uc

b Stanford Graduate School of Business. 518 Memorial Wa ted

a r

ArtiRecRecAccAva

KeyElecPanMed

icympouatbeshethlime, ine

results by focusing on how media commentary affects voteto the i

I am persuaded myself that the good sense of the people will alwaysbe found to be the best army. They may be

e peoprs willpunishard of tons of tthe chauld penmentse to ke

shortcuts and

Journal of Public Economics 94 (2010) 838847

Contents lists available at ScienceDirect

Journal of Public Economics

j ourna l homepage: www.e lsev ie r.com/ locate / jpubehesitate a moment to prefer the latter.Thomas Jefferson, 1787

I may not agree with everything you write or report....But I do sowith the knowledge that when you are at your best then you helpme be at my best. You help all of us who serve at the pleasure ofthe American people do our jobs better, by holding us accoun-

accountability. Voters in the model are sometimes misinformed abouttheir true interests. The incumbent policy maker has better informa-tion about optimal policy choices, and may thus have an incentive topanderto implement a policy that voters believe is in their bestinterest, even though the policy maker's superior informationindicates the voters are wrong (Canes-Wrone et al., 2001; Maskinand Tirole, 2004; Prat, 2005).3table, by demanding honesty, by prev

Corresponding author. Tel.: +1 773 251 9563.E-mail addresses: sashwort@uchicago.edu (S. Ashwo

(K.W. Shotts).1 Tel.: +1 650 725 4510.

0047-2727/$ see front matter 2010 Elsevier B.V. Adoi:10.1016/j.jpubeco.2010.06.013ep that right; and were itve a government withoutvernment, I should not

citizens about the merits of particular policy choices, and thereby, theargument goes, enhance electoral accountability.2

We analyze this Jeffersonian intuition in a formal model of political

left to me to decide whether we should hanewspapers or newspapers without a gobut will soon correct themselves. Ththeir governors: and even their errotrue principles of their institution. Towould be to suppress the only safeguto prevent these irregular interpositifull information of their affairs thro'& to contrive that those papers shothe people. The basis of our governpeople, the very rst object should bled astray for a moment,le are the only censors oftend to keep these to thethese errors too severelyhepublic liberty. Thewayhe people is to give themnnel of the public papers,etrate the whole mass ofbeing the opinion of the

desperately weary of.Barack Obama, 2009

An active media is often thought to be essential to a well-functioning democracy. When politicians are accountable to voters,the people must be well-informed, lest the government respond tomistaken voter impulses. In Jefferson's terms, newspapers, then,ensure that the opinion of the people is kept right by educatingenting us from taking

rth), kshotts@stanford.edu

2 It is, of course, noprovide ex ante edumerits of particular platter, which is whafamous quotation in

3 Swank and Vissetheir model, incumbvetting them, becauand failing to come

ll rights reserved.falling into easy political games that people are soa b s t r a c tt i c l e i n f o

cle history:eived 10 September 2008eived in revised form 18 May 2010epted 11 June 2010ilable online 7 July 2010

words:tionsderingia

Elections sometimes give polbest interest, even though theincentives are typically attenincumbent truly was in theirvoters information about wmonitoring does sometimes epanderingworse. Furthermora yes mansuppressing somthese seemingly paradoxicalasymmetric burden of proofy, Stanford CA 94305-5015, UniDoes informative media commentary red

Scott Ashworth a,, Kenneth W. Shotts b,1

a Unviersity of Chicago Harris School.1155 E. 60th Street, Chicago, IL 60637, United Statesncumbent, based on whether she pursues popular or unpopular policies. 2010 Elsevier B.V. All rights reserved.t clear in Jefferson's letter whether he thcation about policy in general or ex poolicies that politicians had pursued. In tt President Barack Obama had in mindhis speech at the 2009 White House Cr (2006) study a related mechanism inents have an incentive to implementse voters cannot distinguish between tup with an idea for a project at all.rong choice. We explainrs' tendency to apply ane politicians' incentives to pander?

States

akers incentives to pander, i.e., to implement a policy that voters think is in theirlicymaker knows that a different policy is actually better for the voters. Panderinged when voters learn, prior to the election, whether the policy chosen by thet interest. This suggests that themedia can improve accountability by reporting toer an incumbent made good policy choices. We show that, although mediainate the incumbent's incentive to pander, in other cases it makes the problem ofsome circumstances incumbent incentives are improvedwhen themedia acts as

information that indicates the policy maker made the wought newspapers wouldst information about thehis paper we focus on thewhen he cited Jefferson'sorrespondents dinner.a moral hazard context. Inprojects without properlyhe rejecting a bad project

We start with a baseline model without a media, and then add amedia outlet (the newspaper). The newspaper gets privateinformation about which policy best serves voters' interests, and itacts as a commentator, making statements about whether theincumbent chose the correct policy. The newspaper gets to observe

839S. Ashworth, K.W. Shotts / Journal of Public Economics 94 (2010) 838847the policy choice before commenting, so it may, in the event that itsees a weak signal indicating that the incumbent chose the wrongpolicy, act as a yes man, herding on the incumbent's choice(Bikhchandani et al., 1992; Scharfstein and Stein, 1990; Prendergast,1993; Efnger and Polborn, 2001). The newspaper comments beforethe next election and is thus relevant for voters' decision to retain orreplace the incumbent, creating the prospect that media commentarymight discipline an electorally motivated incumbent's policy choice.

Our key question is: How does the presence of the newspaperaffect the incumbent politician's incentives to pander?4 In thinkingabout this question, a crucial distinction is whether the newspapergives information about the incumbent's policy choice or about thestate of the world. Prat (2005) shows that, when information aboutthe state of the world is poor, pandering incentives can be eliminatedby not informing voters of the incumbent's policy choice. But whenpolicy choices are observed, it would seem that incumbents will haveless incentive to pander when the media gives voters informationabout whether the incumbent acted in their best interest. After all,Canes-Wrone et al. (2001) and Prat (2005) both show that panderingincentives are lower the greater is the probability that the votersobserve the true state before the election.

Our model partially conrms this Jeffersonian intuition; in somecircumstances, introducing a media commentator eliminates pander-ing. But themedia does not always eliminate pandering; introducing afallible media can in fact lead to pandering when it would not haveoccurredwithout amedia. An additional surprising result is that a yes-man media is in many circumstances more effective than a truthfulmedia at reducing pandering.

