Weingast B.R., Wittman D. The Oxford Handbook of Political Economy

Подождите немного. Документ загружается.

32 voters, candidates, and parties

generally. Emerging to the response to these four problems, then, is a more compre-

hensive theory of politics.

1 Modeling Performance and Policy

.............................................................................

An early critique of the spatial model came from social psychologist Donald Stokes

(1963). Stokes argued that ideological issues matter relatively little to people com-

pared to assessments of the attributes of the candidates or the performance of the

incumbent administration. All voters may want more economic growth or more

spending on local projects. They may want lower crime and traﬃcsafety.Allmay

desire international peace, democracy, and national unity. All other things being

equal, voters may want someone whom they know and trust, who is honest, or who

is more competent in the administration of government. These are examples of what

Stokes termed valence issues—matters on which voters agree on the general objective

on which they judge, not the candidates’ promises or policies but their personal

attributes or performance in oﬃce.

Evidence of such factors abounds in empirical research on elections and the

economy and on the psychology of voting. Voters reward incumbent presidents for

good times and punish them for bad times.

Another sort of valence advantage is a

politician’s personal vote. Incumbent politicians in the United States have signiﬁcant

personal votes derived from the things they have done for constituents, their higher

levels of name recognition, and simply the selection process.

Also, the governing

party in a parliamentary system may expect to have an electoral advantage in the

general election. The sitting prime minister calls elections, and elections are called

strategically so that the circumstances of the election are most favorable to the sitting

government.

Valence issues are easily incorporated into the simple spatial model. For simplicity

of exposition, I will assume that the valence term captures the competence of the

candidate or party, which in turn translates into better economic performance. A

candidate’s competence is a ﬁxed attribute; it is not something the candidate can

maximize or alter during a campaign. A positive valence means that this candidate

has better than average competence in governing. Voters judge candidates not only

on the basis of their policy promises, but also on the basis of their competence. If

the candidates promise identical policies, then all voters will choose the candidate

with the higher valence term. The diﬀerential between candidates’ valence attributes

is termed their valence advantage (or deﬁcit).

⁴ See Hibbs (this volume) for an extensive discussion.

⁵ See Levitt and Wolfram 1997 and Ansolabehere and Snyder 2002 for evidence on the extent of

personal voting and incumbency advantages. Cain, Ferejohn, and Fiorina 1987 oﬀer an extensive

examination of the sources of the personal vote in the United States and United Kingdom. Jacobson 1997

documents that diﬀerential candidate experience matters in elections.

⁶ Baron 1998 develops a general model of parliamentary governments.

stephen ansolabehere 33

The addition of the valence term alters voter decision-making and the logic of

spatial competition. Speciﬁcally, some voters will choose a more competent candi-

date, even if the other candidate is ideologically closer to those voters’ ideal points.

A simple formulation helps make clear exactly how this more complicated model

works. Let z be the ideal point of a voter along a spatial dimension Z. Suppose that

there are two candidates, A and B. Let candidate A’s and candidate B’s positions along

Zbethepointsa and b, respectively. Let v be the net valence diﬀerence between

candidates. The net utility that voter z receives from candidate A versus candidate B

is: U=v − (z − a)

+(z − b)

. The voter chooses A if Uisbiggerthan0 and she

chooses B if Uislessthan0.

Introducing the valence issue immediately aﬀects the electoral outcome. As with

the simple spatial calculus, one can determine the voter who is indiﬀerent between the

parties and characterize that voter’s ideal policy. That voter’s ideal policy is referred

to as a “cut-point” along the dimension Z. Voters with ideal policies to the left of the

cut-point vote for party A and those with ideal policies to the right of the cut-point

vote for party B. The cut-point is the value of z such that U = 0. Speciﬁcally, if the

candidates are diverged:

c =[(a + b)/2] + [v/(2(a −b))].

The cut-point determines the share of votes won by the candidates. The percentage

of the vote won by the left candidate is the percentage of the electorate with ideal

points to the left of c.

Several empirical implications immediately follow.

First, the median voter is pivotal only in the special case where v equals 0.When

voters ignore valence issues, the cut-point equals the midpoint between the candi-

dates’ positions. However, when voters value valence issues, the cut-point is no longer

the midpoint but shifts to the left or right depending on which candidate enjoys the

valence advantage. And, when the candidates are converged, all voters choose the

advantaged candidate.

