the politician and the judge: accountability in government · the politician and the judge:...

The Politician and the Judge Accountability in Government

By ERIC MASKIN AND JEAN TIROLE

We build a simple model to capture the major virtues and drawbacks of makingpublic officials accountable (ie subjecting them to reelection) On the one handaccountability allows the public to screen and discipline their officials on the otherit may induce those officials to pander to public opinion and put too little weight onminority welfare We study when decision-making powers should be allocated to thepublic directly (direct democracy) to accountable officials (called ldquopoliticiansrdquo) or to nonaccountable officials (called ldquojudgesrdquo) (JEL H1 H7 K4)

The premise behind democracy is that publicdecisions should reflect the will of the peopleBut in most democracies comparatively fewdecisions are made directly by the public1

More often the power to decide is delegated torepresentatives And there is a good reason forthis delegation representatives are usually ex-pected to do a better job2 As specialists inpublic decision-making they are more likelythan the average citizen to have the experiencejudgment and information to decide wisely3 In

any case they have greater incentive than thecitizen to try to do so After all in any largesociety a lone citizen will have strong tempta-tion to free-ride since her chance of actuallyaffecting policy is almost negligible4

Another potential advantage of representativegovernment is that it reduces the risk of ldquotyr-anny by the majorityrdquo Noting the dangers ofdirect democracy5 Madison (1787) writes ldquoItis of great importance to guard one part of thesociety against the injustice of the other partDifferent interests necessarily exist in differentclasses of citizens If a majority be united by a

Maskin Institute for Advanced Study Einstein DrivePrinceton NJ 08540 and Princeton University (e-mailmaskiniasedu) Tirole Institut drsquoEconomie IndustrielleManufacture des Tabacs Bureau MF529-Bat F 21 alleesde Brienne 31000 Toulouse France and GREMAQ (UMR5603 CNRS) Toulouse CERAS (URA 2036 CNRS) Parisand MIT (e-mail tirolecictfr) We are grateful to theparticipants of the CEPR-ESF conference on ldquoThe Institu-tions of Restraintrdquo (Toulouse June 24ndash28 2000) the An-nual Meeting of the European Public Choice Society (ParisApril 18ndash22 2001) the Southeastern International Tradeand Economic Theory Conference (Miami November 16ndash18 2001) the Canadian Economic Theory Conference (To-ronto May 24ndash26 2002) the Modeling the ConstitutionConference (Pasadena May 16ndash17 2003) and many sem-inars for their observations on oral presentations and toMathias Dewatripont John Matsusaka Stephen Morris IanShapiro and two anonymous referees for helpful commentson previous drafts

1 Ballot referendums constitute the largest class of deci-sions made through direct democracy but even in theUnited States and Switzerland where they are especiallypopular they touch on only a small fraction of public policyissues

2 See however the sympathetic 14-page survey on di-rect democracy in The Economist (December 21 1996)Many have argued that once the digital divide is eliminatede-voting will enhance the appeal of referendums

3 The view that governments are better informed thancitizens is emphasized in James Madison (1787) and Abbe

Sieyes (1789) (see also the introduction to Bernard Manin1997)

4 Joseph Schumpeter (1942) puts it in characteristicallyacerbic fashion ldquoThe average citizen expends less disci-plined effort on mastering a political problem than he ex-pends on a game of bridgerdquo

5 Summarizing the case against direct democracy DavidButler and Austin Ranney (1994 pp 17ndash18) write

The main arguments against holding referendums inrepresentative democracies include (1) ordinary cit-izens have neither the analytical skills nor the infor-mation to make wise decisions (2) decisions byelected officials involve weighing the intensity ofpreferences and melding the legitimate interests ofmany groups into policies that will give all groupssomething of what they want (3) decisions made byrepresentatives are more likely to protect the rightsof minorities and (4) by allowing elected officials tobe bypassed and by encouraging officials to evadedivisive issues by passing them on to the votersreferendums weaken the prestige and authority ofrepresentatives and representative government

We have already touched on points (1) and (3) and thesewill figure prominently in our analysis

1034

common interest the rights of the minority willbe insecurerdquo

But if representatives decide for the publicwhat induces them to act in the public interestIn this paper we focus on two motivations inparticular

First we suppose that an official wishes toleave a legacy ie she wants to be rememberedfor great things Indeed in our setup it is notenough for the official that great things be doneshe wants to be the one who does them (al-though we allow for the possibility that herconception of what is ldquogreatrdquo may differ fromthat of the average citizen that is she may benoncongruent with society)6

But the desire to use power to achieve certainends is not the only motive we ascribe to theofficial We also assume that she values being inoffice for its own sake perhaps because sheenjoys the perquisites that come with the jobperhaps because she simply has a taste forwielding influence7

The public can harness these two motives bymaking the official accountable that is by re-quiring her to run for reelection every so oftenHolding reelections creates two major potentialbenefits First it may induce an official who isotherwise inclined (because she is noncongru-ent with society) to act in the public interestBecause the electorate may not always be ableto evaluate the officialrsquos actions directly thiscan be called the ldquomoral-hazard-correctingrdquobenefit of accountability Second reelectionsmay allow the electorate to ldquoweed outrdquo thenoncongruent officials altogether This can beviewed as the ldquoadverse-selection-correctingrdquoeffect

But accountability also carries with it twoserious possible drawbacks In order to get re-elected an official may choose an action notbecause it is right for society but because it ispopular That is she may pander to public opin-ion Although some might call such pandering

ldquoresponsivenessrdquo8 it is in clear conflict with therationale for representative democracy dis-cussed above that representatives can makebetter decisions than ordinary citizens9 Further-more if minority rights are a concern the abil-ity to remove officials from office throughelections may give the majority too much powerto shape the government

A constitutionmdasha specification of who getsto decide whatmdashshould strike a balance be-tween these considerations In this paper wecompare constitutions from the standpoint ofpublic welfare We focus mainly on three dif-ferent modes for making decisions (i) directdemocracy in which the public itself decidesthrough a referendum (ii) representative de-mocracy in which an official subject to reelec-tion (a ldquopoliticianrdquo) decides and (iii) judicialpower in which a nonaccountable official (aldquojudgerdquo) decides10 We also show that in our

6 For example the ldquogreat thingsrdquo she does might consistof favors for a particular interest group in which case herlegacy benefits could include what the interest group doesfor her in return

7 The desire to hold office is a commonly assumedmotive in the political economy literature dating back toRobert Barro (1973) and John Ferejohn (1986) It is ourlegacy motive that is nonstandard (although we believequite realistic)

8 Opinions differed in the 1787 US Constitutional Con-vention on the degree to which officials should be respon-sive to public opinion In this paper however we take theview of delegate George Clymer (see Charles Beard 1913p 193) who wrote that ldquoa representative of the people isappointed to think for and not with his constituentsrdquo Indoing so we adopt the usual political-science definition ofldquorepresentative democracyrdquo as a system in which voters donot instruct their representatives The practice of instructingrepresentatives was widespread in the 18th century (egdeputies to the Estates General in France American statesbefore the 1787 Constitution) The modern version of rep-resentative democracy gained acceptance in the 18th cen-tury in England and with the 1787 Constitution in theUnited States and the 1789 revolution in France See Manin(1997 Ch 5) for more details

9 Received theories of democracy in political scienceeg Robert Dahl (1956) often stress the importance ofrepeated elections for making government responsive to thepublic We argue that ldquomoral-hazard-correctingrdquo respon-siveness is beneficial whereas simply carrying out what thepublic wants can be counterproductive

10 We use the term ldquojudicial powerrdquo to refer to non-accountable officials because in most democraciesjudgesmdashat least at the highest levelsmdashare appointed ratherthan elected In the United States for example all federaljudges are appointed and with lifetime tenure Most of whatwe say about ldquojudgesrdquo however applies equally well toappointed government bureaucrats whose tenure is not ap-preciably affected by their actions

Interestingly (and consistent with the theory developedhere) elected judges in the United States appear to ldquopanderrdquomore (and thus behave more like politicians) than theirappointed counterparts Specifically Timothy Besley andAbigail Payne (2003) analyze employment-discriminationcases and find that in states where judges are subject toreelection discrimination charges are filed at a higher rateThey attribute this finding to an incentive rather than a

1035VOL 94 NO 4 MASKIN AND TIROLE ACCOUNTABILITY IN GOVERNMENT

admittedly extremely simple basic model thewelfare-maximizing constitution generally re-duces to one of these three modes (or a combi-nation thereof)

Of course the requirement that officials run forreelection is not only the only form of account-ability in political life After all most appointedofficials are accountable to their supervisors Elec-toral accountability however is more straightfor-ward to analyze it avoids the need to model themotivations and beliefs of the supervisors

I Overview

A Outline

In Section II we set out a two-period modelwith a homogeneous electorate In each periodthere is a decision to be made between twopossible actions One action is ldquopopularrdquo in thesense that the electorate believes it to be optimalwith better than fair odds The electorate willeither decide itself (direct democracy) or dele-gate the decision to an official who knowswhich action is optimal Each official is eithercongruent (ie she has the same preferences asthe electorate) or noncongruent with societyalthough ex ante the electorate does not knowwhich case holds She also places some weighton holding office for its own sake In the case ofdelegation the first-period official will eitherstand for reelection just before period 2 (in thecase of representative democracy) or remain inoffice automatically for the second period (thecase of judicial power) There may be somechance that before period 2 the electoratelearns whether or not the first-period decisionwas optimal

In Section III we analyze this model for thecase in which there is no chance that the elec-torate learns about the optimality of the first-period decision before period 2 For this ldquono

feedbackrdquo case we show in Section III subsec-tion A that the comparison of our three modesof government turns on the strength of officialsrsquooffice-holding motive When this motive isstrong politicians always pander to public opin-ion11 (choose the popular action) and so thebest form of government (among the three) iseither direct democracy or judicial power (de-pending on how much the electorate knows exante about the optimal action) When it is weakthen politicians always act on their legacy mo-tivation and elections offer some possibility toscreen out noncongruent officials Hence inthat case representative democracy dominatesjudicial power (but still could be inferior todirect democracy if the public has a good ideaex ante of the optimal action) These consider-ations suggest that decisions of sufficientlygreat importance (ones for which the legacymotive is likely to dominate) are best taken bypoliticians rather than by judges (although thesemay also be decisions for which direct democ-racy fares better still)

In Section III subsection B we expand theset of possible governmental systems to includeany scheme in which the decision on whether toretain a first-period official depends on the first-period action We argue that when the office-holding motive is strong no mechanism in thisbroader class dominates judicial power or directdemocracy (assuming that the public cannotcommit itself to a random election) When it isweak however a ldquohybridrdquo mechanism in whichan unpopular official is replaced by direct de-mocracy could be optimal

