Review: The Quality of Democracy Author(s): Andrew Roberts Reviewed work(s): The Macropolity by Robert S. Erikson ; Michael Mackuen ; James Stimson Elections as Instruments of Democracy: Majoritarian and Proportional Visions by G. Bingham Powell Jr. Democracy, Accountability, and Representation by Adam Przeworski ; Susan Stokes ; Bernard Manin

...

Source: Comparative Politics, Vol. 37, No. 3 (Apr., 2005), pp. 357-376 Published by: Ph.D. Program in Political Science of the City University of New York Stable URL: http://www.jstor.org/stable/20072894 Accessed: 27/07/2009 12:47 Your use of the JSTOR archive indicates your acceptance of JSTOR's Terms and Conditions of Use, available at http://www.jstor.org/page/info/about/policies/terms.jsp. JSTOR's Terms and Conditions of Use provides, in part, that unless you have obtained prior permission, you may not download an entire issue of a journal or multiple copies of articles, and you may use content in the JSTOR archive only for your personal, non-commercial use. Please contact the publisher regarding any further use of this work. Publisher contact information may be obtained at http://www.jstor.org/action/showPublisher?publisherCode=phd. Each copy of any part of a JSTOR transmission must contain the same copyright notice that appears on the screen or printed page of such transmission. JSTOR is a not-for-profit organization founded in 1995 to build trusted digital archives for scholarship. We work with the scholarly community to preserve their work and the materials they rely upon, and to build a common research platform that promotes the discovery and use of these resources. For more information about JSTOR, please contact [email protected].

Ph.D. Program in Political Science of the City University of New York is collaborating with JSTOR to digitize, preserve and extend access to Comparative Politics.

http://www.jstor.org

Review

The Quality

of Democracy Andrew

Robert

S. Erikson, Michael

York, Cambridge University G. Bingham Proportional

and James Stimson, Mackuen, 2002. Press,

Article

Roberts

New

The Macropolity,

Powell, Jr., Elections as Instruments of Democracy: Visions, New Haven, Yale University Press, 2000.

Majoritarian

and

and Bernard Manin, Susan Stokes, eds., Democracy, Przeworski, and New York, Cambridge University Press, 1999. Representation, Accountability,

Adam

and Democracy: Neoliberalism Stokes, Mandates New America, York, Cambridge University Press, 2001. Susan

Susan Stokes, ed., Public Support for Market York, Cambridge University Press, 2001.

Reforms

by Surprise

inNew Democracies,

in Latin

New

For the past two decades the main topic of research in comparative politics has been democratization. The question troubling scholars has been why some countries or waver become and stay democratic, while others either do not democratize between democracy and authoritarianism. This research program has been both pro lific and successful. Today, a considerable amount is known about why democracies are distributed around the world the way they are.1 the third wave has largely ended, and there appear to be few new Nevertheless, cases of democratization

or authoritarian relapse to explain. Existing cases, more have been over, analyzed in depth both through careful case studies and aggregate statistical analyses. While more could be done, scholarly attention might yield greater gains by focusing on aspects of democratization besides the transition to and the persistence of democracy. One interesting place to look is in recent work that argues that many of these new democracies do not function as expected. Guillermo O'Donnell, for example, has noted that many Latin American democracies should be termed delegative democracies because presidents, while elected democratically, rule without reference to public opin

357

Comparative

Politics

April 2005

ion or institutional checks.2 For similar reasons, Schmitter and Karl have proposed a def inition of democracy that looks at the presence of accountability rather than elections.3 How well do democracies work? How well do they produce responsive, accountable, and representative governments? This area of research might be called the quality of democracy. It is the next logical step from analysis of democratization. It also offers a relatively underdeveloped research program with much room for innovative work.