To understand why the Jeffersonian intuition can fail, it helps totake a closer look at the basic pandering incentive without a media.Pandering arises because, when the incumbent and the challenger areclose in terms of their prior reputations, voters treat the incumbent'spossible policy choices asymmetrically. They reelect an incumbentwho chooses a popular action unless he is proved wrong, but theyonly reelect an incumbent who chooses an unpopular action if he isproved right. If, at the time of the next election, it is unclear whichpolicy choice was correct, voters will reelect the incumbent if and onlyif he chose the initially popular policy. This asymmetric burden ofproof creates the incentive for panderingif the public is sufcientlyunlikely to learn whether the incumbent's policy choice was correct,then choosing the action with a lower burden of proof is optimal, evenwhen that action is unlikely to be correct.

The media eliminates pandering when it induces the voter to treatinitially popular and initially unpopular actions symmetrically. Forexample, in some circumstances the incumbent is reelected unless hisaction is somehow shown to be incorrect, either by a clear publicsignal that he indeed chose the wrong policy or, in the absence of suchpublic information, by informative media commentary criticizing theincumbent's action. The two actions thus lead to the same burden ofproof, and the incumbent has no incentive to pander.

Focusing on asymmetric voter responses also helps explain thepotential benet of a yes-man media, one unwilling to criticize theincumbent unless it observes overwhelmingly clear information thatthe incumbent chose the wrong policy. When the media is a yes man,its contradictory reports are denitive, i.e., voters know that the

4 Besley and Prat (2006) analyze whether the media can discipline kleptocraticgovernment ofcials. Egorov et al. (2007) analyze tradeoffs faced by an autocrat whocan use a free media to acquire information about bureaucrats' performance, but whoworries the media might instigate a revolution by informing voters that he himself has

performed badly.media only criticizes the incumbent when it is sure that he chose thewrong policy. As a result, voters treat an incumbent criticized by themedia after taking the popular action exactly as they treat anincumbent who is criticized after taking the unpopular action. Thissymmetry gives a yes-man newspaper a leg up on eliminatingpandering. If the media does not act as a yes man, then itscontradictory reports are not denitive, and the incumbent mayhave a prior reputation sufciently superior to that of his challengerthat he canwin in the face of media criticism of the popular action, butnot the unpopular one.

Although pandering is a response to asymmetric voter responses, itis important to note that, in our model, the media is completely evenhanded, i.e., symmetric, in its treatment of politicians and of policiesitdoes not favor a particular candidate, nor does it favor any particularpolicy. This may seem strange given the extant literature's focus onmedia bias, both in studies that empirically estimate bias (Grosecloseand Milyo, 2005; Ho and Quinn, 2008; Gentzkow and Shapiro, 2007)and in studies that examine its origins and relationship to competition inthe media industry (Page, 1996; Arnold, 2004; Baron, 2006; Gentzkowand Shapiro, 2006; Bernhardt et al., 2006). In the analysis that followswe set aside all issues related to bias and competition, i.e., we analyze amodel with a single, unbiased, media outlet. We do so not because wendbias and competition uninteresting; on the contrary,we think theseissues are quite important. However, for the purpose of our analysis it isimportant to use a relatively optimistic model, in which the mediafocuses on providing information rather than trying to push for aparticular policy. Ultimately, our goal is to assess a simple and seeminglycompelling intuitionthat by providing information the media reducesincentives for pandering. Ourmost surprising resultshave todowith thefact that this intuition often fails, for reasons independent of bias. Evenan unbiased media does not necessarily eliminate pandering, andindeed it can sometimes aggravate the problem.

The paper proceeds as follows. We rst introduce the model. Next,we analyze how the media affects incentives for pandering byconsidering a baseline model without a media and then adding inmedia commentary. Finally, we briey extend the model to allow fordifferent media motivations: a reputation-motivated media as well asone that always reports its information and never acts as a yes man.

1. The model

We want to identify the impact of the media's announcement onthe incumbent's incentives to take the correct action. To isolate themedia's impact, we start with a baselinemodel without amedia, usinga simplied variant of the model in Canes-Wrone et al. (2001). Theheart of our analysis modies this baseline model, adding acommentator to explore the media's effect on politicians' incentivesto pander.

1.1. Baseline model

1.1.1. Policies and preferencesIn each of two periods, a policymust be selected from the set {A, B}.

The optimal policy in a period depends on the state of the world inthat period, {A, B}. A representative voter gets payoff 1 for eachperiod in which the policy matches the state, and 0 for each period inwhich policy does not match the state. The state of the world isindependent across the two periods, and in each period state A is morelikely: Pr(=A)=N1/2. The fact that one state is ex ante strictlymore likely than the other is critical for there to be incentives forpandering. We do not introduce additional notation to distinguish thetwo periods, because almost all of the action in the model occurs inthe rst period. There is no discounting.

In period 1, an incumbent policy maker chooses policy xI {A, B}.At the end of this period, the voter can either reelect the incumbent or

replace him with a challenger. A politician gets payoff N0 for

matching his policy choice to the state, plus an ego rent of 1 for eachperiod that he holds ofce.5

1.1.2. Information structureAt the beginning of each period, the voter has no information

about the state, aside from the prior. The policy maker, on the otherhand, gets an informative private signal, s, about the state. This signal's

media's inability to fully convey to voters all of the subtleties of itsinformation. It is important for our analysis that the set of signalsthe newspaper might receive is richer than the set of messages it cansendthis is what makes yes-man behavior possible.

1.3. Equilibrium concept

We focus on perfect accountability equilibria, that is, perfect

Table 1Likelihoods of the newspaper's signals in the two states.

sN Pr(sN|=A) Pr(sN|=B)

AH N 0AL q(1N) (1q)(1N)BL (1q)(1N) q(1N)BH 0 N

840 S. Ashworth, K.W. Shotts / Journal of Public Economics 94 (2010) 838847precision depends on his type, {H, L}. A high quality type learns thetrue state,

Pr s = j = H = 1;

whereas a low quality type gets an imperfect signal,

Pr s = j = L = q N :

By Bayes's Rule,

Pr = B js = B; = L = 1q1q + 1q N12

;

so the restriction that qN ensures that even a low quality policymaker's signal outweighs the prior. Types are private information andthe prior probabilities that the candidates (incumbent and challenger)are high quality are:

Pr I = H = IPr C = H = C :

We say that the election is competitive if I and C are close together.With probability , the voter learns the true rst period state

before election day; otherwise he votes knowing only the policychoice, xI. Formally, the voter's signal is sV {A, B, }, where meansno information. If uncertainty resolves then sV=. A low meansthat either the election is imminent so there is little time forinformation to be publicly revealed or that the policy being chosen isunlikely to produce any easily assessed short run effects.