Section III subsection CndashSection III subsec-tion F consider several extensions Section IIIsubsection C makes it costly for an official tofind out which action is best for her or forsociety and argues that a politician is less likelyto incur that expense than a judge We concludethat highly technical decisions are best taken byjudges Section III subsection D examines theissue of term lengths and points out that opti-mally they should balance the transition costs of

selection effect ie elected judges are likely to be moregenerous in awarding damages (And among states withappointed judges those where judges serve life termsmdashsoneed not worry about reappointmentmdashhave even fewerdiscrimination charges filed) Relatedly Besley and StephenCoate (2003) show that electricity prices paid by retailconsumers are lower in those US states that elect regula-tors The literature also provides evidence (reviewed inBesley and Ann Case 2003) that US governors subject toterm limits behave differently from those who are not

11 As we have already noted there is evidence from theempirical literature that elected and appointed officials be-have differently In addition to the articles cited in footnote10 work by Henning Bohn and Robert Inman (1996) andAndrew Hanssen (1999 2000) suggests that elected judges(whom we would call ldquopoliticiansrdquo) are more independentthan appointed judges

1036 THE AMERICAN ECONOMIC REVIEW SEPTEMBER 2004

replacing officials (which make longer termsdesirable) and the electoratersquos risk aversion(which argues for shorter terms) Section IIIsubsection E asks whether politicians or judgesshould have more discretionary power and sug-gests that politicians should have the edge be-cause their decisions convey useful informationto the public unlike those of judges FinallySection III subsection F takes up the possibil-ity of campaign promises and argues that de-pending on how badly officials want to holdoffice this can be either a boon or a bane

In Section IV we consider the case in whichwith positive probability the electorate learnswhether or not the first-period decision wasoptimal before period 2 We show that if theoffice-holding motive is strong there are two(mutually exclusive) alternatives to the full pan-dering equilibrium we obtain in Section IIIeither (i) officials always choose the optimalaction in the first period (if the probability offeedback is sufficiently high) or (ii) there issome chance that a noncongruent official willpander (a good thing) but otherwise officialsact on their legacy motive (if the probability offeedback is moderate)

Finally in Section V we introduce the pos-sibility of conflict between majority and minor-ity preferences and show that representativedemocracy may do a better job than either directdemocracy or judicial power when the proba-bility that minority preferences should prevail isonly moderate but that judicial power is supe-rior if the risk of minority oppression is suffi-ciently large We interpret the US Constitutionas a device for switching control from elected toappointed officials when minority rights arejeopardized

B Related Literature

This paper borrows from several literaturesOur analysis of representative democracy buildson the theory of elections as a disciplining de-vice (Barro 1973 Ferejohn 1986 and chapters4 and 9 of Torsten Persson and Guido Tabellini2000) These are models of political agency inwhich voters are ex post indifferent about whomthey vote for12 and in which the reelection

motive may induce elected officials to behave incongruence with the electoratersquos wishes Ourdepartures are the introduction of (i) panderingwhich is possible because the electorate hasonly imperfect knowledge of the optimal policyand (ii) adverse selection which arises from thepossibility of officialsrsquo noncongruence

The role of citizensrsquo initiatives as a form ofdirect democracy is studied in Besley and Coate(2000 2003) and John Matsusaka (1992) Thelatter paper provides an empirical analysis of871 Californian ballot propositions and showsthat politicians tend to leave distributional is-sues to the public and to tackle ldquogood govern-mentrdquo issues themselves

Specific points made in this paper relate todisparate parts of the political economy litera-ture The notion that representative democracygives rise to dynamic screening of politicians isreminiscent of Juan Carrillo and Thomas Mari-ottirsquos (2001) more general study of dynamicselection of leadership within a party Papers byPersson et al (1997) Mathias Dewatripont andTirole (1999) Jean-Jacques Laffont and DavidMartimort (1999) and Laffont (2000) presentarguments for separation of powers that differfrom our pandering argument

Work on the rationale for unaccountabil-ity (as opposed to unaccountabilityrsquos conse-quences) is sparse Much of the literature start-ing with Kenneth Rogoff (1985) emphasizes thebenefit of having independent central bankerswith preferences different from those of theelectorate as a commitment device vis-a-vismarkets Antione Faure-Grimaud and DenisGromb (2000) show that agency independencestabilizes policies when bureaucracies are cap-tured by the industry Matthew D McCubbinset al (1987) stress the ex ante control ofagencies

In its emphasis on welfare analysis constitu-tional choice and length of tenure our paper isperhaps most closely related to the literature onposturing in corporate finance This literaturehas argued that the threat of takeovers liquida-tion or replacement induces managers to adoptshort-termist attitudes They accordingly select

12 Moral hazard is the only informational imperfection inmodels following the Barro-Ferejohn tradition The absence

of adverse selection implies that voters learn nothing aboutan official from her behavior and thus are indifferent aboutwhether or not to reelect her Thus disciplining her relies ona particular resolution of this indifference


inefficient investments that pay off quickly(eg Jeremy Stein 1989 Dewatripont andMaskin 1995 Ernst-Ludwig von Thadden1995) or induce income and dividend smooth-ing (eg Drew Fudenberg and Tirole 1995)The analysis of intraparty competition in Ber-nard Caillaud and Tirole (2002) also stresses theimpact of entrenchment and competition forleadership in a mechanism-design frameworkChristopher Avery and Margaret Meyer (2000)study the reliability of recommendations by anevaluator who may be biased in favor of theagent to be evaluated and hired An evaluatorwith career concerns may want to be tough topreserve her reputation vis-a-vis the principalWhile career concerns always benefit the prin-cipal in the basic model Avery and Meyeridentify conditions in particular on the corre-lation of the evaluatorrsquos preferences overtime under which reputational incentives areharmful13

A phenomenon somewhat akin to panderingarises in Stephen Morris (2001) who examinesa model in which an informed advisor is sup-posed to provide information to an uninformedprincipal with the same preferences Morrisshows that despite the coincident preferencesthe advisor may refrain from conveying herinformation truthfully if doing so might jeopar-dize her reputation with the principal (eg in-crease the possibility that the principal thinksthat she is a ldquoracistrdquo) As in our model this hasthe adverse welfare effect of possibly destroy-ing useful information

Finally Brandice Canes-Wrone et al (2001)analyze a career-concern model of political-incumbent policy choice in which as in ourmodel the incumbent has policy expertise andthere is a positive probability of feedback beforethe reelection date However Canes-Wrone etal focus on a quite different set of issues Inparticular the incumbent attempts to signal tal-ent (a high-quality officialrsquos information is bet-ter than that of a low-quality official) rather thancongruence and so the redistributive issuesstudied in Section V below do not arise Alsothe Canes-Wrone et al model takes account-

ability as given and so does not broach thetenure issue that figures so prominently in ourpaper

II The Basic Model

There are two periods 1 and 2 and a pair ab of possible actions in each period (We givethese actions the same labels in both periodsonly for notational convenience In fact theyshould be thought of as different between peri-ods and not necessarily as describable ex ante)

In this basic model all voters have the samepreference ranking14 of the two actions but donot know ex ante what this ranking is Theoptimal actionmdashthe action that voters wouldfavor if they knew their rankingmdashis drawn atrandom independently in each period The prob-ability that it is a is p (1frasl2) which is commonknowledge for the electorate Thus a is theldquopopularrdquo action (in the sense that it would bechosen if the electorate did the choosing itself)and b the ldquounpopularrdquo one The parameter p is ameasure of how much the electorate knowsabout the issue15 Two factors influence thevalue of p technicality and familiarity Techni-calitymdashthe degree to which the issue requiresspecialized knowledge say of economics or sci-encemdashtends to reduce the value of p By con-trast familiaritymdashhow much exposure theelectorate has had to this sort of questionmdashwillraise p (we would classify issues that are largelymatters of basic values as ldquofamiliarrdquo)

We suppose that the electorate is interested inmaximizing its expected utility For most of thepaper (but see Section III subsection D) we willassume that the electorate obtains a payoff of 2if the actions chosen in both periods are opti-mal a payoff of 1 if just one action is optimaland a zero payoff if neither is That is theelectorate is ldquorisk neutralrdquo

Unless the constitution specifies direct de-mocracy the period 1 decision is delegated to

13 Gilat Levy (2000) shows that a careerist judge over-turns precedent more than is socially efficient as suchbehavior is a signal of the judgersquos ability That paper alsolooks at the interaction between this incentive and endoge-nous appeals

14 We can accommodate heterogeneous preferences aslong as they are such that maximizing overall welfare is thesame as maximizing the welfare of the median voter Thusin the basic model we are supposing that the welfare of aminority cannot outweigh that of its complement

15 In the case of a heterogenous electorate p denotes themedian voterrsquos uncertainty about a it is not to be interpretedas the fraction of voters who think a is optimal


an official16 With probability (1frasl2) the of-ficial is ldquocongruentrdquo that is her preferenceranking is the same as the voters would have ifthey were fully informed With probability 1 the official is ldquononcongruentrdquo17 her pre-ferred action differs from that of electorate18

The parameter can be interpreted as both adescription of the underlying pool of candi-dates which for now we will take to be exog-enousmdashbut see the discussion at the end ofSection IVmdashand as a measure of the elector-atersquos ability to screen officials

We suppose that unlike the electorate anofficial knows which action is best for her (andwhich is best for society) This assumption re-flects the officialrsquos greater incentive to be wellinformed (or alternatively her greater expertiseand resources)

To measure an officialrsquos eagerness to holdonto power we introduce a ldquonotional discountfactorrdquo For example suppose that the officialobtains utility G from selecting her preferredaction19 and utility R simply from being inoffice (R may correspond to perks prestigeetc) If the officialrsquos rate of time preference isreflected by discount factor then the trade-offbetween second- and first-period payoffs is

(1) G R

G

since the official always selects her preferredaction when in office at date 2 Notice that could be either greater or smaller than 1

More generally the notional discount factor isdefined as the ratio of the officialrsquos payoff from

remaining in office in period 2 to that from choos-ing her preferred action in period 1 It could beaffected by considerations beyond rent from of-fice-holding and legacy For example if a dis-placed official derived some benefit from hersuccessorrsquos choosing her preferred action then would be somewhat smaller than the above for-mula suggest In principle could also be differ-ent for congruent and noncongruent officials(although for simplicity we assume that it is not)

We suppose that before period 2 the elector-ate learns with probability q whether or not thefirst-period action was the optimal one Withprobability 1 q the electorate learns nothingPresumably the electorate will eventually findout about the optimality of this action and so qcan be thought of as a measure of the speed atwhich feedback accrues

Provided that direct democracy is not calledfor in period 2 the electorate can either main-tain the period 1 official in office (it is requiredto do so if the official was granted a two-periodterm) or draw a new official from the pool ofcandidates (ie elect a challenger) Becauseofficials always choose their preferred action inperiod 2 the new official will select the optimalaction with probability 20

III The No-Feedback Case

Let us assume in this section that the elector-ate obtains no feedback about the first-perioddecision before period 2 (ie q 0) We firstcompare the three benchmark institutions (di-rect democracy representative democracy andjudicial power) next ask whether there exists aninstitution superior to the best of those threeand then go on to consider several extensions

A Comparison of Institutions

Direct Democracy (DD)mdashUnder direct de-mocracy the decision is determined solely byvotersrsquo prior beliefs and so the popular action isselected at each date Thus the electoratersquos ex-pected welfare is