Second, the cut-point need not lie between the two candidates. If the valence

advantage is large relative to the diﬀerence between the candidates, then the ideal

point of the voter who is indiﬀerent between the candidates may lie outside the

interval between the two candidates’ positions. Empirical researchers commonly

assume that the voter who is indiﬀerent between the candidates lies between the

candidates. Typically, the midpoint is used to measure the cut-point. For example,

analyses of surveys such as the National Election Study usually take as given that

the voter indiﬀerent between the candidates lies toward the “center” (Brady and

Sniderman 1985; Rabinowitz and MacDonald 1989). Also, studies of elections and

roll call voting use the midpoint between the candidates as measures of the ideal

point of the indiﬀerent voter (Platt, Poole, and Rosenthal 1992). This assumption

is acceptable if voters ignore valence issues. However, when valence issues or can-

didate attributes take on importance, such as incumbency advantages or economic

performance, this assumption is no longer valid. To estimate the position of the

34 voters, candidates, and parties

indiﬀerent voter one must measure the valence issues as well as the parties’ ideological

positions.

Third, the electoral eﬀect of a candidate’s personal vote, especially the incumbency

advantage, ought to depend on the ideological positions of the candidates. In analysis

of legislative elections, the personal vote is the valence advantage that a candidate

enjoys because he or she has the attributes sought by the constituency. Once in oﬃce,

the things a politician does, such as constituent service, might add to that personal

vote. The formula for the cut-point suggests that the positions of the candidates

or parties should aﬀect the magnitude of a candidate’s personal vote. Suppose for

instance that candidates are drawn from the same pool of talent over time and

that all candidates run under their parties’ ideological labels, but that the distance

between the parties changes over time. When the parties are far apart, changes in

the personal vote and incumbency advantage will alter a candidate’s vote share only

a small amount. However, when the parties are close together the personal vote will

haveamuchlargeraﬀect on election results. The eﬀect of incumbency on votes, then,

ought to depend on the positions of the candidates and parties.

Fourth, the cut-point analysis suggests a particular speciﬁcation for the analysis of

aggregate election results: regress vote shares on the midpoint between the parties and

indicators of performance divided by the distance between the parties. This simple

insight has gone unnoticed in the extensive literatures on national, legislative, and

state elections.

The literature on economic voting and election forecasts is a case in point. An

extensive literature uses economic indicators such as income growth or unemploy-

ment to predict elections. The coeﬃcients from a regression are used to construct

an equation from which one may forecast the vote shares of the parties under the

circumstances that may hold in a future election (see for example Fair 1996). Such

data analyses do not include ideology; hence, voters either do not value ideology or

ideology independently aﬀects the vote. The cut-point analysis implies that the latter

is deﬁnitely untrue. Changes in economic performance are captured by the valence

term v.Changesinv aﬀect the cut-point, but the magnitude of the eﬀect depends on

the distance between the parties, i.e. a − b. When the parties are quite distinct, the

eﬀect of a change in the economy on the vote will be muted. But, when the parties

are nearly converged, slight changes in economic performance will have substantial

eﬀects on the vote. The speciﬁcation that emerges from the cut-point analysis is that

the presidential vote won by the incumbent party should be regressed on estimates

of the midpoint between the candidates and the economic indicators divided by the

distance between the presidential candidates in a given election. The positions of the

candidates can be derived from survey data.

It is worth noting that the comparative statics for the cut-point do not mean that

ideology “matters” more or less in the psychology of individual voters. The utility

function of the voter does not change. Rather, the positioning of the candidates aﬀects

on what basis voters can distinguish the candidates. Analyses of survey results do not

measure the actual structure of voter preferences but revealed preferences, given the

positions of the candidates. I direct this comment primarily at those using survey

stephen ansolabehere 35

data to “test” the spatial theory. A number of odd results in the literature, such

as directional voting (Rabinowitz and MacDonald 1989) or asymmetries in spatial

perceptions (Brady and Sniderman 1985), may result from the assumed positions of

the candidates in survey research analyses or the assumption that the indiﬀerent voter

is the one midway between the candidates.

Stokes’s assertion that valence issues matter, thus, seems correct and correctable.

Inclusion of a valence term can capture assessments of candidate competence, in-

tegrity, and performance. This simple generalization of voter preferences, however,

surely changes the incentives facing candidates and parties, and, therefore, what

positions candidates ought to take in equilibrium. The question of candidate posi-

tioning represents an even more signiﬁcant empirical challenge to the spatial theory

of elections, and one to which I now turn.

2 Why Do Candidates Diverge?

.............................................................................