1.2. Adding the media

The heart of the paper adds a media commentator to the baselinemodel. We will refer to this commentator as a newspaper, thoughobviously the commentator could be some other media outlet. Afterthe incumbent chooses policy, the newspaper makes an announce-ment xN{A, B}, declaring which state of the world it believes is morelikely. The newspaper maximizes the probability that its announce-ment matches the true state, i.e., it is intrinsically motivated to givethe voter its best assessment of the state.6

The newspaper bases its belief about the true state on the prior, theincumbent's action, and a private signal, sN{AH, AL, BL, BH }. The signallikelihoods are given in Table 1. This information structure is, in someways, similar to that of the incumbent. A signal is characterized byboth the state it indicates is more likely and by how precise it is.Signals subscripted by H perfectly reveal the state, while thosesubscripted by L are correct only with conditional probability q. Theprobability of a perfectly revealing signal is N.

There is, however, a crucial difference between the commentatorand the incumbent: the newspaper has only one type, and mayreceive a signal of either precision. Substantively, this just reects the

5 Politicians do not care about policy when they are not in ofce. This is a sufcient,though not necessary, condition to ensure that a low-quality incumbent will not seekto lose ofce in the hopes of being replaced by a higher-quality ofcial.

6 Gentzkow and Shapiro (2006) suggest a different model of media motivations, inwhich the media is concerned about potential customers' beliefs about its quality.Given that our primary focus is on the effects of media behavior, rather than the mediaitself, we use the simpler assumption that the media is motivated by a desire to reportthe truth. In the section on extensions, we analyze the case of reputational

motivations.Bayesian equilibria in which the incumbent matches his action tohis signal. Such an equilibrium has the normatively desirable propertythat the incumbent uses his information optimally to promote thevoter's policy interests. With this focus, we can sharpen our mainquestion: does the presence of a commentator make the existence of aperfect accountability equilibrium more or less likely? And how doesthe existence of a perfect accountability equilibrium depend onfactors such as the competitiveness of the election (|IC|) and theprobability that uncertainty resolves ()?

When no perfect accountability equilibrium exists, there will be apandering equilibrium, in which a low quality incumbent sometimeschooses an action that matches voters' prior beliefs about the correctpolicy, but that does not promote their interests. Details of suchequilibria are available upon request.

2. The baseline model

We can now analyze policy choice in the rst period of our baselinemodel. In this baseline, there is no newspaper and the timing is:

This baseline model is a simplied version of Canes-Wrone et al.(2001). As such, we can appeal to a variant of their Proposition 1 for adelineation of when perfect accountability equilibria exist. Weillustrate their result in Fig. 1. The challenger's probability of beingFig. 1. Baseline model.

high quality, C, is plotted on the horizontal axis, while the probabilityof uncertainty resolution, , is on the vertical axis. For the formalstatement, we need the following notation: Let (xI ,sV) be the voter'sposterior belief about incumbent quality given rst period policy xI,public signal sV, and the (perfect accountability) equilibrium conjec-ture that the incumbent chooses xI=sI. In Fig. 1, these posteriorsdivide the horizontal axis into three regions. In two of the regions, the

3. Media commentary

Nowwe introduce the newspaper, whichmakes an announcementafter observing the incumbent's action. The timing is:

841S. Ashworth, K.W. Shotts / Journal of Public Economics 94 (2010) 838847incumbent has straightforward incentives to follow his signal. In theremaining region, incentives depend on , the probability thatinformation resolves. To state the precise cutoff, let v(L)=1+q bethe value of reelection to a low quality incumbent, let B be a lowquality incumbent's posterior probability the state is A given a signalof B, and dene

2B1 + L2L1B

: 1

Proposition 1. In the baseline model, there is a perfect accountabilityequilibrium if one of the following conditions holds:

1. Challenger reputation is worse than incumbent who chooses xI=B, i.e.Cb (B, ).

2. Challenger reputation is better than incumbent who chooses xI=A, i.e.C (A, ).

3. Uncertainty resolution is likely, i.e. .

We give the basic intuition for this result below. This is importantbecause, although the Canes-Wrone et al. (2001) model providesinteresting insights about pandering, that paper focused on state-ments of equilibria, leaving rather opaque many parts of the logic forhow the model works. So to build on that model we must rst clarifythe foundations.

An implication of Bayes's Rule is that an incumbent whose signalagrees with the prior is more likely to have an accurate signal than isan incumbent whose signal disagrees with the prior: (B, )b (A, ).In this case, if the incumbent follows his signal, the optimal reelectionrule as a function of the voter's information at the time of the electionis that shown in Table 2.

As in several other recent papers (Gentzkow and Shapiro, 2006;Prat, 2005), Canes-Wrone et al. (2001) show that the asymmetry inthis rule creates incentives for pandering, i.e., an incumbent choosinga policy that voters believe is optimal even though the incumbent hasinformation indicating that it is not. Basically, if the challenger'sreputation falls within the gap between (B, ) and (A, ), the voterapplies a higher burden of proof to an incumbent who chooses the exante unpopular action B than to an incumbent who chooses A; hereelects the incumbent when xI=A but not when xI=B. Thisasymmetric burden of proof can lead to pandering.

The intuition is straightforward. In the region C( (B, ), (A, ),) the incumbent and the challenger have similar priorprobabilities of being high quality. Thus, given that the voter updatespositively about the incumbent after he chooses A and negatively whenhe chooses B, the incumbent knows he faces an asymmetric burden ofproof that he must satisfy in order to be reelected, and he has anincentive to choose A. However, if uncertainty resolves he will winreelection if and only if he chose the correct policy, so if the probabilityof uncertainty resolution is sufciently high, i.e., N , there exists aperfect accountability equilibrium.

Table 2Election winner when C( (B,), (A,)).

xI=A xI=B

=A Incumbent Challenger=B Challenger Incumbent Incumbent ChallengerAs usual, we solve the model by working backwards from the end.

3.1. Second period policy and continuation values

As there are no subsequent periods, the period 2 policy maker isconcerned only with the immediate impact of his policy choice. Hegets utility N0 frommatching the state and zero otherwise, and thus,because qN, he will choose the policy corresponding to the state thathis signal indicates is more likely.