16 We will assume that constitutional dictates cannot beevaded For a discussion of this assumption see Barry RWeingast (1997)

17 With a heterogeneous electorate (see footnotes 14 and15) the congruent official can be thought of as representingthe interests of the median voter whereas the ldquononcongru-entrdquo official represents some other group of voters

18 We are supposing that the preferences of the congru-ent official are perfectly aligned with the interests of theelectorate while those of the noncongruent official arediametrically opposed Without any change in the qualita-tive results we could relax this assumption so that thenoncongruent official simply had a higher probability thanthe congruent official of being at variance with the elector-ate in both periods

19 The assumption that this utility accrues only if theofficial herself selects the action captures the legacymotivation

20 The model can be extended to an overlapping-generations framework with two-period-lived officials Thechallengerrsquos incentives are then similar to those of theincumbent one period earlier Since our focus is on theincumbentrsquos behavior our simpler two-period model in-volves little loss of generality (but see footnotes 29 and 31)


WDD 2p

Judicial Power (JP)mdashUnder judicial powerthe official need not worry about reelection andso selects her preferred action in each periodThe electoratersquos expected welfare is therefore

WJP 2

Representative Democracy (RD)mdashThere aretwo cases

(a) Strong office-holding motive ( 1) When 1 the politician values staying in officethrough period 2 above selecting her pre-ferred action in period 1 In equilibrium thepolitician selects the popular action a re-gardless of her true preferences and theoptimal action Voters then reelect her ifand only if she chose a We call this a fullpandering (FP) equilibrium21 Notice that

voters learn nothing about the politicianrsquostype from her first-period decision sinceshe always panders Thus they are in factindifferent between reelecting and not re-electing her Equilibrium welfare is

WRD p

Note that generically

WRD max WDD WJP

This is not surprising since in this caserepresentative democracy is tantamount todirect democracy in period 1 and to un-accountable judicial power in period 2We conclude that when 1 RD isdominated The optimal choice betweenDD and JP turns on whether p or p (see case 1 of Figure 1)

21 FP is not the only equilibrium but we claim that it isthe most reasonable one to focus on The only other pure-strategy equilibrium is that in which the official alwayschooses b (the unpopular action) However this ldquounpopularpanderingrdquo equilibrium is not robust to a small perturbationto the pool of candidate officials Specifically suppose thatwe introduce a small proportion of officials who haveweak office-holding motives (their discount factor is lowerthan 1) a fraction of these are congruent the rest non-congruent In the Appendix we show (see Proposition A1)

that for 0 the FP equilibrium of the text is the uniqueequilibrium in which the official does not randomize

Finally although there are mixed-strategy equilibria inwhich the official randomizes between a and b each ofthese is non-Markovian in the sense that there are twodifferent states (ie two realizations of the uncertaintyabout the officialrsquos congruence and action arsquos optimality) inwhich all actors have exactly the same preferences and yetbehave differently in the two states These equilibria aretherefore eliminated by the requirement that equilibriumstrategies be Markovian (see Proposition A1)

FIGURE 1 OPTIMAL CHOICE BETWEEN MODES


(b) Weak office-holding motive ( 1) When 1 the politician chooses her preferredaction in period 1 ie there is no pander-ing The electorate can now draw inferencesabout the politician from this choice Spe-cifically the posterior probability that thepoliticianrsquos preferences are congruent withthose of the electorate is

p

p 1 1 p

when the official has chosen a in period 1and

1 p

1 p 1 p

when she has selected b Thus in equilib-rium the official will be reelected if andonly if she has chosen the popular actionThe electoratersquos welfare is therefore

WRD 1 p 1 p

(1)p[12p(1)]

Representative democracy now strictlydominates judicial power The two generatethe same first-period behavior but RD al-lows the electorate to replace an officialwhose behavior suggests noncongruenceRepresentative democracy dominates directdemocracy if and only if

(2) 2p 1

1 2

(see case 1 of Figure 1) Note that the

ldquoboundaryrdquo1

21 2 p between

RD and DD in this case is below thecorresponding boundary p betweenJP and DD in case 1 That is DD ismore likely to be optimal in case 1than in case 1 This is because in thelatter case RD could be optimal even ifp provided that the welfare gain fromhaving the opportunity to oust a possiblynoncongruent official is big enough

Observe that in the no-feedback case (q 0)the only benefit of accountability is the possi-bility of screening out noncongruent officialsThat is in this case accountability does notinduce the official to act on behalf of the elec-torate (when 1 the official panders when 1 she acts in her own interest)

We have implicitly been supposing that alldecisions are equally important in the sensethat they deliver the same potential legacypayoff to the official But we can readilyaccommodate variations in this payoff Imag-ine that the legacy payoff is drawn each pe-riod from a probability distribution with meanG Suppose that in period 1 a politician facesan especially important issuemdash eg whetherto go to warmdash generating potential legacypayoff G with G G Then (if the secondperiod issue is not yet known) the notionaldiscount factor for this decision is

G R

G and therefore lower than that in (1) (we arefocusing only on the first-period decision be-cause the magnitude of the legacy payoff doesnot affect second-period behavior) Such a de-cision would thus be less likely to generatepandering and therefore more apt to be bestallocated to representative democracy ratherthan to judicial power (note that the advantageof RD is not that it leads to better first-perioddecisions than JP but that it provides an oppor-tunity to screen officials) We conclude thatceteris peribus decisions generating large leg-acy payoffs should be assigned to RD over JP(for such discriminatory assignment to be pos-sible some aspect of the decision must be de-scribable in advance eg that it entails whetheror not to go to war) Of course issues thatproduce large potential legacy payoffs are likelyto generate large potential social payoffs aswell Notice that if the social payoff from anoptimal first-period decision increases theboundary between RD and DD in Figure 1 ( 1) would move upwards that is RD would beless likely to generate greater expected welfarethan DD This is because as the ratio of first-period to second-period payoffs rises the com-parative benefit from ousting a noncongruentofficial after period 1 declines


B Mechanism Design

Moving beyond the three systems DD RPand JP let us consider a more general systemfor public decision-making in which an offi-cial makes the first-period choice and xa andxb are her probabilities of being reelectedafter selecting the popular and unpopular ac-tions respectively (our model is very simpleand so xa and xb are the only instruments wehave) To obtain an upper bound on socialwelfare we first assume that these probabili-ties xa and xb are contractible (eg that it issomehow possible to write a constitutionalprovision that if an official chooses a she willbe reelected with probability xa) We thenenquire whether the upper bound can bereached through an institution that does notrequire such contractibility

Suppose that the official prefers action i overaction j where i j a b She will select herpreferred action if

1 xj xi

and will pander if

1 xj xi

Note that given her preferences the officialrsquosincentive to pander does not depend onwhether she is congruent or not (this will nolonger be true when we consider the case q 0) That is a congruent official who prefers ahas an objective function that is identical tothat of a noncongruent official who prefers aFollowing footnote 21 let us focus onMarkov equilibrium meaning that these twotypes of officials adopt the same equilibriumbehavior Clearly randomization or pander-ing by the official will not be welfare-maxi-mizing in all the equilibria we discuss belowthe official chooses her preferred action inperiod 1

If 1 the only possible improvementon RD and DD occurs when p Thenfor a range of parameter values the welfare-maximizing system consists of RD in period 1followed by an election in which the votersdecide either to return the incumbent to office(which in equilibrium will occur when the

official has chosen the popular action) or else tohold a referendum on the second-period action(when the official has chosen the unpopularaction) For 1 improving on JP requiresreducing the difference xa xb to deter pander-ing Probabilities xa 1 xb 0 constitute theoptimal choice But contractibility is requiredfor such randomization In its absence DD or JPis optimal

PROPOSITION 1 Expected welfares from di-rect democracy and judicial power are

WDD 2p and WJP 2

When q 0 (ie there is no feedback about theoptimality of the officialrsquos first-period deci-sion) welfare from representative democracydepends on the strength of the officialrsquos office-holding motive (ie whether 1 or 1)When 1

WRD p max WDD WJP

and so representative democracy is dominatedJudicial power is superior to direct democra-cies if and only if p When 1

WRD 1 2p1

and so representative democracy dominatesjudicial power Representative democracyis superior to direct democracy for valuesof and p such that (2) holds

When 1 no other system (except thoserequiring contractible probabilities) is betterthan judicial power (if p) or direct democ-racy (if p) When 1 the hybrid scheme(RD DD)mdashin which there is a first-periodofficial who runs for reelection and if defeatedis replaced by direct democracymdashgenerates ex-pected welfare

WRD DD 2p p2 2p2

If p then the hybrid is better thanrepresentative democracy and if p is nottoo much greater than it is the welfare-maximizing system (in particular it also bet-ter than direct democracy)


C Costly Information Acquisition

We have assumed that the official is perfectlyinformed about the payoff consequences of ac-tions a and b Suppose now that acquiring suchinformation in each period entails a private costc That is unless the official incurs c she has nobetter information than the electorate Assumethat the officialrsquos gain G from selecting herpreferred action satisfies

(3) 1 pG c

This inequality is a necessary condition foran official under any system to acquireinformation

In the case of judicial power this condition isalso sufficient However it is not generally suf-ficient under representative democracy This isbecause information is less valuable to a politi-cian who risks losing her job if she acts on itPut differently pandering does not requirecostly information (the implicit assumption hereis that the cost of ascertaining public opinion isnegligible relative to that of discovering the trueconsequences of a and b)

To confirm this logic note that under RD acongruent official will acquire information onlyif

(4) pG G R c 1 pG c

pG G R c

where the left-hand side is the officialrsquos payoffif she acquires information (and acts on it) Thecorresponding condition for a noncongruent of-ficial is always satisfied It is easy to verify that(4) is a more demanding constraint on G and cthan (3) That is if (4) holds (3) does too If asbefore we take [(G R)G] then (4) canbe rewritten as

c

G 1 p

1 p 1

Note that 1 is no longer a sufficient condi-tion to avert pandering

We conclude that information acquisitionabout the consequences of public decisions isless likely under representative democracy than

under judicial power a conclusion that suggeststhat representative democracy will not accom-modate technical decisions well

By contrast under representative democracywe would expect politicians to expend consid-erable resources to ascertain the prior beliefs ofthe electorate Our model can be generalized toallow the politician to be uncertain about theelectoratersquos view In this case and if 1 thepolitician will want to poll public opinion (pro-vided the cost of doing so is low enough) toknow what the electorate actually favors Anunaccountable official would not incur such ex-penditures which are socially wasteful in ourmodel To sum up accountability provides anincentive for wasteful information acquisitionand a disincentive for acquiring informationabout the optimal decision However becauseour model assumes that learning the beliefs ofthe public is unrelated to determining the opti-mal action this summary perhaps exaggeratesthe case against representative democracy

D Term Lengths

We have been taking term lengths as exoge-nous In particular we have simply assumedthat a judge serves for two periods But wemight have supposed instead that the term isonly one period long

