature
can
influence
theultimate
decision
by
voicing
its
concernspublicly if an administrative agency is heading in the wrongdirection, and the legislature usuallyhas the power to overturn anadministrativeagency ifitcanmusterthe will to do so. Ultimatelythedelegatormayretainthepower toeliminatethedelegation, andtoensureagainst(whatthedelegatorwouldconsidertobe)mistakes,itmaybesufficient forthedelegatetoknow this fact. In informalrelations, involving friends, colleagues, and familymembers, thereare various mechanisms for controlling any delegate. Somedelegatesknow that they are only consultants; others know thattheyhavetheeffectivepowerofdecision. Allthishappensthrough arangeofcues, whichmaybesubtle.
The thirdcomplicationstemsfrom the fact that at the outset,thecostsofasecond-orderdecisionof thiskindmaynotbesolowafterall, sincethepersonorinstitutionmusttakethe timetodecidewhether todelegateatalland ifso, towhom todelegate. Complexissuesmayarise about the composition of any institution receivingthedelegation;theseburdensmaybequitehighandperhapsdecisiveagainst delegation altogether. A multimember institution oftendividessharplyonwhethertodelegateandevenafterthatdecisionismade, itmayhavetroubledecidingon the recipientof thedelegatedauthority.
D.IntrapersonalDelegationsandDelegationtoChance
Thus farwehavebeen discussing cases in which the delegatorexports costs to some other party. What about the intrapersonalcase?On the one hand, there is no precise analogy between thatproblemandthecasesunderdiscussion. On theotherhand, people
7/28/2019 Second Order Decisions
27/44
CHICAGOWORKINGPAPERINLAWANDECONOMICS 26
confronted with hard choices can often be understood to havechosen to delegate the power of choice to their future selves.Consider, forexample, suchdecisionsaswhether tobuyahouse, tohaveanotherchild, togetmarriedordivorced, tomovetoanewcity;insuchcasesagentswhoprocrastinatemayunderstandthemselvestohavedelegatedthedecisiontotheirfutureselves.There are two possible reasons for this kind of intrapersonal
delegation, involving timing and content respectively. You maybelieveyouknowwhattherightdecisionis, butalsobelieveitisnotthe right timetobemaking thatdecision, oratleastnot the righttimetoannounceitpublicly. Alternatively, youmaynotknowwhat
the
right
decision
is
and
believe
that
your
future
self
will
be
in
abetterpositiontodecide. Youmay think that your future selfwillhave more information, suffer less or not at all from cognitivedifficulties, bias, ormotivationalproblems, orbeinabetterpositiontoassumethe relevantresponsibility. Perhapsyou are feeling underpressure, sufferingfromillness, ornotsureofyourjudgmentjustyet.In suchcases, thequestionof intrapersonal, intertemporalchoice isnot so far from the problem of delegation to others. It is evenpossible to see some overlapping principal-agent problems withsimilarmechanismsofcontrol, aspeopleimpose certain constraintsontheirfutureselves.
From the standpoint of the agent, then, the strategy of small
steps, like delay, can be seen as a form of delegation. Also, thestrategyofdelegationitselfmayturn into thatofpickingwhen thedelegateisachancedevice. WhenImakemyfuturedecisiondependon which card I draw frommydeck of cards, I've delegated mydecisiontotherandomcard-drawingmechanism, therebyeffectivelyturningmydecisionfromchoosingtopicking.
V. High-Low(withSpecialReference toRulesandPresumptions)
We have seen that people often make second-order decisionsthatarethemselvescostly, simplyinordertoreduce the burdens of
later
decisions
in
particular
cases.
When
this
process
is
working
well,
thereismuchtodobeforethefactofdecision, butonce thedecisionisinplace, thingsaregreatlysimplified.