With this second period strategy in hand, the optimal election ruleis easy to derive. The voter wants second period policy to match thestate. And he knows that whoever he elects will follow his signal inthe second period. Thus the voter elects whichever candidate hebelieves is more likely to be a high quality type. Writing (h) for thevoter's assessment of the probability that the incumbent is a high typegiven history h, an optimal election rule is:

reelect the incumbent if and only if h C

where C is the probability that the challenger is high quality.7

Given voter behavior and second period concerns, we can derive anexpression for the incumbent's indirect payoff function for period1. Hegets an immediate payoff of if he picks the correct policy, xI=, inperiod 1. He gets an additional positive payoff if he is reelected:

v 1 + I I = H + qI I = L ; 2

where I[p] is the indicator function for the proposition p. The rstterm in the expression, 1, is the direct payoff from holding ofce,while the second term reects the probability that the incumbent willchoose the right action if he is reelected. Note that the second termdepends on the incumbent's type.

An incumbent with type I who sees signal sI in period 1 thuschooses policy xI to solve:

maxxI A;Bf g

PrxI = jsI ; I + vIPr hC jxI; sI :

The rst term is the incumbent's direct payoff from his rst periodpolicy choice, while the second term is his payoff from reelectionweighted by the probability of reelection conditional on his signal andpolicy choice.

3.2. Simplifying voter beliefs

The incumbent's reelection chances depend on the voter's beliefsabout his type, so these beliefs, especially the posterior (h), play acentral role in the analysis. With media commentary, it can becomplicated to directly calculate this probability, which depends onthe newspaper's strategy, the incumbent's strategy, and their realized

7 To be precise if (h)NC the voter must reelect, if (h)bC the voter must removethe incumbent, and if (h)=C the voter is indifferent. The incumbent plays a purestrategy in a perfect accountability equilibrium so only for knife-edge parameter

values will it be the case that (h)=C.

actions. Fortunately, focusing on perfect accountability equilibriasubstantially simplies these calculations.

Updating is particularly straightforwardwhen uncertainty resolves

Lemma 2. Under Assumption 1, in any perfect accountability equilibri-um, the newspaper is a partial yes man. When the newspaper sees astrong signal, AH or BH , it follows this signal, announcing xN=A or xN=Brespectively. However, if the newspaper sees a weak signal, AL or BL, it

842 S. Ashworth, K.W. Shotts / Journal of Public Economics 94 (2010) 838847and the voter observes the true state, sV=. In this case, the voterknows for sure whether the incumbent's policy choice was correct. Ifthe incumbent chose thewrong policy, then the voter's conjecture thatthe incumbent follows his signal and the fact that only the low type canget a wrong signal combine to imply that the probability theincumbent is high quality is 0. On the other hand, if the incumbentchose the correct policy, then the voter knows that sI=. There aretwo ways this can happen: the incumbent is high quality or he is lowquality and got the right signal. Thus Bayes's Rule gives the voter'sbelief about the probability that the incumbent is high quality as

II + q1I

:

What if the voter does not observe the true state? Because highability types always set xI= sI= in a perfect accountabilityequilibrium, we can write the posterior in a simple form.

Lemma 1. Assume that the incumbent matches his action to his signal.Then

h = Pr = xI jh :

The proofs for this lemma and several other results are in theAppendix. Intuitively, the voter rst revises his beliefs about the truestate of the world, Pr(=xI | h), and then uses those beliefs about thestate as weights to form beliefs about the incumbent's type. Thesebeliefs are a weighted average of the best case belief, +, when theincumbent's policy choice is known to be correct, and the worst casebelief, 0, when the incumbent's policy choice is known to be wrong.

3.3. Newspaper behavior

The newspaper sometimes acts as a yes man, ignoring its ownsignal and simply following the incumbent's lead. To see this, supposethe newspaper sees a weak signal that B is the correct policy but alsosees the incumbent choose xI=A. In a perfect accountabilityequilibrium, the newspaper infers that the incumbent's signal wasA, so the two signals point in different directions. And the newspaperknows that the incumbent, who might be a high type, gets a signalthat is more accurate (on average) than its own, weak, signal. The netimpact of these observations is to tilt the newspaper's posteriortowards =A. Along with the prior bias towards A, this ensures thatthe newspaper believes the probability that =A is greater than 1/2.And because the newspaper wants to match its announcement to thestate, it will ignore its signal and announce xN=A.

Wewant to explore the implications of a yes-mannewspaper in thesimplest possible case. To this end, we assume that the incumbent issufciently likely to be high quality, and hence to correctly match thepolicy and the state, so the commentator will be a yes man even whenthe incumbent's policy choice is B. The following assumption sufces8:

Assumption 1.

Iq21

q21 + 1 :

With the assumption, the commentator's behavior is straightfor-ward to characterize.

8 Without this assumption, conditions for the existence of a perfect accountabilityequilibrium will be a mixture of the conditions we derive in the current section andthose we discuss below for the case of a nonstrategic media. We leave the details to

the interested reader.always says that the incumbent was right, announcing xN=xI .

Proof. Recall that the newspaper seeks to maximize the probabilitythat xN=. Thus the claim follows from two observations. First, whenit sees a strong signal, the newspaper knows that its signal is correctwith probability 1, so it announces its signal truthfully. Second, a weaksignal AL or BL is outweighed by the incumbent's signal, which istruthfully revealed in a perfect accountability equilibrium. To see thislast point, use Bayes's Rule to write the newspaper's posterior as

Pr = A jsI = B; sN = AL

=1I1qq

1I1qq + 1 I + 1Iq 1q

which is less than or equal to 1=2 because, under Assumption 1,I q21q21 + 1.

If the newspaper gets imprecise information then it acts as a yesman, always announcing that the incumbent chose the correct policy.With precise information, on the other hand, the newspaper reportsits true signal rather than acting as a yes man. So the model predictsthat media commentators sometimes suppress evidence that incum-bents havemade policymistakes. It is particularly noteworthy that weget this result even though the media's sole objective is to giveaccurate information about the state of the world. In particular, theresult is not driven by collusion or side payments because in ourmodel the newspaper cannot be bought off by the incumbent.9

As an aside, it is also worth noting that our model can be extendedto analyze MN1 media outlets, who move sequentially, in the sensethat rst the incumbent chooses policy, then the rst media outletmakes its announcement, then the second media outlet makes itsannouncement, and so on. In such a model, as long as no previousmedia report has contradicted the incumbent's policy choice, mediaoutlets who receive a weak signal AL or BLwill act as yes men, whereasthosewho see a denitive signal AH or BHwill follow their signals. Thusour model can be easily reinterpreted to cover this case, by changingthe media's (collective) probability of seeing a perfectly revealingsignal from N in the variant we present here to N=1(1N)M .