Actually as the model stands there is nodifference (in welfare terms) between judgesserving one period or two both deliver expectedwelfare 2 But one reason for this coincidenceis that we have been assuming voters are essen-tially risk neutral Let us now suppose insteadthat they are risk averse Specifically let V(2)( 2) V(1) and V(0) ( 0) be the utilitiescorresponding to two one and zero optimaldecisions over two periods where V is a strictlyconcave function22 Then two-period terms stillproduce expected welfare 2 but one-periodterms now generate welfare 22 2(1 )V(1) Thus because V(1) 1 (from riskaversion) one-period terms are better it is lessrisky to have two draws from the candidate poolthan just one

Of course by this logic we should replace

22 That V be concave is not the only reasonable possi-bility If for example the payoff from an optimal decisionin the first period were enhanced by an optimal second-period decision then V might be convex


officials as often as possible which clearly doesnot make sense After all there are setup costsassociated with every regime change not theleast of which is the learning-by-doing that anew official must undertake The optimal termlength will therefore strike a balance betweensetup costs and risk aversion23

E Discretionary Power

Judges and other nonaccountable officialstypically have narrower spheres of action andless discretionary power than accountable offi-cials such as legislators Thus even such apowerful institution as the US Supreme Courtcan consider only the cases brought before itand moreover it is constrained (at least inprinciple) to decide them according to the ex-isting law or the Constitution By contrast theUS Congress can set its own agenda and passany law it wishes (although there may be a riskthat the law will be struck down by the Court)One might ask whether such differences can beexplained in our framework

We cannot accommodate different degreesof discretion in our basic two-action modelbut a simple extension allows us to do so Asbefore let the social payoffs from the optimaland nonoptimal actions be 1 and 0 respec-tively Introduce a status-quo alternative toactions a and b This status-quo action yieldsknown social welfare [0 1] The questionis whether an official should be given thediscretion to choose between a and b or berequired to stick with the status quo Theanswer depends on the value of Discretionshould be given to the official if is smallerthan some cutoff

Under judicial power the optimal value of is

JP WJP

2

Under representative democracy and assumingthat the politician does not pander (otherwise

representative democracy is dominated)24 wehave

RD WRD

2 JP

Discretion confers a benefit under represen-tative democracy that is not available underjudicial power it provides information thathelps society weed out noncongruent officialsAccordingly elected officials should have morediscretion than unelected ones

F Campaign Promises

In our model so far the only informationunder RD that voters have about an official atthe time of reelection is the action that she tookin the first period But in real elections theelectorate typically has more to go on than justthe candidatesrsquo records In particular their cam-paign promises may be informative Let usbriefly examine how the analysis changes whencandidates can make promises during the elec-tion after period 1 about what they would do inperiod 2 We suppose that by that time theyknow the optimal second-period action

If a promise implied no commitment then inour model it would play no substantive role atall Regardless of her true intentions a candi-date would utter whatever helped her to getreelected and so such pronouncements wouldbe worthless Let us suppose instead that if acandidate promises to take an action then whenelected she must carry out that pledge (perhapsbecause the loss of face from not doing sowould be too great) Assume furthermore thatchallengers as well as the incumbent can makesuch promises

There are two cases depending on whether ornot candidates are willing to carry out theirnonpreferred actions simply to get elected25 Ifthey are willing to do so then in effect weobtain pandering in both periods (there cannot

23 When officials are accountable an additional consid-eration enters the picture the length of term will affect theirincentive to pander Because this factor is more compli-cated we will not take it up here

24 We assume here that the status quo if chosen isselected for two periods

25 In our basic model an official always gets nonnega-tive utility (either R or G R) from holding office How-ever we can generalize this setup to allow for the possibilitythat an official who chooses her nonpreferred action derivesnegative utility overall in which case she would not prom-ise that action simply to get elected


be an equilibrium in which congruent and non-congruent officials are separated because thelatter officials would have the incentive to imi-tate the former merely to get elected)

In the case in which candidates pledge onlytheir preferred actions the electorate can inferwhich action is optimal with high probability ifthere are sufficiently many candidates the op-timal action will be promised by approximatelya fraction p of candidates the other action byapproximately a fraction 1 p26 Thus withenough candidates equilibrium will entail theelectorate selecting a candidate among thosewho have pledged the ldquomore promisedrdquo action(the action that a higher proportion of candi-dates have pledged to carry out)

Thus the welfare implications of campaignpledges depend critically on which case we arein In the former case pledges lead to the dete-rioration of RD through pandering in each pe-riod But in the latter welfare in both periodsapproaches the theoretical limit of 1 as the num-ber of candidates grows (this gets at the intuitiveidea that the electorate benefits from politicalcompetition)

IV Feedback

A Forward-Looking Pandering

Let us now assume that parameter q is posi-tive so that with positive probability the elec-torate learns before period 2 whether thefirst-period action was optimal Expected wel-fares under direct democracy and judicial powerare unchanged under direct democracy feed-back is irrelevant since optimal decisions areindependent across periods under judicialpower it is also irrelevant since the officialcannot be voted out Thus as before

WDD 2p

and

WJP 2

Next consider representative democracy Let

xa and xb denote the (sequentially rational) equi-librium probabilities that the official is retainedin office when no feedback is obtained In theAppendix we show that if the fraction of can-didate officials with weak office-holding mo-tives (their is smaller than 1) is positive (evenif arbitrarily small) then in the event of feed-back an official is reelected in equilibrium ifand only if she chose the optimal action inperiod 1 (see the proof of Proposition A2)

There are four possible ldquotypesrdquo of official inthe first period In particular let (C a) corre-spond to a congruent official when a is theoptimal first-period action (and hence the offi-cialrsquos preferred action) Let (N b) refer to anoncongruent official when the optimal actionis b (so that the official prefers a) (C b) and (Na) are defined analogously Let 13 denote thedifference between the officialrsquos payoff fromchoosing a and that from choosing b

13C a 1 q 1 qxa xb

13N b 1 q 1 qxa xb

13N a 1 q 1 qxa xb

13C b 1 q 1 qxa xb

We have listed the four 13s in descending order forthe case q 1 When q 1 then 13(N a) 13(N b) the ranking is otherwise unchanged Itwill make the analysis more interesting to sup-pose throughout that 127

There are three possible equilibria28 two ofwhich are mutually exclusive

Full Pandering EquilibriummdashAs in SectionIII full pandering entails that the official alwaysselects the popular action in equilibrium and isreelected if and only if she adheres to equilib-rium The necessary and sufficient condition forsuch an equilibrium to exist is that 13 (C b) 0when xa 1 and xb 0 that is

26 If there is a cost to making campaign pledges thensome candidates who prefer the nonoptimal action mayrefrain from making pledges at all since they are unlikely tobe elected

27 For 1 the possibility of feedback does not alterthe officialrsquos first-period behavior she always chooses herpreferred action But feedback allows the electorate to learnwith certainty whether or not the official is congruent

28 Here we are assuming that as in footnote 21 a smallproportion of officials who have weak office-holding mo-tives are introduced and this proportion is then sent to zero


1 2q 1

Thus the official must have a strong office-holding motive and feedback must be suffi-ciently slow As noted in the previous sectionFP makes representative democracy unattrac-tive since welfare is then p which is lessthan max WDD WJP

Forward-Looking PanderingmdashIn this nextkind of equilibrium the official selects the op-timal action in the first period regardless of herown preference That is types (C a) and (N a)select action a and types (C b) and (N b)choose action b In the absence of feedback theofficial is always reelected A necessary andsufficient condition for such an equilibrium toexist is that 13(N b) 13(N a) or as we notedabove

q 1

We call the officialrsquos behavior in this equilib-rium forward-looking pandering (FLP) to re-flect the fact that she is pandering not to thecurrent electorate but rather to voters of thefuture who may have received feedback abouther first-period performance29 FLP illustratesthe disciplinary role traditionally attributed toelections in representative democracy (indeedit constitutes a pure moral-hazard-correcting ef-fect there is no correction of adverse selection)It generates welfare

1 WJP

In the Appendix (Proposition A2) we show thatif q (1 )2 the unique limit of perfectBayesian equilibria when a proportion of of-ficials with weak office-holding motives is in-troduced and then sent to zero is the FLPequilibrium in which an official is reelected ifand only if either (i) there is feedback and shehas chosen the optimal action or (ii) there is nofeedback and she has chosen action a Further-more Proposition A3 shows that if (1 )2 q 1 then the only two possible limits as 3 0

are an FLP equilibrium in which if there is nofeedback the electorate randomizes over reelec-tion (this limit always exists for this range ofparameter values) and a full pandering equilib-rium (if (1 2q) 1)

Partial PanderingmdashWhen

q 1

then FLP is no longer an equilibrium The onlyequilibrium possibility (other than full pander-ing) is that types (C a) and (N b) select actiona type (C b) selects action b and type (N a)selects her preferred action b with probability yand panders with probability 1 y where

1 p

1 p 1 py

or

y 1

p 1

Thus (N a) sometimes panders but the otherthree types never do

To compute social welfare in this partialpandering (PP) equilibrium note that withoutfeedback the electorate is indifferent betweenreplacing the official and keeping her regard-less of her first-period behavior (because aftereither choice of action the conditional proba-bility that she is congruent is ) So withoutfeedback second-period welfare always equals and overall welfare is therefore

1 p1 y

1 2p 1 WJP

The PP equilibrium generates greater welfarethan JP in the first period either an officialchooses her preferred action (as in JP) or ifnoncongruent [more specifically if of type (Na)] possibly panders (which enhances welfare)PP also generates greater welfare than JP in thesecond period if there is no feedback after thefirst period the two are the same But if there isfeedback PP improves the likelihood of a con-

29 If we instead adopted an overlapping-generationsframework the official would no longer always choose theoptimal action in equilibrium even in the case q 1Instead we would obtain an equilibrium very much likewhat we call partial pandering (see footnote 31)


gruent official since noncongruent officials aremore likely to be thrown out of office Noticethat the PP equilibrium incorporates bothadverse-selection-correcting and moral-hazard-correcting effects In the Appendix we show(Proposition A4) that when 1 and q 1one limit of perfect Bayesian equilibria as theproportion of officials with weak office-hold-ing motives goes to zero is a PP equilibrium30

The only other possible limit [if (1 2q) 1]is the full pandering equilibrium

To summarize we have

PROPOSITION 2 Suppose that officialsrsquo of-fice-holding motives are strong ( 1) Feed-back (q 0) creates scope for sociallybeneficial forward-looking pandering31 (whenq 1) or partial pandering (when q 1) andeliminates full pandering when (1 2q) 1These three sorts of equilibria are the onlylimiting possibilities when a proportion of offi-cials with weak office-holding motives is intro-duced and then sent to zero

Thus the equilibria of this model exhibit thefull range of potential benefits from account-ability In the absence of feedback (q 0Section III) the reelection process may helpcounteract adverse selection but does nothing tosolve the moral hazard problem either 1 inwhich case the politician panders and represen-tative democracy is dominated or 1 inwhich case behavior is the same as withoutaccountability and elections merely offer theprospect of removing noncongruent officialsWith rapid enough feedback (q 1) theforward-looking pandering equilibrium existsand completely solves the moral hazard prob-lem in the first period but does not help against

adverse selection With moderately fast feed-back (q 1 q 0) the partial panderingequilibrium exists and to some extent counter-acts both moral hazard and adverse selection