7/28/2019 Second Order Decisions
28/44
27 Second-OrderDecisions
A.DiverseRules,DiversePresumptions
Wehave suggested that rules and presumptions belong in thiscategory, andfrequentlythisistrue. Butthepointmustbequalified;somerulesandpresumptionsdonotinvolvehighburdensofdecisionbefore the fact. For example, a rule might be picked rather thanchosendriveon the right-hand sideof the road, orspoonsto theright, forkstothe left. Especiallywhen what it is important is toallow all actors to coordinate on a single course of conduct, thereneed be little investment in decisions about the content of therelevantrule. Arulemightevenbeframednarrowly, soastoworkasakindofsmallstep. Acourtmightdecide, forexample, that a law
excludinghomosexualsfrom the armed services is unconstitutional,andthisdecisionmightbeframedasarule;butthecourtsopinioncouldbeissuedinsuchawayastoleaveundecidedmostother issuesinvolving the constitutional status of homosexuals. Rules oftenembodysmallsteps. Ofcoursethe same points can bemade aboutpresumptions, which aresometimespickedrather than chosen andwhichmightbequitenarrow.
For present purposes we focus on situations in which aninstitutionoranagentiswillingtodeliberateagooddealtogeneratea rule or a presumption that, once in place, turns out greatly tosimplify (without impairing and perhaps even improving) futuredecisions. This is a familiar aspiration in law and politics. Alegislaturemight, forexample, decidein favor of a speed limit law,partlyinordertoensurecoordinationamongdrivers, andpartlyasaresultofaprocessofbalancingvariousconsiderationsaboutrisksandbenefits. Peopleareespeciallywillingtoexpendagreatdealofefforttogeneraterulesin two circumstances:(1)when planning and fairnoticeareimportantand(2)when alargenumberofdecisionswillbemade.26
Inmostwell-functioning legal systems, for example, it is clearwhat isandwhat isnotacrime. People need to know when theymay be subject to criminal punishment for what they do. TheAmericanConstitution istaken torequireadegreeofclarityin the
crim
in
al
law,
and
every
would-b
e t
yrant
kn
ows th
at r
ules m
ay b
eirritatingconstraints on his authority. So too, the law of contract26
SeeLouis Kaplow, Rules andStandards:An EconomicAnalysis, 42DukeLJ557(1992).
7/28/2019 Second Order Decisions
29/44
7/28/2019 Second Order Decisions
30/44
29 Second-OrderDecisions
presumption)willbeaccurate, andthat itwill actually be followed.Third, and most obviously, High-Low is better when notrustworthy delegate is available, or when it seems unfair to askanotherpersonorinstitution tomakethe relevantdecision. Hencelegislatures tend in the direction of rule-likejudgment when theyhave little confidence in the executive; in America, parts of theCleanAirAct are a prime example of a self-conscious choice ofHigh-Low over delegation. Liberal democracies take theseconsiderations as special reasons to justify rules in the context ofcriminallaw:The lawdefiningcrimesisreasonably rule-like, partlybecauseofthe importanceofcitizenknowledgeaboutwhat counts
as
a
crime, partly
because
of
a
judgment
that police
officers
andcourtscannotbetrustedtodefinethecontentofthelaw.When would High-Low be favored over Low-Low (picking,
small steps)?The interest in planning ishighly relevanthere andoftenpushesin thedirectionofsubstantialthinking in advance. Iftheagentorinstitutionhasfaithinitsabilitytogenerateagoodruleorpresumption, itdoesnotmakemuch sensetoproceedbyrandomchoiceorincrementally. Hence legislatureshaveoftendisplacedthecommonlawapproachofcase-by-casejudgmentwith clearrulessetout in advance; in England and America, this has been a greatmovementof the twentieth century, largely because of the interestin planning and decreased faith in the courts ability to generate
goodoutcomesthroughsmallsteps.Of course mixed strategies are possible. An institution may
producearuleto cover certain cases butdelegatedecision in othercases; or a delegate may be disciplined by presumptions andstandards;oranareaof law, orpractical reason, may be covered bysomecombinationofrule-boundjudgmentandsmallsteps.