3.4. Voter updating

A history at which the voter votes is a triple h=(xI, xN, sV), so wecan write (xI, xN, sV) for her posterior belief that the incumbent ishigh quality, under the conjecture that the incumbent sets xI= sI andthat Lemma 2 describes the newspaper's strategy. Lemma 1 lets uscalculate these posteriors for every history that the voter mightobserve. What really matters is their ranking.

Lemma 3. With media commentary,

0 = A; xN ;B = B; xN ;A = B;A; = A;B;b B;B;b A;A;b A; xN ;A = B; xN ;Bb1:

Proof. The newspaper disagrees with the incumbent only if it has asignal that showswith certainty that the incumbent choose thewrongpolicy. Since this can only happen if the incumbent is low quality, wehave 0=(B, A, )=(A, B, ).

9 For a model in which incumbents may buy off the media, see Besley and Prat

(2006).

It remains to show that (B, B, )b(A, A, ). By Lemma 1, thisinequality is equivalent to

Pr = B jxI = B; xN = B bPr = A jxI = A; xN = A

where

Pr = B jxI = B =1 I + q1I

1I + q1I + 1q1I1N

and

Pr = A jxI = A =I + q1I

I + q1I + 11q1I1N:

Substituting and simplifying, the inequality reduces to

Table 3Election winner when Cb(B,B,).

xI=A xI=B

=A Incumbent Challenger=B Challenger IncumbentxN=A Incumbent ChallengerxN=B Challenger Incumbent

Table 4Election winner when C((A, A, ), (A, xN, A)) or CN (A, xN, A).

C((A, A, ), (A, xN, A)) CN (A, A)

xI=A xI=B xI=A xI=B

=A Incumbent Challenger =A Challenger Challenger=B Challenger Incumbent =B Challenger ChallengerxN=A Challenger Challenger xN=A Challenger ChallengerxN=B Challenger Challenger xN=B Challenger Challenger

843S. Ashworth, K.W. Shotts / Journal of Public Economics 94 (2010) 8388471

1 + 1q1I1NI + q1I

1b

1

1 + 1q1I1NI + q1I1

which holds because N1/2. Two aspects of this posterior ranking are critical to the structure of

the equilibrium. First, we again have pandering incentives, because(B, B, )b(A, A, ). Second, the partial yes-man nature of thenewspaper's strategy implies that 0=(B, A, )=(A, B, ). Thenewspaper only disagrees with the incumbent's policy choice if itreceives a precise signal that the incumbent chose the wrong policy.So if xNxI the voter knows that xI; because a high qualityincumbent never chooses xI, the voter thus knows for sure thatthe incumbent is low quality.

Fig. 2 illustrates the conditions under which a perfect account-ability equilibrium exists, depending on the parameters of the model.Lemma 3 divides the horizontal axis into several regions forchallenger quality C; in each region, the voter uses a differentelection rule in a perfect accountability equilibrium. We next derivethose rules and determinewhether the incumbent will actually followhis signal given those rules.

3.5. Reelection and policy choice

Lemma 3 gives us four cases to consider. If C(B, B, ), then theincumbent wins unless either uncertainty resolves and his choice waswrong, or the newspaper announces that he chose the wrong policy.These two events can happen only if he actually chose the wrongpolicy, because a partial yes-man newspaper only criticizes theincumbent if its signal denitively reveals he was wrong. Theseelectoral incentives which are summarized in Table 3, which showsthe voter's action as a function of the incumbent's policy choice (xI=Aor xI=B) and the voter's best information about the optimality of thischoice at the time of the election (=A, =B, xN=A, or xN=B).Given this voter behavior, the incumbent strictly prefers to match theFig. 2. With media commentary.policy to his signal, xI=sI. Thus there is a perfect accountabilityequilibrium in region 1 of Fig. 2.

If C((A, A, ), (A, xN, A)), the incumbent wins only ifuncertainty resolves and his choice was correct: xI=sV . Because hissignal is informative, this gives him a strict incentive to match hissignal. And if CN(A, xN, A) the incumbent can never win, so heprefers to follow his signal. These electoral incentives, which aresummarized in Table 4, imply that there is a perfect accountabilityequilibrium in region 3 of Fig. 2.

In contrast, if C((B, B, ), (A, A, )) then there are nontrivialpandering incentives, as summarized in Table 5, which corresponds toregion 2 of Fig. 2. If the incumbent plays A hewins reelection unless heis proved wrong, either by the public signal revealing a state differentthan the policy or by a media contradiction. If he plays B, in contrast,he only wins if he is proved correct by the public signalmediacommentary in support of his policy is insufcient to convince thevoters to reelect him.

The relevant incentive constraint for pandering in this case is easyto derive.

Lemma 4. If C((B, B, ), (A, A, )), the incumbent chooses A if andonly if

+ v I 2Pr = A jsI; I 1 + 1 v I Pr xN = A jsI ; I 0:

As in the baseline model, there is no incentive problem for anincumbent who sees signal A. Such an incumbent has a posteriorPr(=A|sI=A,I )N1/2, and the incentive constraint is satised.

The case of signal sI=B is more interesting. For either type ofincumbent, the rst term in the inequality in Lemma 4 gives theincumbent an incentive to followhis signal because Pr(=A|sI ,I)b1/2.The second term, however, represents an incentive to pander. Our nextresult shows that it is pandering by low, rather than high, qualityincumbents, which is most difcult to deter.

Lemma 5. Given the electoral incentives in Table 5, for any parametersat which a low quality incumbent wants to follow a B signal, a highquality incumbent also wants to follow a B signal.

As in the baseline model, whether a low quality incumbent willfollow a signal sI=B depends on the probability of uncertaintyresolution. Dene the notation

BPr = A jsI = B; I = L

and

Pr xN = A jsI = B; I = L = B + 1B 1N :

when the incumbent is slightly ahead, but can destroy the symmetrywhen the incumbent is moderately behind.

844 S. Ashworth, K.W. Shotts / Journal of Public Economics 94 (2010) 838847Then Lemma 4, implies that the incumbent follows his signal if andonly if

+ v L 2B1 + 1 v L 0

or

2B1 + vL2vL1BvL1

: 3

In summary, we have derived the following set of results:

Proposition 2. In the model with media commentary, there is a perfectaccountability equilibrium if one of the following conditions holds:

1. Cb (B, B, ), i.e., challenger reputation is worse than an incumbentwho chooses xI=B and is supported by the newspaper, whichannounces xN=B.

2. C(A, A, ), i.e., challenger reputation better than an incumbentwho chooses xI=A and is criticized by the newspaper, whichannounces xN=A.