B Performance Measurement WithinGovernment

We have seen that an increase in the qualityq of performance measurement can boost thecase for representative democracy Indeed thereis a potential advantage from raising q beyondthe direct benefit of inducing a forward-lookingor partial pandering equilibrium We have beentaking the pool of potential officials as givenexogenously But in practice it will depend onthe cost a candidate incurs from running foroffice and her expected payoff from holdingoffice Notice that moving from full panderingto forward-looking or partial pandering raisesthe payoff for congruent officials and reducesthat for noncongruent officials Thus throughthe self-selection it promotes such a move islikely to improve the composition of the pool

We have assumed that q is exogenous butinstitutional design in practice affects perfor-mance measurement Of particular interest hereis the creation of independent evaluation boards(eg the US General Accounting Office) towrite detailed opinions about the performanceof elected officials (eg the budget designers)Such evaluation boards do not share controlwith elected officials Rather they perform anadvisory or monitoring role They can beviewed as raising the quality q of the elector-atersquos performance measurement

Although we often take such institutions forgranted we might enquire into their rationalewhy is the monitor most often32 an unaccount-able body And why are the evaluated officialsgenerally accountable

The answer to the latter question is straight-forward in our model improved performancemeasurement pays off only if the evaluee isaccountable Indeed the effect of performancemeasurement grows with the degree ofaccountability

The former question is more challenging Atentative answer could be that if an official has

30 In Section III we rule out equilibrium in which theofficial randomizes by appealing to Markov equilibrium(see footnote 21 and Proposition A1) Although partialpandering entails randomization by the official it is consis-tent with the Markovian requirement because once q 0the different types of official no longer have the samepreferences

31 In footnote 29 we noted that if we used an overlap-ping-generations model instead of our two-period once-offframework we would no longer obtain a FLP even in thecase q 1 Instead we would get a PP equilibrium inwhich (i) types (C a) (C b) and (N b) choose theirpreferred actions (ii) type (N a) panders with probability yand (iii) the official is reelected only when there is feedbackthat she chose the optimal first-period action

32 To be sure there also exist Congressional investiga-tion committees in the United States


the incentive to pander to the electorate then somight an accountable monitor In that case theintroduction of monitors would do little to ex-pose panderers

V Tyranny of the Majority

We now relax the assumption of a homoge-neous electorate in order to investigate minoritypolitics We will argue that relative to JP rep-resentative democracy gives more weight to themajority and is therefore more likely to under-value the minority interests

The tyranny of the majority is an old theme ofpolitical thought We have already quoted Mad-ison on this point In The Federalist PapersAlexander Hamilton proposed that the judiciaryshould control the encroachments and oppres-sions of the representative body and that insu-lating it from accountability is important for thisrole

If then the courts of justice are to beconsidered as the bulwarks of a limitedConstitution against legislative encroach-ments this consideration will afford astrong argument for the permanent tenureof judicial offices since nothing will con-tribute so much as this to that independentspirit in the judges which must be essen-tial to the faithful performance of so ar-duous a duty

Still despite the risks that RD poses for minor-ity rights we will suggest that it safeguardsthem better than direct democracy (recall Butlerand Ranney 1994 on the pitfalls of democracysee footnote 5)

We introduce a variant of the basic model inwhich there are two groups the majority and theminority To simplify the analysis we assumethat the majority knows that action a is its op-timal choice Similarly the minority knows thataction b is best for it What is not known iswhich action maximizes overall welfare Withprobability x the majority should prevail be-cause action a imposes only a small negativeexternality on the minority let B 0 be theoverall social benefit of action a over action b inthat case (we normalize social welfare underaction b to be 0) With probability 1 x theminority should prevail action a imposes alarge externality on the minority and generates

overall net loss L 0 relative to action b B canbe interpreted as the social loss when the mi-nority blocks a socially desirable movewhereas L is the social loss from a move thatoppresses the minority

Finally we assume that there are three typesof officials those who are congruent with themajority (labeled ldquoMrdquo and having probabilityM) those who are congruent with the minority(labeled ldquomrdquo and having probability m) andthose who balance the two groupsrsquo interests andso have preferences in line with social welfare(labeled ldquoWrdquo and having probability W) Thepreferences of the three types of officials aresummarized in Table 1

Let us look at the welfare implications of ourthree benchmark institutions

Direct democracy Under DD the majorityalways prevails and so per-period social wel-fare is

xB 1 x L

Judicial power A nonaccountable official se-lects her preferred policy and so per-periodwelfare under JP is

M xB 1 x L m 0 W xB

As one would expect direct democracy faresbetter relative to judicial power when the lossfrom the externality is small (L small) orunlikely ( x large) and when officials tend tobe biased toward interest groups (W low)

Representative democracy The analysis ofRD is similar to that of Section III Let us beginwith the pandering case 1 In that case evena W official who knows that the minority shouldprevail prefers to cater to the majority in period1 As in Section III RD delivers the sameoutcome as DD in period 1 and the same as JPin period 2 It is therefore dominated

Next suppose that 1 so that there is nopandering Action a is chosen by an M official

TABLE 1mdashOFFICIALSrsquo PREFERRED ACTIONS

Case

Type of official

M m W

Majority should prevail a b aMinority should prevail a b b


and with probability x by a W official Actionb is selected by an m official and with proba-bility 1 x by a W official Thus the majoritywill not reelect an official who has chosen band so representative democracy weeds out mand (less effectively) W officials

PROPOSITION 3 For all (i) there existthresholds x and x (x x for 1 x x for 1 where 0 x x 1) suchthat if x x JP is optimal if x x xRD is optimal if x x DD is optimal wherex is the probability that the majority shouldprevail (ii) Similarly there exist thresholds

BL and BL BL BLif 1 such

that ifB

L B

L JP is optimal if B

L

B

L B

L RD is optimal if B

L

B

L DD

is optimal where BL is the ratio of the welfareloss from minority blocking to that from major-ity oppression

PROOFLet wDD WDD2 wJP WJP2 and w

xB Then

WRD wJP M wDD m wJP

Wxw 1 xwJP

WDD WJP and WRD are increasing in x WJP

is the highest of the three for x 0 WDD thehighest for x 1 and for the value x suchthat WDD WJP

WRD WDD WJP

since

w wDD wJP

Thus for low values of x or BL JP is opti-mal these are the ranges for which minorityrights are most important and RD and DD leadto the majority-preferred decision too often Formoderate values of x or BL RD is optimal RDis better than DD because it entails some chancethat the decision will be taken by an m or W

official but RD does not overweight m officialsrsquoinfluence as does JP Finally for high values ofx or BL DD is optimal (since in this sectionwe assume away imperfect knowledge aboutthe optimal action on the part of the elector-ate we might as well let the majority decidedirectly if the minority stands to lose rela-tively little)

The Constitution as a Decision-AllocatingDevicemdashProposition 3 suggests a possible in-terpretation of the US Constitution Let ussuppose that x is rarely so high that DD isdesirable Then it becomes desirable to distin-guish between those cases in which x is (mod-erately) high (so that RD is optimal) and thosein which x is low (JP is optimal)

The Constitution provides a means of makingthis distinction operational If a decision bearson some Constitutional guarantee this is a signthat x is low ie that a minorityrsquos rights arein jeopardy And indeed the decision mech-anism in such a case is to assign the decisionto the federal courts the embodiment of ju-dicial power In the absence of a Constitu-tional issue however the presumption is thatx is not especially low and the policy deci-sion remains in the realm of representativedemocracy

VI Summary and Avenues for FurtherResearch

The paperrsquos main findings can be summa-rized as follows

(1) Accountability has two potential benefits Itallows voters to remove officials whose in-terests appear to be noncongruent with theelectorate but also gives noncongruent of-ficials some incentive to act as though theywere congruent (through the effect offorward-looking or partial pandering)

(2) However accountability may encourage of-ficials to pander to the electorate and over-look minority interests

(3) Nonaccountability is most desirable when(a) the electorate is poorly informed aboutthe optimal action (b) acquiring decision-relevant information is costly and (c) feed-back about the quality of decisions is slowTherefore technical decisions in particular


may be best allocated to judges or ap-pointed bureaucrats

(4) The most important decisions should betaken by elected rather than nonaccountableofficials (although direct democracy mayhave the edge over representative democ-racy for such decisions)

(5) The discretion of nonaccountable officialsshould be more limited than that of account-able ones

(6) Nonaccountability is preferable when themajorityrsquos preferences are very likely toinflict large negative externalities on theminority However representative democ-racy is better in this case than direct democ-racy and for moderate probabilities ofnegative externalities may constitute adesirable compromise between the twoextremes

This paper is only a first step in the analysisof how constitutional design affects publicchoices Many other issues of interest come tomind First extending the model to more thantwo periods would lead to a richer set offeasible institutions For example in a four-period model the policy of giving an officialan initial tenure of two periods and then mak-ing her mandate renewable by vote in each ofthe last two periods can be shown to makesome sense Second the analysis could beextended to an international context we couldget at the idea that elected officials may havea harder time establishing credibility interna-tionally (eg in arms talks or in negotiatingwith the IMF) because of their incentive topander domestically That is pandering tomultiple audiences may be difficult Third wecould study alternative nomination processesfor judges and agency commissioners ratherthan maintain our current assumption thatthey are simply selected at random Fourthwe could enrich the model to allow for thepossibility that campaign contributions affectpoliticiansrsquo choices Fifth the model could beextended to allow elected officials to ldquopassthe buckrdquo by calling for a public referendumFinally some of our analysis might be appliedto the media People often read newspapersor watch television networks that confirmtheir prejudices in other words the me-dia pander in much the same way that politi-cians do

We hope that these extensions and others willbe pursued in future research

APPENDIX

We shall suppose that there is a small pro-portion of officials with weak office-holdingmotives (so that they will choose actions in thefirst period according to their true preferences)Call these the ideological officials A proportion of these are congruent the remainder 1 are noncongruent

PROPOSITION A1 When 1 and q 0the unique pure-strategy perfect Bayesian equi-librium of RD in the limit when 3 0 is a purepandering equilibrium in which the official al-ways chooses a in period 1 and is reelected ifand only if she chooses a The same conclusionholds for mixed-strategy equilibria if we imposethe Markov requirement of footnote 21

PROOFThroughout assume that 0 Suppose

contrary to the proposition there is a pure-strategy equilibrium in which all types of non-ideological officials choose b in period 1 Thenthe choice of a will lead the electorate to believethat the chooser is a congruent ideological of-ficial with probability p[p (1 )(1 p)] and so will reelect her By contrast thechoice of b will lead the electorate to believethat the official is noncongruent with probabil-ity exceeding 1 and so will not reelectThus nonideological type (C a) cannot chooseb in period 1 after all