C.PrivateDecisions:OrdinaryPeople, IntrapersonalCollectiveActionProblems,andRecoveringAddicts
Thus far we have been stressing public decisions. In theirindividual capacity, people frequently adopt rules, presumptions, orself-conscious routines in order to guidedecisions that they know
might, in individual cases, be too costly to make or be madeincorrectlybecauseoftheirownmotivationalproblems. Sarahmightdecide, forexample, that shewill turndown allinvitationsfor out-of-town travelin themonth ofSeptember, orJohnmight adopt a
7/28/2019 Second Order Decisions
31/44
CHICAGOWORKINGPAPERINLAWANDECONOMICS 30
presumptionagainstgoing toanyweddings or funerals unless theyinvolveclosefamilymembers, orFredmightmakeuphismindthatatdinnerparties, hewilldrinkwhatever thehost isdrinking. Rules,presumptions, androutinesofthiskindare an omnipresent featureofpracticalreason;sometimestheyarechosenself-consciouslyandasan exercise of will, but often they are, or become, so familiar andsimple that they appear to the agent not to be choices at all.Problemsmayarisewhen apersonfindsthathecannot sticktohisresolution, and thus High-Low may turn into High-High, andthingsmaybeasifthe second-orderdecisionhadnotbeenmadeatall.
Some
especially
important
cases
involve
efforts
to
solve
the
kindsof intertemporal, intrapersonal problems that arise when isolated,small-stepfirst-orderdecisions are individually rational but produceharm to the individualwhen taken in the aggregate. These casesmight be described as involving intrapersonal collective actionproblems.27Consider, forexample, thedecisiontosmokeacigarette(right now), or to have chocolate cake for desert, or to have analcoholicdrinkafterdinner, ortogambleonweekends. Smallsteps,which are rational choices when taken individually and whichproducenetbenefitswhentakenon theirown, can leadtoharmorevendisasterwhen theyaccumulate. There ismuch room here forsecond-orderdecisions. As a self-control strategy, a person might
adopt a rule: cigarettes only after dinner; no gambling, ever;chocolate cake only on holidays; alcohol only at parties wheneveryone else is drinking. A presumption might sometimes workbetter:apresumptionagainstchocolatecake, with thepossibility ofrebuttalon specialoccasions, when celebrationisin theairand thecakelooksparticularlygood.Well-known privateagenciesdesignedtohelppeoplewith self-
controlproblems(AlcoholicsAnonymous, GamblersAnonymous)haveastheirbusinessthedevelopmentofsecond-orderstrategiesofthisgeneralkind. Themoststrikingcasesinvolverecoveringaddicts,butpeoplewho are not addicts, andwho arenot recovering fromanything, often make similar second-order decisions. When self-control is particularly difficult to achieve, an agent may seek to
27
Cf. Thomas Schelling, Self-Command in Practice, in Policy, and in aTheoryofRationalChoice, 74Am. Econ. Rev. 1(1984).
7/28/2019 Second Order Decisions
32/44
31 Second-OrderDecisions
delegateinstead. Whether adelegation(Low-High) ispreferabletoa rule or presumption (High-Low) will depend in turn on thevariousconsiderationsdiscussedabove.
VI. Low-Low(withSpecialReference toPickingandSmall Steps)
A.Equipoise,Responsibility,andCommitment
Whymight an institution or agent pick rather than choose?Whenwouldsmallstepsbebest?At the individual level, it can beobviousthatwhen youare in equipoise, youmight aswell pick; it
simply isnot worthwhile to go through the process of choosingwith itshigh cognitive or emotional costs. As we have seen, theresultcanbepicking inboth low-stakes (cereal choices) and high-stakes(employmentopportunities)settings. Pickingcanevenbesaidtooperateasakindofdelegation, wheretheobjectof thedelegationisfate,andtheagent loses the sense of responsibility thatmightaccompany an all-things-consideredjudgment. Thus some peoplesortouthardquestionsbyresortingtoachancedevice(likeflippingacoin).