3. N, i.e., uncertainty resolution is likely.

3.6. Assessing the Jeffersonian intuition

We can see the impact of the media on the existence of a perfectaccountability equilibrium by comparing the conditions in Proposi-tions 1 and 2. In each case, the set of parameter values allowing aperfect accountability equilibrium is characterized by (i) an interval ofchallenger reputations in which there are pandering incentives, and(ii) a threshold level for the probability of the public signal requiredfor incumbents to not succumb to these incentives.

Regarding the threshold probability for the public signal, ourresults are unambiguous: media commentary makes incumbents lesslikely to pander in the event that there is some potential gain fromdoing so. This follows from direct comparison of Eqs. (1) and (3),showing that N, i.e., commentary reduces the cutoff for theprobability of resolution necessary to ensure a perfect accountabilityequilibrium. The intuition is as follows: whereas in the baselinemodelthe incumbent could attain the electoral benets of pandering simplyby choosing policy A, in the model with the newspaper the electoralbenets of pandering are less certain because they also depend, inpart, on the newspaper's actions.

Our results are less clear-cut, however, regarding the interval ofchallenger reputations for which there are pandering incentives. Inparticular, the interval where A and B are treated asymmetrically isshifted in the presence of media commentary.

Table 5Election winner when C((B, B, ), (A, A, )).

xI=A xI=B

=A Incumbent Challenger=B Challenger IncumbentxN=A Incumbent ChallengerxN=B Challenger Challenger B; b B; B; and A; b A; A; :

So, although pandering incentives are reduced for some incum-bents who face relatively weak challengers (because (B, )b(B, B,)) they are also increased for some incumbents who face strongchallengers (because (A, )b(A, A, )). Intuitively, the key toeliminating pandering is to get the voter to apply the same burden ofproof after each policy choice the incumbent might make. Anewspaper that is informative about the state helps bring this aboutWe now briey note two additional implications of our model.First, comparing the equilibriumwith and the equilibriumwithout thenewspaper can be interpreted as a simple comparative static exerciseon media quality: the model with no newspaper corresponds to anewspaper whose signals are random noise independent of thestate.10 Given that, it's worth asking whether improvements innewspaper accuracy can induce pandering more generally. Theanswer is yes. Improvements in newspaper accuracy both reduce and shift the s to the right. Thus for C just greater than (A, A, ) and sufciently low, increasing the newspaper's signal accuracy candestroy a perfect accountability equilibrium.

Second, if the media is sufciently likely to get a fully revealingsignal, then pandering does not happen unless the challenger is aheadof the incumbent, CNI.11 Work on selection effects of repeatedelections imply that this condition is unlikely to be satisedif theincumbent won ofce at least partly on quality grounds in the rstplace, his quality is likely greater than that of a fresh draw from thepool of candidates (Ashworth and Bueno De Mesquita, 2008). Thissuggests that a partial yes-man newspaper will do an effective job ofeliminating pandering, so long as its signal is likely to be accurate.

4. Extensions

We now extend the model in two different directions. First, weanalyze what happens if the media, rather than being motivated by adesire to make the correct announcement, is instead motivated by itsreputation for quality among consumers. Second, we analyze the caseof a media that always reports honestly whether its signal indicatesthat the incumbent's policy choice was correct. The rst extensionrepresents a more pessimistic and cynical view of the media, whereasthe second addresses what would seem to be a best case scenario formedia incentives.

4.1. Reputational motivations

To analyze reputational motivations, we build on the model ofmedia bias from Gentzkow and Shapiro (2006). Doing so requiressome reinterpretation of our model. In particular, assume that thenewspaper can be either high or low quality. A high qualitynewspaper receives perfect signals, whereas a low-quality onereceives signals that are correct with conditional probability q. Theprobability that the newspaper is high quality is N. The newspaperwants to appear high-quality in the long run, i.e., it care about theexpectation of citizenconsumers' beliefs about their quality once thetrue value of becomes publicly known. Let (xI, xN, ) beconsumers' beliefs about the probability that the newspaper is highquality given the incumbent's action, the newspaper's announcement,and the true state of the world.

Although a full analysis of equilibria for all parameters is beyondthe scope of this paper, it is straightforward to show that if thenewspaper is likely to be high quality, i.e., N is high, then there existsan equilibrium in the extension with a reputation-motivated news-paper inwhich behavior is identical to that derived in ourmainmodel.

To see why, start by assuming that a high-quality newspaper isalways truthful whereas a low-quality newspaper chooses xN=xI.Bayes's Rule implies that consumers' posterior beliefs are as follows:

A;A;A = B;B;B = N A;A;B = B;B;A = 0 A;B;B = B;A;A = 1:

4

10 We thank an anonymous referee for suggesting this interpretation.11 To see this, observe that if N is large enough, we have (B, B, )NI .

Lemma 6. With truthful newspaper commentary,

0 = A; xN ;B = B; xN ;A b B;A; b A;B; b B;B; b A;A; b A; xN ;A = B; xN ;B b1:

The lemma follows straightforwardly from Bayes's Rule. The keyinequalities generating pandering incentives are (B, A, )b(A, B, )and (B, B, )b(A, A, ). Both have a simple intuition. Suppose thenewspaper disagreeswith the incumbent's policy choice and the voterdoes not learn the true state of the world. The observations (xI=A,xN=B) and (xI=B, xN=A), have the same likelihoods, so the voter'sprior belief that A is the more likely state of the world implies that hehas more condence that the incumbent was correct if he chose A.Similarly, if the newspaper's announcement agrees with the incum-bent's choice, the incumbent is more likely to be high quality if thispolicy agrees with the prior, so (B, B, )b(A, A, ).

845S. Ashworth, K.W. Shotts / Journal of Public Economics 94 (2010) 838847Off the equilibrium path, we use beliefs such that if the incumbentchooses an action, the newspaper criticizes it, and this criticism turnsout to be mistaken, the consumer believes that the newspaper is lowquality.

A;B;A = B;A;B = 0:

The low quality newspaper's decision to herd or not to herddepends on these consumer beliefs as well as on the newspaper's ownbelief about the probability that the state of the world is=A. For thelow quality newspaper, let N (xI, sN, )=Pr(=A|xI, sN, ). Note thatthis belief depends on , the probability that a low quality incumbentpanders after seeing a signal of B.