Suppose next that there exists a pure-strategyequilibrium in which some nonideological typeschoose a and others choose b Then for suf-ficiently small the probability of an officialrsquosbeing congruent conditional on her choosing ain the first period will either be strictly greaterthan or strictly less than In the former casean official will be reelected if and only if shechose a and in the latter if and only if she choseb So in either case all nonideological officialswill have the incentive to behave the same way(in order to get reelected) a contradiction Thusthe proposition is established for pure-strategyequilibria

Next allow for mixed-strategy equilibria butimpose the Markov requirement Suppose thatat least one type of nonideological official


chooses a with positive probability in equilib-rium Because 1 note that the (C a) and (Nb) types have a stronger preference for a over bthan do the (C b) and (N a) types Hence inequilibrium the former group chooses a with atleast as high probability as the latter groupFurthermore from the Markov assumption the(C a) and (N b) types must play a with thesame probability Similarly the (C b) and (N a)officials must play a with the same probabilityThus since the proportion of (C a)s to (N b)sis p(1 )(1 p) ((1 )) the elec-torate will attach a probability greater than tothe officialrsquos being congruent if a is chosen andso will reelect her [This is so even if the (C b)and (N a) types choose a with probability 1because in this case the ideological officialswill tip the balance in favor of reelection] Sym-metrically an official will fail to be reelected ifshe chooses b

Because officials are reelected if and only ifthey choose a nonideological officials will optfor a since 1

PROPOSITION A2 When q (1 ) 2 theunique limit of perfect Bayesian equilibria as 3 0 is an FLP equilibrium in which an officialis reelected if and only if she has chosen (i) theoptimal action when there is feedback or (ii)action a if there is no feedback

PROOFWe claim first that in equilibrium with 0

if the electorate obtains feedback about an offi-cialrsquos first-period choice the official will be re-elected if and only if the decision was optimal

To see this suppose that a is the chosendecision and that the electorate has learned thatit is optimal If in equilibrium no type ofnonideological official chooses a with positiveprobability when a is optimal the electoratewill infer from the feedback that a congruentideological official has chosen a and so willreelect Similarly if in equilibrium some type ofnonideological official chooses a with positiveprobability when a is optimal then the proba-bility that (C a) chooses a must be at least asbig as the probability that (N a) does so [since(C a)rsquos preference for a is stronger than that of(N a)] Thus if it incorporates the possibility ofa congruent ideological official the probabilitythat an official is congruent conditional on ahaving been chosen and revealed optimal is

strictly greater than and so the electorate willagain reelect the official

Suppose instead that a has been revealed tobe nonoptimal If in equilibrium no nonideo-logical type chooses a with positive probabilitywhen a is nonoptimal the electorate will inferfrom the feedback that a noncongruent ideolog-ical official has chosen a and so will not reelectSimilarly if in equilibrium some nonideologi-cal official chooses a with positive probabilitywhen a is nonoptimal then the probability that(N b) chooses a must be at least as big as theprobability that (C b) does so [since (N b)rsquospreference for a is stronger than that of (C b)]Thus including the possibility of a noncongru-ent ideological official the probability that anofficial is noncongruent conditional on a havingbeen chosen and revealed nonoptimal is strictlygreater than 1 and so the electorate willnot reelect the official establishing the claim fora The argument for b is entirely symmetric

Now consider a nonideological type (N b)official If she chooses b her payoff from theabove claim is at least q If instead shechooses a her payoff is at most 1 (1 q)But by hypothesis the former exceeds the lat-ter Hence in equilibrium type (N b) cannotchoose b and the same for (C b) [since (C b)rsquospreference for b is even stronger] Similarlytypes (N a) and (C a) choose a in equilibriumestablishing that any equilibrium must be FLP

Suppose that a has been chosen in the firstperiod In equilibrium nonideological types (Ca) and (N a) and ideological types (C a) and(N b) choose a Hence if there has been nofeedback the probability that the official is con-gruent conditional on the choice of a is strictlygreater than and so she will be reelectedSimilarly she will not be reelected if she haschosen b and there is no feedback

PROPOSITION A3 When1

2 q 1

one limit of PBEs as 3 0 is an FLP equilib-rium in which if there is no feedback the elec-torate randomizes over reelection The onlyother possible limit [if (1 2q) 1] is a fullpandering equilibrium

PROOFFix 0 From the proof of Proposition A2

in any equilibrium if the electorate obtains


feedback about an officialrsquos first-period choicethen the official will be reelected if and only ifthe decision was optimal Suppose that thereexists an equilibrium in which nonideologicaltype (N a) chooses b with positive probabilityThen (N b) and (C b) both choose b withprobability 1 [since for q 1 their preferencefor b is even stronger than that of (N a)]Hence without feedback [and in view of theideological types (N a) and (C b) who chooseb] the probability that an official is noncongru-ent conditional on b having been chosen isstrictly greater than 1 and so an officialwho chooses b will not be reelected This meansthat (N a)rsquos payoff from choosing b is 1 Bycontrast if she chooses a she will be reelected(whether or not there is feedback) and so herpayoff will be which by hypothesis isgreater a contradiction We conclude that (N a)must choose a with probability 1 in equilibriumwhich implies that the same is true of (C a)

Suppose that there exists an equilibrium inwhich nonideological type (N b) chooses b withprobability 1 In that case the probability with-out feedback that an official is noncongruentconditional on her having chosen b is strictlygreater than 1 (thanks to the fact that in theabsence of feedback the probability that anideological official is noncongruent conditionalon her having chosen b is strictly greater than1 ) and so (N b)rsquos payoff from b is q Bycontrast if she chooses a she will be reelectedif there is no feedback and so her payoff will be1 (1 q) which by hypothesis is greatera contradiction We conclude that in any equi-librium (N b) must choose a with positiveprobability

Consider an equilibrium in which nonideologi-cal type (N b) randomizes between a and b Then

(A1) q 1 q 1 1 q13

where the left- and right-hand sides of equation(A1) correspond to the payoffs from b and arespectively and and 13 are the probabilities ofreelection in the absence of feedback when re-spectively b and a have been chosen From equa-tion (A1) and hypothesis we obtain

1 13 0

That is when there is no feedback the elec-torate must randomize between reelecting and

not reelecting either when the official haschosen a or when she has chosen b (or inboth cases) This implies that the probabilityof an officialrsquos being congruent conditionalon a having been chosen is Now thenonideological types other than (N b) whochoose a in equilibrium are (C a) and (N a)and the ideological types are (C a) and (Nb) Hence the probability that (N b) choosesa must be only big enough to offset the effectof the ideological officials Thus as 3 0equilibrium converges to one in which (N b)[and hence (C b)] plays b with probability1mdashie an FLP equilibriummdashand the elector-ate randomizes over reelection when there isno signal as the proposition claims

The only remaining equilibrium possibility isthat nonideological type (N b) chooses a withprobability 1 Now if (C b) does so too thenwe are done because this will be a pure pan-dering equilibrium Hence assume that (C b)chooses b with positive probability If this prob-ability is high enough to outweigh the effect ofthe ideological type (N a) who chooses b thenwithout feedback an official who chooses b willbe reelected implying that (N b)rsquos payoff fromb is whereas that from a is only 1 a contra-diction of the fact that (N b) chooses a Weconclude that the probability that (C b) choosesb must be sufficiently small and converge to 0as 3 0 Thus in the limit we obtain a fullpandering equilibrium as claimed

PROPOSITION A4 When 1 and 0 q 1 one limit of PBEs as 3 0 is a PPequilibrium (which is also a Markov equilib-rium) The only other possible limit [if (1 2q) 1] is a full pandering equilibrium

PROOFFix 0 Again it can be shown that in any

equilibrium an official is reelected when thereis feedback if and only if her first-period deci-sion was optimal Suppose that there exists anequilibrium in which nonideological type (N a)chooses b with probability 1 Then the proba-bility that an official is noncongruent condi-tional on her having chosen b is greater than1 and so without feedback an officialchoosing b will not be reelected Thus (N a)rsquospayoff from b is 1 whereas her payoff from a is a contradiction since the latter is bigger Weconclude that nonideological type (N a) must


choose a with positive probability in equilib-rium If she also chooses b with positive prob-ability then since 0 q 1 implies that 13(C a) 13 (N b) 13 (N a) 13 (C b) type(C b) will choose b and types (C a) and (N b)will choose a the equilibrium is PP Notice thatbecause all four types have different prefer-ences this is also a Markov equilibrium

Assume therefore that (N a) chooses a withprobability 1 Now if nonideological type (C b)chooses b with high enough probability to out-weigh the effect of the ideological types (C b)and (N a) who choose b then without feed-back an official who chooses b will be re-elected implying that (N a)rsquos payoff from b is1 (1 q) whereas that from a is q But byhypothesis the former is bigger than the lattera contradiction We conclude that nonideologi-cal type (C b) can choose b with probability atmost on the order of But as 3 0 (C b)rsquosstrategy converges to one in which a is chosenwith probability 1 implying that the limitingequilibrium is full pandering

REFERENCES

Avery Christopher and Meyer Margaret ldquoDe-signing Hiring and Promotion ProceduresWhen Evaluators are Biasedrdquo Mimeo KSGHarvard and Oxford 2000

Barro Robert ldquoThe Control of Politicians AnEconomic Modelrdquo Public Choice Spring1973 14 pp 19ndash42

Beard Charles An economic interpretation ofthe Constitution of the United States (1913)New York MacMillan (1952 edition)

Besley Timothy and Case Ann ldquoPolitical Insti-tutions and Policy Choices Evidence fromthe United Statesrdquo Journal of Economic Lit-erature March 2003 41(1) pp 7ndash73

Besley Timothy and Coate Stephen ldquoNon-Majoritarian Policy Outcomes and the Roleof Citizensrsquo Initiativesrdquo Mimeo LSE 2000

ldquoElected versus Appointed RegulatorsTheory and Evidencerdquo Journal of the Euro-pean Economic Association September2003 1(5) pp 1176ndash206

Besley Timothy and Payne Abigail ldquoJudicialAccountability and Economic Policy Out-comes Evidence from Employment Dis-crimination Chargesrdquo Mimeo LSS and Mc-Master University 2003

Bohn Henning and Inman Robert ldquoBalancedBudget Rules and Public Deficits Evidencefrom the US Statesrdquo National Bureau ofEconomic Research (Cambridge MA)Working Paper No 5533 1996

Butler David and Ranney Austin Referendumsaround the world The growing use of directdemocracy Washington DC American En-terprise Institute 1994

Caillaud Bernard and Tirole Jean ldquoParties asPolitical Intermediariesrdquo Quarterly Journalof Economics November 2002 117(4) pp1453ndash89

Canes-Wrone Brandice Herron Michael andShotts Kenneth ldquoLeadership and PanderingA Theory of Executive PolicymakingrdquoAmerican Journal of Political Science 200145 pp 532ndash50

Carrillo Juan and Mariotti Thomas ldquoElectoralCompetition and Politicians Turnoverrdquo Eu-ropean Economic Review January 200145(1) pp 1ndash25