Small steps, unlike a random process, are a form of choosing.Highschoolstudentstend todatein thisspirit, atleastmostofthetime; often adults do too. Newspapers and magazines offer trialsubscriptions;the sameistruefor book clubs. Often advertisers (orfor that matter prospective romantic partners) know that peopleprefer small steps and they take advantage of that preference (nocommitments). In the first years of university, studentsneed notcommit themselvestoanyparticular course of study; they can takesmallstepsinvariousdirections, samplingastheychoose.On the institutional side, consider lotteries for both jury and
militaryservice. The appeal of a lottery forjury service stems fromthe relativelylowcostsof operating the system and the belief thatany alternative device for allocationwould producemoremistakes,becauseitwould depend on a socially contentiousjudgment about
wh
o
sh
ouldb
e
servin
g
on
jur
ies,
with
possibly
destruct
iver
esult
s
for
thejurysystemitself. Thekeypointisthatthejuryissupposedtobeacross-sectionofthecommunity, andarandomprocessseemstobethe best way of serving that goal (as well as the fairest way of
7/28/2019 Second Order Decisions
33/44
7/28/2019 Second Order Decisions
34/44
33 Second-OrderDecisions
of toomuch information. A particular casemayhave a surplus ofapparentlyrelevantdetails, andperhapsfuturecaseswill lackone ormore of the relevant features, and this will be the source of theconcernwith creatingwideprecedents. The existenceof(interalia)featuresXorYincaseA, missingincaseB, makesithazardous togenerate a rule in caseA that would govern case B. The narrowwriting and reception of the Supreme Courts decision in thecelebratedAmish case, allowing an exemption of Amish childrenfrommandatorypublicschooling, isanexample.Quite apart from the pressures of inadequate information, too
much information, and disagreement, small steps might make
special
sense
in
view
of
the pervasive
possibility
of
changedcircumstances. Perhaps things will be quite different in the nearfuture;perhapsrelevantfactsandvalueswillchange, andthusarulethat iswellsuitedtopresentconditionsmay become anachronistic.Thusitispossiblethatanydecisioninvolving theapplicationofthefirst amendment to new communications technologies, includingtheinternet, shouldbenarrow, becauseabroaddecision, renderedatthistime, wouldbesolikelytogowrong. On thisview, asmallstepisbestbecauseofthelikelihoodthatabroadrulewouldbemistakenwhenappliedtocasesnotbeforethecourt.
In an argument very much in this spirit, Joseph Raz hasconnected a kind of small stepthe form usually produced by
analogical reasoningto the special problems created by one-shotinterventions intocomplexsystems.29InRazview, courtsreasonbyanalogy in order to prevent unintended side-effects from largedisruptions. Similarly supportive of the small-step strategy, theGermanpsychologistDietrichDorner has done some illuminatingcomputerexperimentsdesignedtoseewhetherpeoplecanengage insuccessful social engineering.30 Participants are asked to solveproblems faced by the inhabitants of some region of the world.Through the magic of the computer, many policy initiatives areavailable to solve the relevant problems (improved care of cattle,childhood immunization, drilling more wells). But most of theparticipantsproduceeventualcalamities, becausetheydonot seethecomplex, system-wide effects of particular interventions. Only the
29
JosephRaz, TheAuthorityofLaw(1985).30
SeeDietrichDorner, TheLogicofFailure(1994).
7/28/2019 Second Order Decisions
35/44
CHICAGOWORKINGPAPERINLAWANDECONOMICS 34
rareparticipantisabletoseeanumberofstepsdown the roadtounderstand themultiple effects of one-shot interventions on thesystem. The successful participants are alert to this risk and takesmall, reversiblesteps, allowingplanning tooccurovertime. HenceDorner, along with others focussing on the problems created byinterventions into systems,31 argues in favor of small steps. Judgesfacesimilarproblems, and incrementaldecisions are a goodway ofrespondingtotheparticularproblemofbounded rationality createdbyignoranceofpossibleadverseeffects.
Fromthesepointswecanseethatsmallstepsmaybebetterthanrulesorthandelegation. Often an institution lacksthe information
to
generate
a
clearpath
for
the
future;
oftenno
appropriate
delegatehasthatinformation. Ifcircumstancesarechanging rapidly, anyruleorpresumptionmightbeconfounded by subsequentdevelopments.What is especially important is that movement in any particulardirectionshouldbereversibleifproblemsarise.The analysis is similar outside of the governmental setting.