The key condition that needs to hold to ensure newspaper herdingis that when xI=B, and sN=A the newspaper announces B. With areputation-motivated newspaper, the relevant condition is that

NB;A;B;A;A + 1NB;A;B;A;BNB;A;B;B;A+ 1NB;A;B;B;B;

which simplies to

NNB;A;

1NB;A; : 5

Eq. (5) is most difcult to satisfy when =0, so we can derive asufcient condition on N for there to exist an equilibrium in areputational model with behavior that is identical to behavior that wecharacterized in our main model. Specically

NNB;A;0

1NB;A;0 : 6

The intuition behind this condition is straightforward. When thenewspaper is likely to be high quality and the high quality newspaperfollows its signals, then for the low quality newspaper the key is tomake the right announcement. Even though, as noted in Eq. (4)consumers will update more positively about the newspaper'sstatement if it chooses xNxI and is proved correct than if it choosesxN=xI and is proved correct (i.e., they update to 1 rather than to N), itis nonetheless the case that for N sufciently high, the newspapermaximizes its expected reputation by maximizing its probability ofmaking the correct announcement.

4.2. Truthful media

We now briey sketch the case of a media that truthfully reportswhether its signal indicates that the incumbent's policy decision wascorrect, i.e., the newspaper is nonstrategic or it is duty-bound toannounce what policy its signal indicates was better, without trying tolearn from the incumbent's policy choice. This case is analytically thesame as a media that makes an announcement at the same time thatthe incumbent chooses policy.

As in our main model, (h) is the voter's assessment of theprobability that the incumbent is high quality, given history h. Withthe newspaper's announcement, h is a triple (xI, xN, sV); we use (xI, xN,sV) to denote the voters' posterior given rst period policy xI,newspaper announcement xN, and public signal sV (always underthe conjecture that there is a perfect accountability equilibrium, soxI=sI). Lemma 1 makes it straightforward to calculate these poster-iors for every history that the voter might observe. For our purposes,

the key aspect of these posteriors is their ordinal ranking.In such circumstances, the incumbent is willing to pander unlessthe probability of uncertainty resolution is sufciently high. Specif-ically, using the notation Pr(sN=A|sI=B, I=L), the cutoff toensure that a perfect accountability equilibrium exists is

2B1 + vL2vL1BvL1

: 7

Fig. 3 illustrates conditions under which a perfect accountabilityequilibrium exists, depending on the parameters of the model.

We now turn to the question of how the newspaper's truthfulnessaffects incumbent policy choice. It is natural to conjecture that apartial yes-man newspaper is worse than a newspaper that truthfullyreports its signal. But this conjecture is only partly correct. A simplecomparison of Eqs. (1), (3), and (7) conrms that bb, tting wellwith the intuition that the more honestly the newspaper reports itssignal, the more benecial will be the effects on policy choice.However, a focus solely on the probability of uncertainty resolutionmisses part of the story. Pandering incentives also depend on thedifference in the two candidates' reputations, and, for some values ofC , there is pandering in the simultaneous model but not in thesequential model. Specically, the set of challenger qualities for whichpandering can occur with a truthful media i.e., ((B, A, ), (A, B,))((B, B, ), (A, A, )), is neither a strict superset nor a strictsubset of ((B, B, ), (A, A, )), the set of challenger qualities, fromProposition 2, for which pandering can occur in our mainmodel. Thus,at least for some parameter values, a truthful newspaper actuallymakes things worse than a nonstrategic newspaper that truthfullyreports xN=sN.

For some intuition, recall that pandering incentives arise when thevoter applies different burdens of proof to the incumbent dependingon which policy he chose. With a truthful media, an incumbent with amoderate lead over the challenger may pander because he gets theFig. 3. With truthful media.

benet of the doubt when he chooses the popular action and thecommentator disagrees. A partial yes-man newspaper, on the otherhand, only contradicts the incumbent if its signal perfectly revealsthat the incumbent's choice was wrong. Thus the voter no longer

846 S. Ashworth, K.W. Shotts / Journal of Public Economics 94 (2010) 838847gives the popular action an easier burden of proof, and pandering iseliminated.12

5. Discussion

Previous models of accountability suggest that a policy maker'sincentives to pander to public opinion can be reduced by either hidinginformation about the incumbent's policy choice (Prat, 2005) or byintroducing a neutral media commentator who makes informativeannouncements about whether the incumbent's actions truly pro-moted voters' interests. We have shown that the actual effect ofintroducing a commentator is far more subtle than this intuitionsuggests. The real key to eliminating pandering is to induce the voterto apply the same burden of proof to the incumbent after eachpossible action that he may take, and we nd that although thepresence of the media reduces pandering incentives for incumbentswho faceweak challengers it may increase pandering incentiveswhenthe challenger is strong.

It is likely that this mechanism is also a work in the other family ofpandering models, in which the voter's uncertainty is not over theincumbent's competence but rather over his policy preferences (e.g.,Maskin and Tirole, 2004). In those models, like the one we consider,the incumbent can have an incentive to take the ex ante popular actionwhen the voter will respond asymmetrically to different policies.Although proof will have to await future research, it seems likely thatinformative media commentary has the potential to induce suchasymmetric treatment, and thus pandering, in those models as well.

A particularly interesting implication of the model is that yes-manbehavior by the media, while denying information to the voter, canactually provide better incentives for the incumbent to choose thecorrect policy, compared to a media that always reports truthfullywhether its signal indicates that the incumbent's policy choice iscorrect. This result is surprising, because one might think that mediadeference to the incumbent would inevitably impede the process ofaccountability. Instead, deference in the face of low-precision mediainformation can be necessary for incumbents to take correct actionsonly if the commentator herds can the incumbent be safe from thefear that a mistaken critique of an already unpopular policy will leadhis to be dismissed for taking the right action.

However surprising this result may be, the intuition behind it isnot completely new. In fact, Jefferson made a similar argument, aquarter of a century after his famous letter to Carrington. We closewith a quotation from a 1811 letter to Colonel William Duane, editorof the Philadelphia Aurora, in which Jefferson argued that journalistsshould be aware of their own limitations and hesitant to rush tojudgment of elected ofcials:

I think an Editor should be independent, that is, of personalinuence, and not be moved from his opinions on the mereauthority of any individual. But, with respect to the generalopinion of the political section with which he habitually accords,his duty seems very like that of a member of Congress. Some ofthese indeed think that independence requires them to followalways their own opinion, without respect for that of others. Thishas never been my opinion, nor my practice, when I have been ofthat or any other body. Differing, on a particular question, fromthose whom I knew to be of the same political principles with

12 An informal statement is that, with a strategic newspaper, disagreement with theincumbent is more informative than in the case of the truthful newspaper. Thisshould not be confused with the formal notion of informativenessthe two

information structures are not ordered by Blackwell's garbling criterion.myself, and with whom I generally thought and acted, aconsciousness of the fallibility of the human mind, and of myown in particular, with a respect for the accumulated judgment ofmy friends, has induced me to suspect erroneous impressions inmyself, to suppose my own opinion wrong, and to act with themon theirs...As far as my good will may go, for I can no longer act, Ishall adhere to my government executive and legislative, and, aslong as they are republican, I shall go with their measures,whether I think them right or wrong; because I know they arehonest, and are wiser and better informed than I am.