Dahl Robert A preface to democratic theoryChicago University of Chicago Press 1956

Dewatripont Mathias and Maskin Eric SldquoCredit and Efficiency in Centralized andDecentralized Economiesrdquo Review of Eco-nomic Studies October 1995 62(4) pp 541ndash55

Dewatripont Mathias and Tirole Jean ldquoAdvo-catesrdquo Journal of Political Economy Febru-ary 1999 107(1) pp 1ndash39

Faure-Grimaud Antoine and Gromb DenisldquoPublic Trading and Private IncentivesrdquoWorking paper FMG 2000

Ferejohn John ldquoIncumbent Performance andElectoral Controlrdquo Public Choice 198650(1ndash3) pp 5ndash25

Fudenberg Drew and Tirole Jean ldquoA Theory ofIncome and Dividend Smoothingrdquo Journalof Political Economy February 1995 103(1)pp 75ndash93

Hanssen Andrew ldquoThe Effect of Judicial Insti-tutions on Uncertainty and the Rate of Liti-gation The Election versus Appointment ofState Judgesrdquo Journal of Legal Studies Jan-uary 1999 28(1) pp 205ndash32

ldquoIndependent Courts and Administra-tive Agencies An Empirical Analysis of theStatesrdquo Journal of Law Economics and Or-ganization October 2000 16(2) pp 534ndash71

Laffont Jean-Jacques Incentives and politicaleconomy Oxford Oxford University Press


2000Laffont Jean-Jacques and Martimort David

ldquoSeparation of Regulators against CollusiveBehaviorrdquo RAND Journal of EconomicsSummer 1999 30(2) pp 232ndash62

Levy Gilat ldquoLegal Precedent and Appellate Re-view with Careerist Judgesrdquo Mimeo LondonSchool of Economics 2000

Madison James ldquoFederalist 10rdquo in AlexanderHamilton James Madison and John Jayeds The Federalist Papers [1787] NewYork Penguin 1981

Manin Bernard The principles of representa-tive government Cambridge CambridgeUniversity Press 1997

Matsusaka John ldquoEconomics of Direct Legis-lationrdquo Quarterly Journal of EconomicsMay 1992 107(2) pp 541ndash71

McCubbins Matthew D Noll Roger G andWeingast Barry R ldquoAdministrative Proce-dures as Instruments of Political ControlrdquoJournal of Law Economics and Organiza-tion Fall 1987 3(2) pp 243ndash77

Morris Stephen ldquoPolitical Correctnessrdquo Jour-nal of Political Economy April 2001 109(2)pp 231ndash65

Persson Torsten Roland Gerard and TabelliniGuido ldquoSeparation of Powers and PoliticalAccountabilityrdquo Quarterly Journal of Eco-

nomics November 1997 112(4) pp 1163ndash202

Persson Torsten and Tabellini Guido Politicaleconomics Explaining economic policyCambridge MA MIT Press 2000

Rogoff Kenneth ldquoThe Optimal Degree of Com-mitment to an Intermediate Monetary Tar-getrdquo Quarterly Journal of EconomicsNovember 1985 100(4) pp 1169ndash89

Schumpeter Joseph (1942) Capitalism social-ism and democracy 3rd Ed New YorkHarper Torchbooks 1962

Sieyes Abbe ldquoDire de lrsquoAbbe Sieyes sur laQuestion du Veto Royalrdquo Versailles Bau-doin Imprimeur de lrsquoAssemble e Nationale1789

Stein Jeremy ldquoEfficient Capital Markets In-efficient Firms A Model of Myopic Cor-porate Behaviorrdquo Quarterly Journal ofEconomics November 1989 104(4) pp655ndash 69

von Thadden Ernst-Ludwig ldquoLong-Term Con-tracts Short-Term Investment and Monitor-ingrdquo Review of Economic Studies October1995 62(4) pp 557ndash75

Weingast Barry R ldquoThe Political Foundationsof Democracy and the Rule of Lawrdquo Amer-ican Political Science Review June 199791(2) pp 245ndash63


common interest the rights of the minority willbe insecurerdquo

But if representatives decide for the publicwhat induces them to act in the public interestIn this paper we focus on two motivations inparticular

First we suppose that an official wishes toleave a legacy ie she wants to be rememberedfor great things Indeed in our setup it is notenough for the official that great things be doneshe wants to be the one who does them (al-though we allow for the possibility that herconception of what is ldquogreatrdquo may differ fromthat of the average citizen that is she may benoncongruent with society)6

But the desire to use power to achieve certainends is not the only motive we ascribe to theofficial We also assume that she values being inoffice for its own sake perhaps because sheenjoys the perquisites that come with the jobperhaps because she simply has a taste forwielding influence7

The public can harness these two motives bymaking the official accountable that is by re-quiring her to run for reelection every so oftenHolding reelections creates two major potentialbenefits First it may induce an official who isotherwise inclined (because she is noncongru-ent with society) to act in the public interestBecause the electorate may not always be ableto evaluate the officialrsquos actions directly thiscan be called the ldquomoral-hazard-correctingrdquobenefit of accountability Second reelectionsmay allow the electorate to ldquoweed outrdquo thenoncongruent officials altogether This can beviewed as the ldquoadverse-selection-correctingrdquoeffect

But accountability also carries with it twoserious possible drawbacks In order to get re-elected an official may choose an action notbecause it is right for society but because it ispopular That is she may pander to public opin-ion Although some might call such pandering

ldquoresponsivenessrdquo8 it is in clear conflict with therationale for representative democracy dis-cussed above that representatives can makebetter decisions than ordinary citizens9 Further-more if minority rights are a concern the abil-ity to remove officials from office throughelections may give the majority too much powerto shape the government

A constitutionmdasha specification of who getsto decide whatmdashshould strike a balance be-tween these considerations In this paper wecompare constitutions from the standpoint ofpublic welfare We focus mainly on three dif-ferent modes for making decisions (i) directdemocracy in which the public itself decidesthrough a referendum (ii) representative de-mocracy in which an official subject to reelec-tion (a ldquopoliticianrdquo) decides and (iii) judicialpower in which a nonaccountable official (aldquojudgerdquo) decides10 We also show that in our

6 For example the ldquogreat thingsrdquo she does might consistof favors for a particular interest group in which case herlegacy benefits could include what the interest group doesfor her in return

7 The desire to hold office is a commonly assumedmotive in the political economy literature dating back toRobert Barro (1973) and John Ferejohn (1986) It is ourlegacy motive that is nonstandard (although we believequite realistic)

8 Opinions differed in the 1787 US Constitutional Con-vention on the degree to which officials should be respon-sive to public opinion In this paper however we take theview of delegate George Clymer (see Charles Beard 1913p 193) who wrote that ldquoa representative of the people isappointed to think for and not with his constituentsrdquo Indoing so we adopt the usual political-science definition ofldquorepresentative democracyrdquo as a system in which voters donot instruct their representatives The practice of instructingrepresentatives was widespread in the 18th century (egdeputies to the Estates General in France American statesbefore the 1787 Constitution) The modern version of rep-resentative democracy gained acceptance in the 18th cen-tury in England and with the 1787 Constitution in theUnited States and the 1789 revolution in France See Manin(1997 Ch 5) for more details

9 Received theories of democracy in political scienceeg Robert Dahl (1956) often stress the importance ofrepeated elections for making government responsive to thepublic We argue that ldquomoral-hazard-correctingrdquo respon-siveness is beneficial whereas simply carrying out what thepublic wants can be counterproductive

10 We use the term ldquojudicial powerrdquo to refer to non-accountable officials because in most democraciesjudgesmdashat least at the highest levelsmdashare appointed ratherthan elected In the United States for example all federaljudges are appointed and with lifetime tenure Most of whatwe say about ldquojudgesrdquo however applies equally well toappointed government bureaucrats whose tenure is not ap-preciably affected by their actions

Interestingly (and consistent with the theory developedhere) elected judges in the United States appear to ldquopanderrdquomore (and thus behave more like politicians) than theirappointed counterparts Specifically Timothy Besley andAbigail Payne (2003) analyze employment-discriminationcases and find that in states where judges are subject toreelection discrimination charges are filed at a higher rateThey attribute this finding to an incentive rather than a




I Overview

A Outline


























II The Basic Model













(1) G R

G
















WDD 2p


WJP 2




WRD p


WRD max WDD WJP








p

p 1 1 p


1 p

1 p 1 p


WRD 1 p 1 p

(1)p[12p(1)]


(2) 2p 1

1 2


ldquoboundaryrdquo1

21 2 p between




G R



B Mechanism Design



1 xj xi

and will pander if

1 xj xi





WDD 2p and WJP 2


WRD p max WDD WJP


WRD 1 2p1



WRD DD 2p p2 2p2





(3) 1 pG c




(4) pG G R c 1 pG c

pG G R c


c

G 1 p

1 p 1





D Term Lengths











JP WJP

2



RD WRD

2 JP


F Campaign Promises











IV Feedback



WDD 2p

and

WJP 2




13C a 1 q 1 qxa xb

13N b 1 q 1 qxa xb

13N a 1 q 1 qxa xb

13C b 1 q 1 qxa xb








1 2q 1



q 1


1 WJP




q 1


1 p

1 p 1 py

or

y 1

p 1



1 p1 y

1 2p 1 WJP































xB 1 x L


M xB 1 x L m 0 W xB





Case

Type of official

M m W






that ifB

L B


L

B

L B


L

B

L DD



xB Then

WRD wJP M wDD m wJP

Wxw 1 xwJP



WRD WDD WJP

since

w wDD wJP

















APPENDIX



















2 q 1








(A1) q 1 q 1 1 q13


1 13 0










REFERENCES











































I Overview

A Outline


























II The Basic Model













(1) G R

G
















WDD 2p


WJP 2




WRD p


WRD max WDD WJP








p

p 1 1 p


1 p

1 p 1 p


WRD 1 p 1 p

(1)p[12p(1)]


(2) 2p 1

1 2


ldquoboundaryrdquo1

21 2 p between




G R



B Mechanism Design



1 xj xi

and will pander if

1 xj xi





WDD 2p and WJP 2


WRD p max WDD WJP


WRD 1 2p1



WRD DD 2p p2 2p2





(3) 1 pG c




(4) pG G R c 1 pG c

pG G R c


c

G 1 p

1 p 1





D Term Lengths











JP WJP

2



RD WRD

2 JP


F Campaign Promises











IV Feedback



WDD 2p

and

WJP 2




13C a 1 q 1 qxa xb

13N b 1 q 1 qxa xb

13N a 1 q 1 qxa xb

13C b 1 q 1 qxa xb








1 2q 1



q 1


1 WJP




q 1


1 p

1 p 1 py

or

y 1

p 1



1 p1 y

1 2p 1 WJP































xB 1 x L


M xB 1 x L m 0 W xB





Case

Type of official

M m W






that ifB

L B


L

B

L B


L

B

L DD



xB Then

WRD wJP M wDD m wJP

Wxw 1 xwJP



WRD WDD WJP

since

w wDD wJP

















APPENDIX



















2 q 1








(A1) q 1 q 1 1 q13


1 13 0










REFERENCES


























































II The Basic Model













(1) G R

G
















WDD 2p


WJP 2




WRD p


WRD max WDD WJP








p

p 1 1 p


1 p

1 p 1 p


WRD 1 p 1 p

(1)p[12p(1)]