The median voter theorem is perhaps the single most important theoretical result in

modern political science. It is simple and elegant; it is widely applied to the study of

elections and government; and, by most accounts, it is false.

In a two-candidate election or two-party system, the simple spatial model predicts

that the competing candidates will promise the same policy—that preferred most by

the median voter. The settings most likely to approximate the assumptions of this

theory are the US Congress and presidency and the UK Parliament. Extensive studies

of British and American elections ﬁnd that the parties in fact take divergent positions,

and neither party represents the median voter’s ideal policy.

Political economists take the median voter theorem the same way many biologists

view the Hardy–Weinberg Law. The strong empirical prediction is almost surely

wrong and, as a result, one of the basic assumptions is incorrect. The challenge is

to ﬁnd the factor or factors that explain divergence.

Many explanations have been oﬀered—including primary elections, incumbency

advantages, political parties, uncertainty and risk aversion, interest group contribu-

tions, candidates’ personal preferences, election laws, and potential third parties. I

will discuss four of these explanations here.

⁷ An important literature in political science examines the policy positions of the parties and their

candidates in the US House of Representatives. One line examines the roll call votes of candidates who

have served in the House and run against each other, as happens when one candidate beats an

incumbent (Fiorina 1974; Ansolabehere, Snyder, and Stewart 2001). The second line examines surveys of

candidates for oﬃce (Erikson and Wright 1997; Ansolabehere, Snyder, and Stewart 2001). While both

face sample selection problems, they show very similar patterns. There is a large gap between the average

Democratic candidate and the average Republican candidate overall, and the candidates competing with

each other within a given district almost always take diﬀerent positions, with the Democrat on the left

and the Republican on the right.

36 voters, candidates, and parties

Valence issues. Valence issues yield a median voter result with divergence. When

one candidate or party expects to have a valence advantage the candidates will stake

out divergent positions along the policy dimension. Ansolabehere and Snyder (2000)

and Wittman (2005) provide general conditions under which equilibria will exist and

the characteristics of equilibria. In the simplest case, two candidates seek to win oﬃce

(rather than vote maximization); one has a ﬁxed advantage that all voters consume;

and voters’ utilities are as described in the previous section. The candidate with the

valence advantage locates at the median voter and wins with certainty. The other

candidate can locate anywhere. If voters have random utilities and candidates maxi-

mize the probability of winning, the advantaged candidate will locate at the median

and the disadvantaged candidate will locate some distance away from the median in

order to maximize his chance of winning. If the disadvantaged candidate is too close

to the advantaged candidate voters will only be able to distinguish the candidates

on the basis of the valence advantage and the advantaged candidate will have very

high probability of winning. If the disadvantaged candidate locates very far from the

median, he will attract few votes.

Contestation. Competition in the spatial model is limited to position-taking by a

ﬁxed number of candidates (two). The possibility of entry changes the logic, and leads

to divergence. If entry is possible, a third party or candidate may wish to contest the

election. Suppose the two parties are located at the median, as predicted by the simple

theory. If the third party locates at a position a small distance from the median, the

third party will win the largest share of the votes. Suppose the third party locates

slightly to the left of the median. That party will win all votes to the left of that

position—nearly half of the electorate—while the two incumbent parties will split the

vote to the right. As a result the third party would win, and clearly has an incentive to

enter.

Palfrey (1984) shows that the possibility of entry leads to an equilibrium between

the two parties in which neither is at the median and the third party has no incentive

to enter. His argument requires a reﬁnement of the Nash equilibrium, as one must

consider subgame perfect strategies. One equilibrium is for the incumbent parties to

locate at the fortieth percentile and sixtieth percentile points. They will split the vote

evenly.Athirdpartyhasnoincentivetoenterontherightorontheleft,asdoing

so will only make the party furthest away a certain winner, and the third party has

no chance of winning. Neither party will want to move toward the middle. Doing so

will prompt a third party to form slightly to the outside of that party. The third party

would then win the most votes. By taking symmetric positions about the median,

the two incumbent parties can therefore eﬀectively box out entrants and will retain a

ﬁfty–ﬁfty chance of winning.

The calculus of entry will diﬀer with alternative electoral institutions. In a propor-

tional representation system, for instance, a third party would be guaranteed some

representation if it entered. It may then become part of a coalition government. Cox

(1996) considers a wide range of electoral institutions and the eﬀects that they have

on the number of parties or candidates that will contest. Electoral institutions, Cox

argues, determine the number of candidates and the positions they will take. Only

stephen ansolabehere 37

the plurality rule single-member district systems create an incentive for two-party or

two-candidate elections.