6. Proofs

Proof of Lemma 1. Consider the following learning process. Thevoter rst learns h, and then learns the true state. (The second stepmay or may not be redundant). The voter's belief at the time of theelection corresponds to the intermediate stage of this process.

At the nal stage of the process, the voter's belief about incumbentquality is either 0 or +. Since a probability of an event is just theexpected value of an indicator function, the martingale property ofBayesian updating implies (h)=Pr(I=H|h)=E(Pr(I=H|h,)),where the expectation is with respect to the realization of the nalstage of learning. But the expectation is just Pr(=A|h)Pr(I=H|h,=A)+Pr(=B|h)Pr(I=H|h,=B), which gives the result.

Proof of Lemma 4. The incumbent wins if he is proved correct or ifhe chooses A and the newspaper agrees. Then choosing A gives payoff

+ v I Pr = A jsI ; I + 1 v I Pr sN = A jsI; I ;

while choosing B gives payoff

+ v I 1Pr = A jsI ; I :

Thus the incumbent chooses A if and only if

+ v 2Pr = A jsI; I 1 + 1 v I Pr sN = A jsI ; I 0: 8

Proof of Lemma 5. Based on Eq. (8), the difference between a lowtype's gain from choosing A rather than B, and the high type's gain is

= + L 2B1 + 1 L B 1B 1N + + H 1 H 1N :

Substituting in for v (L) and v (H) from Eq. (2) yields

= + 1 + q 2B1 + 1 1 + q B + 1B 1N + 1 + 1 1 + 1N :

9

At =0, this reduces to

2B1 + 1 + q B + 1B 1N + 1 + 1N :

We claim this expression is positive. Since B+(1B)(1N)N(1N) it sufces to show

2B1 + 1 + q 1N + 1 + 1N N 02B N 1q 1N

2 1q

1q + 1 q N 1q 1N

2 N 1q 1N + 1 q 1N :

This last expression holds because N (1q)(1N) and N1/2N(1) q (1 ).N

Having established that (0)N0, we now differentiate Eq. (9) withrespect to to get

= 1 + q 2B1 1 + q B + 1B 1N + 1 + + 1 + 1N

= 2B1 B + 1B 1N + 2N+ q 2B1 q B + 1B 1N + 2N

= 2N B + 2N 1q 1B

10

Because 2NN0, 1B(0,1), and q(0,1) Eq. (10) is strictlygreater than zero. Because the difference is positive at =0 andincreasing everywhere is sufcient to ensure that if the low type'sincentive constraint is satised, the high type's incentive constraint isalso satised.

Acknowledgments

We thank the seminar audiences at Berkeley, Duke, Harvard/MIT,Maryland, Montreal, Princeton, Stanford GSB, and Yale for usefulcomments.

References

Arnold, R. Douglas, 2004. Congress, the Press, and Political Accountability. PrincetonUniversity Press, Princeton, NJ.

Ashworth, Scott, Bueno de Mesquita, Ethan, 2008. Electoral selection, strategicchallenger entry, and the incumbency advantage. Journal of Politics 70 (4),10061025.

Baron, David P., 2006. Persistent media bias. Journal of Public Economics 90 (12), 136.Bernhardt, Dan, Krasa, Stefan, Polborn, Mattias, 2006. Political polarization and the

electoral effects of media bias. Journal of Public Economics 92 (56), 10921104.Besley, Timothy, Prat, Andrea, 2006. Handcuffs for the grabbing hand? The role of the

media in political accountability. American Economic Review 96 (3), 720736.Bikhchandani, S., Hirshleifer, D., Welch, I., 1992. A theory of fads, fashion, custom, and

cultural change as informational cascades. Journal of Political Economy 100,9921026.

Canes-Wrone, Brandice, Herron, Michael C., Shotts, Kenneth W., 2001. Leadership andpandering: a theory of executive policymaking. American Journal of PoliticalScience 45 (3), 532550.

Efnger, Matthias R., Polborn, Mattias K., 2001. Herding and anti-herding: a model ofreputational differentiation. European Economic Review 45 (3), 385403.

Egorov, Georgy, Sergei Guriev, and Konstantin Sonin. 2007. Media freedom,bureaucratic incentives, and the resource curse. New Economic School Typescript.

Gentzkow,Matthew, Shapiro, Jesse, 2006. Media bias and reputation. Journal of PoliticalEconomy 114 (2), 280316.

Gentzkow, Matthew, Shapiro, Jesse M., 2007. What Drives Media Slant? Evidence fromU.S. Daily Newspapers. University of Chicago Typescript.

Groseclose, Tim, Milyo, Jeffrey, 2005. A measure of media bias. Quarterly Journal ofEconomics 120 (4), 11911237.

Ho, Daniel E., Quinn, Kevin M., 2008. Measuring explicit political positions of media.Quarterly Journal of Political Science 3 (4), 353377.

Maskin, Eric, Tirole, Jean, 2004. The politician and the judge: accountability ingovernment. American Economic Review 94 (4), 10341054.

Page, Benjamin I., 1996. Who Deliberates? The University of Chicago Press, Chicago.Prat, Andrea, 2005. The wrong kind of transparency. American Economic Review 95 (3),

862877.Prendergast, Canice, 1993. A theory of yes men. American Economic Review 83 (4),

757770.Scharfstein, D., Stein, J., 1990. Herd behavior and investment. American Economic

Review 80, 465479.Swank, Otto H., Visser, Bauke, 2006. Do elections lead to informed public decisions?

Public Choice 129, 435460.

847S. Ashworth, K.W. Shotts / Journal of Public Economics 94 (2010) 838847

Does informative media commentary reduce politicians' incentives to pander?The modelBaseline modelPolicies and preferencesInformation structure

Adding the mediaEquilibrium concept

The baseline modelMedia commentarySecond period policy and continuation valuesSimplifying voter beliefsNewspaper behaviorVoter updatingReelection and policy choiceAssessing the Jeffersonian intuition

ExtensionsReputational motivationsTruthful media

DiscussionProofsAcknowledgmentsReferences