Agentsmight take small steps because they lack the informationthat would enable them to generate a rule or presumption, orbecausethedecisiontheyfaceisuniqueandnotlikelytoberepeated,sothatthereisnoreasonforaruleorapresumption. Orsmallstepsmayfollow from the likelihoodofchangeovertime, from the factthatalargedecisionmighthaveunintendedconsequences, or from
thewishtoavoidoratleasttodeferthe responsibilityfor large-scalechange.
VII. SummaryandConclusions
A.Second-OrderStrategies
ThediscussionissummarizedinTable2. Recallthatthetermslowandhigh refertotheoverallcostsof the decision, which arenotnecessarilybornebythe sameagent:with Low-High the costs aresplitbetweendelegatoranddelegate;withHigh-Lowtheymaysplitbetweenan institution (which makes the rules, say) and an agent
(wh
o
followsth
erules)
.
31
SeeJamesScott, SeeingLikeAState(1998).
7/28/2019 Second Order Decisions
36/44
35
Table2:Second-OrderStrategies
Strategies Examples PotentialAdvantages PotentialDisadvantages AppropriateContext
1. low-high:delegation
spouses, friends;administrativeagencies
relieffromdirectresponsibility forultimatedecisions;increasedchanceforgoodoutcomes
problemsrelatingtotrust,fairness, andresponsibility; possiblehighcosts indecidingwhetherandtowhomtodelegate
availabilityofappropriateandtrustworthydelegate;highburdensof, orperceived likelihoodoferrorin, decisionbydelegator
2. low-low:picking, smallsteps, variousheuristics
Anglo-Americancommon law;lotteries;bigpersonaldecisions
lowoverall costs;reversibility; copingwithchangeandwithunintendedconsequences
difficultyofplanning;highaggregatedecisioncosts;multiplemistakes
equipoise/symmetryofpreferencesorvalues;aversionofdrasticchanges;fearofunanticipatedconsequences
3. high-low:rules,presumptions,routines
speedlimitlaws;legal formalism;criminal law;recoveringaddicts;rigid
people
lowcostsofnumerousdecisionsonceinplace;uniformity;facilitatesplanning
difficultyofgeneratinggoodrulesorpresumptions;mistakesonceinplace
sheernumberofanticipateddecisions/decisionmakers;repetitivenatureoffuturedecisions;need forplanning, confidence in
ab
ilit
yt
o
gen
er
at
e
ex
ant
edecisions
4. high-high Hamlet;HenryJamescharacters;dysfunctionalgovernments
none (unlessdecisioncostsareactuallypleasanttoincuranddecisionsendupbeinggood)
paralysis; unpopularity;individualor institutionalcollapse
agencyor institutioncannotdootherwise
7/28/2019 Second Order Decisions
37/44
CHICAGOWORKINGPAPERINLAWANDECONOMICS 36
B.DoPeopleMake Second-OrderDecisions?
We have not yet discussed an important underlying issue: dopeople, orinstitutions, actuallymakeaself-consciousdecisionaboutwhich second-order strategy to favor, given the menu ofpossibilities?Sometimesthisisindeedthecase. Alegislaturemay, forexample, deliberateanddecidetodelegateratherthan toattempt togenerate rules; a court may choose, self-consciously, to proceedincrementally; having rejected the alternatives, a President mayrecommend a lottery system rather than other alternatives foradmittingcertainalienstothecountry. Thus itispossible to think
of
cases
in
which
an
in
stitution
or
a
person
expressly
makes
an
all-
things-considereddecisionin favorofone or another second-orderstrategy.
Sometimes, however, a rapid assessment of the situation takesplace, rather than a full or deliberative weighing of alternativecoursesofaction. This isoften thecase in privatedecisions, wherejudgments often seem immediate. Indeed, second-order decisionsmight be too costly if they were a product of an on-the-spotoptimizing strategy; so taken, they would present many of theproblemsoffirst-orderdecisions.