Candidate preferences. Theresolutiontotheparadoxesofrationality,Ihavesug-

gested, is to treat voting as consumption. People like to vote, and they are motivated

to choose the candidate who is in their judgement the best alternative rather than to

maximize the returns to voting. It is reasonable to project the same motivations to

politicians. Wittman (1983) shows that if candidates have policy preferences and seek

oﬃce to get their preferences implemented, then competing candidates will oﬀer

divergent positions. See also Besley and Coate’s (1997) citizen-candidate model.

The labor market for politicians may sustain the expression of candidate prefer-

ences in electoral competition. The labor market for politics is hierarchical—those in

higher positions typically started their political careers in local oﬃces, such as school

committees. Local posts are usually part-time or volunteer jobs. The appeal of such

posts is the ability to make a diﬀerence in the community, rather than the pay. Those

who get involved in local government, then, are motivated at ﬁrst by ideological or

“consumption” beneﬁts, rather than by the value of oﬃce.

At higher levels of government most elected oﬃcials (federal and state legislators

and statewide oﬃcers) are attorneys and businessmen. They are not compensated

well in most democracies compared with their professional opportunities. In order to

sustain such a pool of high-quality candidates, it may be the case that these politicians

need non-monetary compensation in order to stay involved in politics. Voters may be

willing to tolerate some extremism in order to get more professional people in oﬃce.

One objection to this account is that since anyone can run, a citizen near the

median will run for oﬃce and win. A centrist has a clear incentive to enter the political

arena if the sitting politician is an extremist: the centrist will surely win. Even if the

median voter does not run, successive elections will choose candidates increasingly

close to the median. So the median is the limit to a sequence of elections.

Ideological candidates may be sustained through the candidate selection process.

In most democracies candidates are selected through party organizations or legisla-

tive caucuses. Those organizations may wish to select candidates consistent with a

particular ideology. For example, unions in the United Kingdom have considerable

inﬂuence on the nomination of candidates for Parliament under the Labour Party

label. In the United States, primary elections tend to screen candidates. Primary

electorates are drawn from the liberal and conservative segments of the ideological

spectrum, and moderates have a very diﬃcult time emerging from this process

(Aldrich 1980). Those doing the nominating may not wish to trade oﬀ ideology for

winning, especially if the circumstances of a given election favor the opposing party.

Political parties’ platform choices. Downs assumed that political parties are teams—

groups of individuals working for a common goal. The individual’s utilities within a

party are the same and the same as the party’s. Most of the subsequent research has

maintained this assumption. Cox (1996), for example, uses interchangeably parties

and candidates.

Politicians within a party do not have identical interests. If one assumes that

candidates simply seek to win their own seat, each politician will want the party’s

38 voters, candidates, and parties

platform to be identical to the preference of the median voter in his district. Those

in far-left seats want the party to move to the left so as to guarantee a win in their

districts; those in center-left seats want to pull the party toward the middle in order

to secure their seats.

Collective decision-making within parties in which there are conﬂicting interests

leads to predictions of policy divergence. Snyder (1994)andSnyderandTing(2002)

present models of elections in which parties must ﬁrst make collective decisions to

choose a platform under which all must run. In equilibrium, the parties oﬀer distinct

platforms. Those on the left of the left party and the right of the right party keep their

parties from converging.

Party platform choice, contestation, candidate selection, and valence issues have

emerged as elements of a richer political economy theory of elections. Each of these

elements is suﬃcient to explain the empirical regularity of policy divergence between

candidates or parties. But each captures somewhat diﬀerent electoral forces. A model

with valence issues, for example, predicts that the advantaged candidate will take

a centrist policy position, while the logic of party platform choice predicts that

neither party takes a convergent policy position. None of these models, however,

yields the sharp prediction against which data can be compared. As a result, the

median voter theorem remains an important baseline against which we compare

behavior.

3 Twin Paradoxes

.............................................................................

More troubling than the failure of the median voter theorem, the political economy

theory of elections leads immediately to two paradoxes of rationality. An Economic

Theory of Democracy—and liberalism generally—posits that individuals seek their

own self-interest. However, a theory of democracy based on narrow self-interest,

such as income maximization, leads immediately to two predictions: rational non-

participation and rational ignorance. Citizens should not take the time to vote or

expend the considerable eﬀort to stay abreast of public aﬀairs. In order to have

meaningful elections people must vote and understand the choices they face in or-

der to even have meaningful elections. As such, the theory apparently implies that

democracy will fail.