(2) 2p 1

1 2


ldquoboundaryrdquo1

21 2 p between




G R



B Mechanism Design



1 xj xi

and will pander if

1 xj xi





WDD 2p and WJP 2


WRD p max WDD WJP


WRD 1 2p1



WRD DD 2p p2 2p2





(3) 1 pG c




(4) pG G R c 1 pG c

pG G R c


c

G 1 p

1 p 1





D Term Lengths











JP WJP

2



RD WRD

2 JP


F Campaign Promises











IV Feedback



WDD 2p

and

WJP 2




13C a 1 q 1 qxa xb

13N b 1 q 1 qxa xb

13N a 1 q 1 qxa xb

13C b 1 q 1 qxa xb








1 2q 1



q 1


1 WJP




q 1


1 p

1 p 1 py

or

y 1

p 1



1 p1 y

1 2p 1 WJP































xB 1 x L


M xB 1 x L m 0 W xB





Case

Type of official

M m W






that ifB

L B


L

B

L B


L

B

L DD



xB Then

WRD wJP M wDD m wJP

Wxw 1 xwJP



WRD WDD WJP

since

w wDD wJP

















APPENDIX



















2 q 1








(A1) q 1 q 1 1 q13


1 13 0










REFERENCES













































(1) G R

G
















WDD 2p


WJP 2




WRD p


WRD max WDD WJP








p

p 1 1 p


1 p

1 p 1 p


WRD 1 p 1 p

(1)p[12p(1)]


(2) 2p 1

1 2


ldquoboundaryrdquo1

21 2 p between




G R



B Mechanism Design



1 xj xi

and will pander if

1 xj xi





WDD 2p and WJP 2


WRD p max WDD WJP


WRD 1 2p1



WRD DD 2p p2 2p2





(3) 1 pG c




(4) pG G R c 1 pG c

pG G R c


c

G 1 p

1 p 1





D Term Lengths











JP WJP

2



RD WRD

2 JP


F Campaign Promises











IV Feedback



WDD 2p

and

WJP 2




13C a 1 q 1 qxa xb

13N b 1 q 1 qxa xb

13N a 1 q 1 qxa xb

13C b 1 q 1 qxa xb








1 2q 1



q 1


1 WJP




q 1


1 p

1 p 1 py

or

y 1

p 1



1 p1 y

1 2p 1 WJP































xB 1 x L


M xB 1 x L m 0 W xB





Case

Type of official

M m W






that ifB

L B


L

B

L B


L

B

L DD



xB Then

WRD wJP M wDD m wJP

Wxw 1 xwJP



WRD WDD WJP

since

w wDD wJP

















APPENDIX



















2 q 1








(A1) q 1 q 1 1 q13


1 13 0










REFERENCES









































WDD 2p


WJP 2




WRD p


WRD max WDD WJP








p

p 1 1 p


1 p

1 p 1 p


WRD 1 p 1 p

(1)p[12p(1)]


(2) 2p 1

1 2


ldquoboundaryrdquo1

21 2 p between




G R



B Mechanism Design



1 xj xi

and will pander if

1 xj xi





WDD 2p and WJP 2


WRD p max WDD WJP


WRD 1 2p1



WRD DD 2p p2 2p2





(3) 1 pG c




(4) pG G R c 1 pG c

pG G R c


c

G 1 p

1 p 1





D Term Lengths











JP WJP

2



RD WRD

2 JP


F Campaign Promises











IV Feedback



WDD 2p

and

WJP 2




13C a 1 q 1 qxa xb

13N b 1 q 1 qxa xb

13N a 1 q 1 qxa xb

13C b 1 q 1 qxa xb








1 2q 1



q 1


1 WJP




q 1


1 p

1 p 1 py

or

y 1

p 1



1 p1 y

1 2p 1 WJP































xB 1 x L


M xB 1 x L m 0 W xB





Case

Type of official

M m W






that ifB

L B


L

B

L B


L

B

L DD



xB Then

WRD wJP M wDD m wJP

Wxw 1 xwJP



WRD WDD WJP

since

w wDD wJP

















APPENDIX



















2 q 1








(A1) q 1 q 1 1 q13


1 13 0










REFERENCES










































p

p 1 1 p


1 p

1 p 1 p


WRD 1 p 1 p

(1)p[12p(1)]


(2) 2p 1

1 2


ldquoboundaryrdquo1

21 2 p between




G R



B Mechanism Design



1 xj xi

and will pander if

1 xj xi





WDD 2p and WJP 2


WRD p max WDD WJP


WRD 1 2p1



WRD DD 2p p2 2p2





(3) 1 pG c




(4) pG G R c 1 pG c

pG G R c


c

G 1 p

1 p 1





D Term Lengths











JP WJP

2



RD WRD

2 JP


F Campaign Promises











IV Feedback



WDD 2p

and

WJP 2




13C a 1 q 1 qxa xb

13N b 1 q 1 qxa xb

13N a 1 q 1 qxa xb

13C b 1 q 1 qxa xb








1 2q 1



q 1


1 WJP




q 1


1 p

1 p 1 py

or

y 1

p 1



1 p1 y

1 2p 1 WJP































xB 1 x L


M xB 1 x L m 0 W xB





Case

Type of official

M m W






that ifB

L B


L

B

L B


L

B

L DD



xB Then

WRD wJP M wDD m wJP

Wxw 1 xwJP



WRD WDD WJP

since

w wDD wJP

















APPENDIX



















2 q 1








(A1) q 1 q 1 1 q13


1 13 0










REFERENCES









































B Mechanism Design



1 xj xi

and will pander if

1 xj xi





WDD 2p and WJP 2


WRD p max WDD WJP


WRD 1 2p1



WRD DD 2p p2 2p2





(3) 1 pG c




(4) pG G R c 1 pG c

pG G R c


c

G 1 p

1 p 1





D Term Lengths











JP WJP

2



RD WRD

2 JP


F Campaign Promises











IV Feedback



WDD 2p

and

WJP 2




13C a 1 q 1 qxa xb

13N b 1 q 1 qxa xb

13N a 1 q 1 qxa xb

13C b 1 q 1 qxa xb








1 2q 1



q 1


1 WJP




q 1


1 p

1 p 1 py

or

y 1

p 1



1 p1 y

1 2p 1 WJP































xB 1 x L


M xB 1 x L m 0 W xB





Case

Type of official

M m W






that ifB

L B


L

B

L B


L

B

L DD



xB Then

WRD wJP M wDD m wJP

Wxw 1 xwJP



WRD WDD WJP

since

w wDD wJP

















APPENDIX



















2 q 1








(A1) q 1 q 1 1 q13


1 13 0










REFERENCES











































(3) 1 pG c




(4) pG G R c 1 pG c

pG G R c


c

G 1 p

1 p 1





D Term Lengths











JP WJP

2



RD WRD

2 JP


F Campaign Promises











IV Feedback



WDD 2p

and

WJP 2




13C a 1 q 1 qxa xb

13N b 1 q 1 qxa xb

13N a 1 q 1 qxa xb

13C b 1 q 1 qxa xb








1 2q 1



q 1


1 WJP




q 1


1 p

1 p 1 py

or

y 1

p 1



1 p1 y

1 2p 1 WJP































xB 1 x L


M xB 1 x L m 0 W xB





Case

Type of official

M m W






that ifB

L B


L

B

L B


L

B

L DD



xB Then

WRD wJP M wDD m wJP

Wxw 1 xwJP



WRD WDD WJP

since

w wDD wJP

















APPENDIX



















2 q 1








(A1) q 1 q 1 1 q13


1 13 0










REFERENCES














































JP WJP

2



RD WRD

2 JP


F Campaign Promises











IV Feedback



WDD 2p

and

WJP 2




13C a 1 q 1 qxa xb

13N b 1 q 1 qxa xb

13N a 1 q 1 qxa xb

13C b 1 q 1 qxa xb








1 2q 1



q 1


1 WJP




q 1


1 p

1 p 1 py

or

y 1

p 1



1 p1 y

1 2p 1 WJP































xB 1 x L


M xB 1 x L m 0 W xB





Case

Type of official

M m W






that ifB

L B


L

B

L B


L

B

L DD



xB Then

WRD wJP M wDD m wJP

Wxw 1 xwJP



WRD WDD WJP

since

w wDD wJP

















APPENDIX



















2 q 1








(A1) q 1 q 1 1 q13


1 13 0










REFERENCES












































IV Feedback



WDD 2p

and

WJP 2




13C a 1 q 1 qxa xb

13N b 1 q 1 qxa xb

13N a 1 q 1 qxa xb

13C b 1 q 1 qxa xb








1 2q 1



q 1


1 WJP




q 1


1 p

1 p 1 py

or

y 1

p 1



1 p1 y

1 2p 1 WJP































xB 1 x L


M xB 1 x L m 0 W xB





Case

Type of official

M m W






that ifB

L B


L

B

L B


L

B

L DD



xB Then

WRD wJP M wDD m wJP

Wxw 1 xwJP



WRD WDD WJP

since

w wDD wJP

















APPENDIX



















2 q 1








(A1) q 1 q 1 1 q13


1 13 0










REFERENCES









































1 2q 1



q 1


1 WJP




q 1


1 p

1 p 1 py

or

y 1

p 1



1 p1 y

1 2p 1 WJP































xB 1 x L


M xB 1 x L m 0 W xB





Case

Type of official

M m W






that ifB

L B


L

B

L B


L

B

L DD



xB Then

WRD wJP M wDD m wJP

Wxw 1 xwJP



WRD WDD WJP

since

w wDD wJP

















APPENDIX



















2 q 1








(A1) q 1 q 1 1 q13


1 13 0










REFERENCES




































































xB 1 x L


M xB 1 x L m 0 W xB





Case

Type of official

M m W






that ifB

L B


L

B

L B


L

B

L DD



xB Then

WRD wJP M wDD m wJP

Wxw 1 xwJP



WRD WDD WJP

since

w wDD wJP

















APPENDIX



















2 q 1








(A1) q 1 q 1 1 q13


1 13 0










REFERENCES












































that ifB

L B


L

B

L B


L

B

L DD



xB Then

WRD wJP M wDD m wJP

Wxw 1 xwJP



WRD WDD WJP

since

w wDD wJP

















APPENDIX



















2 q 1








(A1) q 1 q 1 1 q13


1 13 0










REFERENCES















































APPENDIX



















2 q 1








(A1) q 1 q 1 1 q13


1 13 0










REFERENCES




















































2 q 1








(A1) q 1 q 1 1 q13


1 13 0










REFERENCES












































(A1) q 1 q 1 1 q13


1 13 0










REFERENCES











































REFERENCES









































the politician and the judge: accountability in government · the politician and the judge:...

Documents