As in the case of first-orderdecisions, it oftenmakes sense toproceed with what seems best, rather than to maximize in any
systematic
fashion,
simply
because
the
former
way
of proceeding
iseasier (and thusmaymaximize once we considerdecision costs ofvarious kinds). For individuals, the salient features of the contextusuallysuggestaparticularkindofsecond-orderstrategy. Often thesameistrueforinstitutionsaswell.Theseareintendedasdescriptive points about the operation of
practical reason. But at the political level, and occasionally at theindividualleveltoo, itwouldbebettertobemore explicit and self-consciousaboutthediversepossibilities, soastoensurethat societiesand institutions donot find themselvesmaking bad second-orderdecisions. Itispossible, forexample, tofindpathologicallyrigidrules;theSentencingGuidelinesareoftencriticized on this ground, and
whetherornotthecriticismisjust, pathologicalrigidityisaproblemfor societies as well as individuals. Legal formalists, like JusticeAntoninScalia, repeatedlyargueforaHigh-Low strategy, buttheydosowithout engaging thepragmatic issues at stake, andwithout
7/28/2019 Second Order Decisions
38/44
37 Second-OrderDecisions
showing that thisstrategyispreferabletothe realistic alternatives.32
Oftenacourtorevenastatewoulddobesttoproceedviasmallsteps,insuchawayastoensurereversibility;thisisan importantmeansofavoidingtheproblemsassociatedwithsocialplanning, evenforthosewho donot believe that a general antipathy to state planning iswarranted. But it ispossible to find circumstances in which smallstepsleadtodisaster, bypreventing those whomustdealwith thelaw frompredictingitscontent. Peopledonotgenerallymakeself-conscioussecond-orderdecisions, andoftenthisisfortunate;butthediscussionherehasbeenintendedasaninitialsteptowardmaking itpossible to bemore systematic and conscious about the relevant
options.
C.Conclusion
In thecourseofmakingdecisions, peopleareoften reluctant tocalculatethecostsandbenefitsofthealternatives. Insteadtheyresortto second-order strategiesdesigned to reduce the burdens of first-order decisions while producing a tolerably low number ofsuboptimal outcomes. This is a pervasive aspect of the exercise ofpracticalreason, andsecond-orderdecisionshavelargeconsequencesforindividuals, forinstitutions, andforsocieties.
Some such strategies involve high initial costs but generate arelatively simple, low-burden mechanism for deciding subsequent
cases.
These
strategies,
often
taking
the
form
of
rules
orpresumptions, seem best when the anticipated decisions arenumerousandrepetitiveand when advance planning is important.Other strategiesinvolveboth low initial costs and low costs at thetimeofmakingtheultimatedecision. Theseapproachesworkwhena degree of randomization is appealing on normative grounds(perhaps because choices are otherwise in equipoise, or becausenoone should or will take responsibility for deliberate decision), orwhen thedecisionistoodifficulttomake(becauseof thecognitiveoremotionalburdensinvolvedin thechoice) or includes toomanyimponderablesandariskoflargeunintendedconsequences.
Stillotherstrategiesinvolvelow initial costs but high, exported
costs at the time of decision, as when a delegation is made toanotherpersonorinstitution, or(inametaphor)toonesfutureself.
32
SeeAntoninScalia, AMatterofInterpretation(1997).
7/28/2019 Second Order Decisions
39/44
CHICAGOWORKINGPAPERINLAWANDECONOMICS 38
Delegations can take many different forms, with more or lesscontrolretainedbythepersonorinstitutionmaking thedelegation.Strategiesofdelegationmakesensewhenadelegateisavailablewhohas relevant expertise (perhaps because he is a specialist) or isotherwisetrustworthy(perhapsbecausehedoesnot sufferfrombiasor some other motivational problem), or when there are specialpolitical, strategic, or other advantages to placing the responsibilityfor decision on some other person or institution. Delegations cancreateproblemsofunfairness, aswhen delegatesare burdenedwithtasksthattheydonotvoluntarilyassume, orwouldnotassumeunderjust conditions, andwhen the delegation is inconsistent with the
social
role
of
the
delegator,
such
as
a
legislature
or
a
court.