The argument goes as follows. Self-interested citizens, the theory prescribes, make

electoral choices and take actions, such as voting or reading about politics, to maxi-

mize the expected beneﬁts of a given action net the cost of those activities. A utility-

maximizing citizen will choose a level of political activity, such that the marginal

expected beneﬁt of the activity equals the marginal cost. This is the familiar formula-

tion of utility maximization from consumer theory in microeconomics.

Utility maximization in the electoral setting, though, leads to the conclusion that

people should vote very infrequently. The marginal expected beneﬁt consists of two

stephen ansolabehere 39

components—the beneﬁt received if one’s favored candidate wins less the beneﬁt

lost if the opposition wins and the marginal eﬀect of voting on the election out-

comes. Marginal expected beneﬁts approach zero if there is no diﬀerence between

the candidates or if one’s vote has an imperceptible eﬀect on the election outcome.

If the median voter theorem holds, then the candidates do not diﬀerentiate and

there is no diﬀerential beneﬁt to voting. Even if the candidates take distinct posi-

tions, the marginal eﬀect on the outcome is about nil if the electorate is suﬃciently

large.

This logic implies that only those individuals whose votes make the diﬀerence be-

tween winning and losing will vote. Ledyard (1984) proves that the Nash equilibrium

of a participation game is for voters to use a mixed strategy in which the probability

of voting is approximately zero for most voters, and slightly higher from those near

the median in a spatial model.

Downs further observes that this logic implies that an individual will not devote

much eﬀort to learning about the electoral choices. It is not an optimizing strategy

to devote considerable time and eﬀort to learning the details of the policy choices

presented by competing candidates and parties. It is rational to be politically ignorant.

Compounding this logic further is the incentive to free ride, as described in Mancur

Olson’s Logic of Collective Action. If ideologically similar people are going to vote or

learn about politics or participate in similar ways, why should I? Utility maximization

implies that individuals will give only small amounts of money to political campaigns

and will spend little time attending political meetings or engaging in other forms of

political action.

These two paradoxes are deeply troubling. They raise a fundamental question

about the power of the theory to explain vote choice, as well as turnout. Once the

individual is in the voting booth, what does the theory really predict about how

the individual will behave? If the action taken is to vote and to vote for a speciﬁc

candidate or party, the marginal expected utility of voting for a given candidate is still

exceedingly small. If voters have random utility functions, as is commonly conjec-

tured by empirical researchers working in this tradition, then the random component

surely dwarfs the marginal expected utility calculations of voting for one candidate or

another. It is hard, then, to entertain the notion that citizens are maximizing utility

when they vote.

One response to this logic is to reject the theory. Donald Green and Ian Shapiro

express this view most sharply in The Pathologies of Rational Choice. Reviewing the

empirical literature they argue that the basic empirical predictions regarding turnout

and competition suggest that the economic theory of democracy ought to be rejected

on empirical grounds. They are most troubled by the implications of the theory for

participation. Rational non-participation ought to be common, except in very close

or uncertain elections. However, closeness does not lead to much higher turnout, and

in very uncompetitive races people still vote at high rates—around 50 per cent in US

national elections and higher in other countries.

The paradoxes of rationality, however, apply broadly to democratic theory, not just

to economic reasoning. These problems are common sense, and have been considered

40 voters, candidates, and parties

in one form or another by those working in all of the main intellectual traditions of

political science. Judge Learned Hand (1952, 93), for instance, considered these two

paradoxes in reﬂecting on his own behavior in his essay “Democracy: its presump-

tions and realities” delivered to the Federal Bar Association more than two decades

before Downs published his treatise. Hand wrote:

My vote is one of the most unimportant acts of my life; if I were to acquaint myself with the

matters on which it ought really to depend, if I were to try to get a judgment on which I was

willing to risk aﬀairs of even the smallest moment, I should be doing nothing else, and that

seems a fatuous conclusion to a fatuous undertaking. Because, if all were done, for what after

all does my single voice count among so many? Surely I can play my part better in the society

where I chance to be, if I stick to my last, and leave governing to those who have the temerity

to accept the job.

I ﬁrst came across reference to this essay not in a critique of the spatial model or

political economy, but in the classic book Voting, written by Bernard Berleson, Paul

Lazarsfeld, and William McPhee—three of the leading sociologists of the behavioral

revolution. They had no answer for Hand’s observation, except to describe “political

man” as somewhere between narrowly self-interested and purely other regarding.