Hencedelegationscanbetroublingfromthepointofviewofdemocracyortheseparationofpowers.Thefinalsetofcasesinvolvehigh costsbothexanteandatthe
time of decision, as in certain hopelessly indecisive fictionalcharacters, and in highly dysfunctional governments. We havemerely gestured in the direction of this strategy, which can beconsidered best only on the assumption that bearing high overallcostsofdecisionisanaffirmativegoodorevensomething torelish.This assumptionmight appear peculiar, but it undoubtedly helpsexplain some otherwise puzzling human behaviorbehavior thatoftenprovidesthemotivationtoconsidertheother, morepromising
second-orderdecisionsdiscussedhere.
7/28/2019 Second Order Decisions
40/44
ThisWorkingPaper isapreliminarydraftdated5/24/98. Allrights
reserved.
Readers
with
comments
should
address
them
to:
CassR. SunsteinKarlN. LlewellynDistinguishedServiceProfessorofJurisprudenceTheLawSchoolTheUniversityofChicago1111E. 60thStreetChicago, IL60637
7/28/2019 Second Order Decisions
41/44
ChicagoWorkingPapersinLawandEconomics(SecondSeries)
1. WilliamM. Landes, CopyrightProtection ofLetters, DiariesandOtherUnpublishedWorks:AnEconomicApproach (July1991).
2. RichardA. Epstein, ThePathtoTheT.J.Hooper:TheTheoryandHistoryofCustomintheLawofTort(August1991).
3. Cass R. Sunstein, On Property and Constitutionalism(September1991).
4. Richard A. Posner, Blackmail, Privacy, and Freedom of
Contract(February1992).5. Randal C. Picker, Security Interests, Misbehavior, and
CommonPools(February1992).6. TomasJ. Philipson&RichardA. Posner, OptimalRegulation
ofAIDS(April1992).7. DouglasG. Baird, Revisiting Auctions in Chapter 11 (April
1992).8. William M. Landes, Sequential versus Unitary Trials: An
EconomicAnalysis(July1992).9. WilliamM. Landes&RichardA. Posner, The Influence of
EconomicsonLaw:AQuantitativeStudy(August1992).10. AlanO. Sykes, TheWelfareEconomicsofImmigrationLaw:
A Theoretical Survey With An Analysis of U.S. Policy(September1992).
11. Douglas G. Baird, 1992 Katz Lecture: ReconstructingContracts(November1992).
12. Gary S. Becker, The Economic Way of Looking at Life(January1993).
13. J. MarkRamseyer, CrediblyCommitting toEfficiencyWages:CottonSpinningCartelsinImperialJapan(March1993).
14. Cass R. Sunstein, Endogenous Preferences, EnvironmentalLaw(April1993).
15. RichardA. Posner, WhatDo Judges and JusticesMaximize?
(The
Same
Thin
g
Everyon
e
Else
Does)
(April
1993).
16. LucianAryeBebchukandRandalC. Picker, BankruptcyRules,ManagerialEntrenchment, andFirm-SpecificHuman Capital(August1993).
7/28/2019 Second Order Decisions
42/44
17. J. MarkRamseyer, ExplicitReasonsforImplicitContracts:TheLegalLogictotheJapaneseMainBankSystem(August1993).
18. WilliamM. LandesandRichardA. Posner, TheEconomicsofAnticipatoryAdjudication(September1993).
19. Kenneth W. Dam, The Economic Underpinnings ofPatentLaw(September1993).
20. Alan O. Sykes, An Introduction to Regression Analysis(October1993).
21. Richard A. Epstein, The Ubiquity of the Benefit Principle(March1994).
22. RandalC. Picker, An Introduction toGameTheory and theLaw(June1994).
23. WilliamM. Landes, Counterclaims: An Economic Analysis(June1994).
24. J. MarkRamseyer, TheMarket forChildren:Evidence fromEarlyModernJapan(August1994).
25. RobertH. GertnerandGeoffreyP. Miller, SettlementEscrows(August1994).
26. Kenneth W. Dam, Some Economic Considerations in theIntellectualPropertyProtectionofSoftware(August1994).
27. CassR. Sunstein, RulesandRulelessness, (October1994).28. DavidFriedman,MoreJusticeforLessMoney:AStepBeyond
Cimino
(Decemb
er
1994).
29. Daniel Shaviro, Budget Deficits and the IntergenerationalDistributionofLifetimeConsumption(January1995).
30. Douglas G. Baird, The Law and Economics of ContractDamages(February1995).
31. Daniel Kessler, Thomas Meites, and Geoffrey P. Miller,Explaining Deviations from the Fifty Percent Rule: AMultimodalApproachtotheSelectionofCases forLitigation(March1995).
32. GeoffreyP. Miller, Das Kapital: Solvency Regulation of theAmericanBusinessEnterprise(April1995).
33. RichardCraswell, FreedomofContract(August1995).
34.
J.
Mark
Ramseyer,
Public
Choice
(November
1995).
35. KennethW. Dam, IntellectualPropertyinanAgeofSoftwareandBiotechnology(November1995).
7/28/2019 Second Order Decisions
43/44
36. Cass R. Sunstein, Social Norms and Social Roles (January1996).
37. J. MarkRamseyerandEricB. Rasmusen, JudicialIndependencein Civil Law Regimes: Econometrics from Japan (January1996).
38. RichardA. Epstein, TransactionCostsandPropertyRights:OrDoGoodFencesMakeGoodNeighbors?(March1996).
39. CassR. Sunstein, TheCost-BenefitState(May1996).40. WilliamM. LandesandRichardA. Posner, TheEconomicsof
LegalDisputes Over the Ownership of Works of Art andOtherCollectibles(July1996).
41. John R. Lott, Jr. andDavidB. Mustard, Crime, Deterrence,andRight-to-CarryConcealedHandguns(August1996).
42. CassR. Sunstein, Health-HealthTradeoffs(September1996).43. DouglasG. Baird, The Hidden Virtues of Chapter 11: An
Overview of the la and Economics of FinanciallyDistressedFirms(March1997).
44. RichardA. Posner, Community, Wealth, andEquality(March1997).
45. WilliamM. Landes, The Art of Law and Economics: AnAutobiographicalEssay(March1997).
46. CassR. Sunstein, BehavioralAnalysisofLaw(April1997).
47.
John
R
.Lott
,
Jr.
and
Kerm
it
Dan
iel,
Term
Limits
and
Electoral Competitiveness: Evidence from Californias StateLegislativeRaces(May1997).
48. Randal C. Picker, Simple Games in a Complex World: AGenerativeApproachtotheAdoptionofNorms(June1997).
49. Richard A. Epstein, Contracts Small and Contracts Large:Contract Law through the Lens of Laissez-Faire (August1997).
50. Cass R. Sunstein, Daniel Kahneman, and David Schkade,AssessingPunitive Damages (with Notes on Cognition andValuationinLaw)(December1997).
51. William M. Landes, Lawrence Lessig, and Michael E.
Solimine,
Judicial
Influence:
A
Citation
Analysis
of
FederalCourtsofAppealsJudges(January1998).
7/28/2019 Second Order Decisions
44/44
52. JohnR. Lott, Jr., ASimpleExplanation forWhyCampaignExpenditures are Increasing: The Government is GettingBigger(February1998).
53. RichardA. Posner, ValuesandConsequences:AnIntroductiontoEconomicAnalysisofLaw(March1998).
54. Denise DiPasquale and Edward L. Glaeser, Incentives andSocialCapital:AreHomeownersBetterCitizens?(April1998).
55. Christine Jolls, Cass R. Sunstein, and Richard Thaler, ABehavioralApproachtoLawandEconomics(May1998).
56. JohnR. Lott, Jr., DoesaHelpingHandPutOthersAtRisk?:Affirmative Action, Police Departments, and Crime (May1998).
57.Second-Order Decisions, Cass R. Sunstein and EdnaUllmann-Margalit(June1998).