That has become the received answer for Learned Hand’s observation among so-

ciologists and pluralist thinkers in political science, but it is not an explanation.

Discarding the political economy approach because it does not explain participation

seems fatuous—no social science model as yet truly explains participation. For his

part, Learned Hand took this paradox of democracy to be a failing of the traditional

theories of democracy.

It is not, then, a failure of the political economy approach in particular, but a

failure of social sciences generally that we do not yet have a coherent theory for why

individuals seeking their own needs will become informed, will vote, and, given that

they do vote, will vote in ways consistent with their interests and beliefs.

A second response to the paradoxes of rational ignorance and non-participation

is to take on the matter directly, beginning with the assumption that people are

self-interested. Generally speaking, the most promising approaches move away from

calculations of citizens’ personal returns or narrow self-interest, and introduce either

the consumption value of voting or mechanisms for coordinating individuals to

participate, especially groups.

Perhaps the ﬁrst attempt to resolve the paradox of non-participation was that

of William Riker and Peter Ordeshook (1968). They add to the basic mathematical

formulation introduced by Downs a constant term that captures a citizen’s sense of

duty to vote. Individuals with suﬃciently high duty will vote, and those with a low

sense of duty will not.

The introduction of citizen duty represents a fundamental shift in the logic away

from the view that people pursue their narrow self-interest, such as the maximization

of income. Citizen duty does the trick. It generates a prediction of positive turnout,

but it is not theoretically satisfying. The assumption oﬀers little guidance for un-

derstanding who will vote and who will not vote, or who will devote themselves to

stephen ansolabehere 41

learning about the issues and candidates and who will not. Citizen duty becomes the

new black box.

In addition, the citizen duty term does not address the issue of what people will do

once in the voting booth. In Riker and Ordeshook’s formulation, those who vote are

the citizens for whom their sense of duty is of suﬃciently high value to oﬀset the cost

of voting. However, if we still maintain that voters maximize their expected beneﬁts,

the marginal beneﬁt of voting for one candidate over another remains inﬁnitesimal.

Voters, then, are nearly tossing coins in the polling booth.

A further reﬁnement on the consumption motivation is “expressive voting.” This

argument holds that people who care more about the outcome in the sense of having

agreaterutilitydiﬀerential in their assessments of the candidates will be more likely

to vote.

This depends on the shapes of utility functions, but some simple implica-

tions may be derived. One simple implication is that individuals who are indiﬀerent

between the candidates (lying between the candidates on a spatial dimension) will

be less likely to vote. There is some empirical support for this claim as self-described

moderates and political independents do vote less than ideological extremists and

strong partisans (see Rosenstone and Hansen 1993, 273).

Most contemporary political economy theorizing tacitly embraces this assump-

tion. Formal models of elections explicitly or implicitly assume that voters merely

choose the ideologically closer candidate or party. If voters care only about the

ideological distance function once in the voting booth then either the loss function is

the motivation for behavior or the reason for voting is independent of one’s political

preferences. It does not matter to the individual voter in such analyses that the

action taken has only an inﬁnitesimal aﬀect on the expected utility. Voters, then, do

not optimize expected utility; they merely consume based on the proximity of the

choices.

Another way out of the paradox of non-participation is to embed individuals

within social groups. A particularly promising example is the work of Coate and

Conlin (2002). They build on John Harsanyi’s notion of rule utilitarianism and

develop the notion of group rule utilitarianism. The argument goes roughly as fol-

lows. Suppose that we all have the same preferences (to pass a certain measure), but

that because of rational non-voting the measure would not pass. A rule utilitarian

solution holds that we would all be better oﬀ if everyone voted than if no one

voted, so the group has an incentive to create a mechanism which induces all to

vote.

Coate and Conlin extend this argument to the situation where there are conﬂicting

interests or preferences at stake in an election. Individuals belong to social groups and

those groups are the foundation for political competition. For example, in a vote over

property taxes in a local community, there are property owners and renters. Among

⁸ For an exposition of this idea, see Schuessler 2000.

⁹ One may take this reasoning further. If the group can exert this degree of coordination, then surely

it would arrive at the rule of requiring just one of the members to vote. If just one person voted the

measure would pass. That member would bear the same cost of voting as when all people voted, but no

one else would bear the cost of voting. Hence, group utility would be higher than if all people voted.