The Field of Democratic

Quality

Research on democratic quality potentially includes virtually any aspect of politics in democratic countries. The field expands even further if quality is interpreted in both its positive and normative senses. However, the phrase should mainly be used to study issues inherent in the democratic process, rather than those common to all political systems. From this perspective, the central concern of democratic quality is the connec tions between citizens and their elected representatives. Democracy distinguishes links between the prefer itself from other forms of government by institutionalizing ences of citizens and the actions of politicians.4 These linkages stem directly from the key institution of democracy: free and fair elections. in three powerful ways. First, they Elections connect citizens and policymakers allow citizens to select the rulers and policy directions they prefer. Second, they give as they govern. to respond to public preferences these rulers strong incentives to recall rulers they judge to be unsuccessful. The study Finally, they allow citizens of the quality of democracy is intimately related to these three processes, later as and accountability. The quality of a democracy defined mandates, responsiveness, concerns

the nature and strength of these links. The study of democratic quality ismotivated by the failure of free and fair elec tions sometimes to foster these three links. Citizens may be unable to make good choices (whether from a lack of good options or information); politicians may gov ern without reference to public opinion; and unsuccessful politicians may escape electoral punishment. In these cases democracy is not fulfilling its potential. The five books under review all focus on democratic quality in this sense. They look at the linkages described above. They examine the degree to which politicians respond to public opinion, follow through on their promises, and are held account able for their actions. Each book focuses on three issues central in understanding the quality of democracy. The first is the proper normative standard for judging citizen elite linkages. In practice, the three standards suggested above are not always mutu ally compatible, and trade-offs often have to be made between them. The second is the best way to measure how well these standards are achieved. This problem turns as well as theoretically. The third is a theory out to be quite tricky methodologically

358

Andrew Roberts quality. Why do some democracies have higher quali others? is, stronger linkages?than ty politics?that While these issues have often been investigated, they have yet to receive attention as part of a larger, connected research program. They have also been relatively neglected in the field of comparative politics. Of course, until recently democratic regimes were in theminority, but the global expansion of democracy has brought them to the forefront. about what causes democratic

The books reviewed here open important paths in the study of democratic quality, but one of the largest questions of all remains open. Do the linkages inherent in democratic quality is actually improve human welfare? Ultimately, democracy own as as Do its for sake for much its consequences. strong linkages make important citizens better off? Eventually itwill be necessary to address the Burkean argument that politicians often need to be delinked from the public. This question will come to the forefront

in the next generation

of studies.

Previous Work In an introductory and Representation, essay to Democracy, Accountability, for thinking Przeworski, Stokes, and Manin helpfully lay out a set of definitions is responsive "if it adopts about mass-elite linkages. In their terms, a government that are signaled as preferred by citizens," where signals include public policies if it follows opinion polls and political action.5 A government ismandate-responsive through on the promises itmakes in campaigns. That is, the government responds to it has been given in winning the election. Finally, a government is the mandate if "citizens can discern representative from unrepresentative accountable govern ments and can sanction them appropriately."6 The authors additionally define a gov ernment as representative if it acts in the best interest of citizens where best interest can be demarcated in different ways. These definitions will be followed in the remainder of the article. They are illustrated in Figure 1. Responsiveness, mandates, and accountability have all been the subject of empirical investigation in the past. However, studies have focused mainly on theUnited States and a handful of advanced democracies. Partially for this reason they have also not probed deeply into the causes of quality but have rather tried to show only the existence (or nonexistence) of a certain type of quality. Moreover, they have not been combined into a single research program but rather have proceeded in fits and starts. has received the lion's share of the attention in studies of democ Responsiveness ratic quality. A large part of democracy's normative appeal is that it will produce is a governments that follow the will of the people (though whether responsiveness has the of become Political scientists, good thing debate).7 subject recently mainly have devoted considerable resources to determining whether policy Americanists, choices

reflect public opinion.

359

Comparative Figure

Politics

1 Citizen-Elite

April 2005 Linkages Responsiveness Mandate Responsiveness

Preferences

-> Signals

-> Mandates

-> Policies

-> Outcomes

Accountability

from Przeworski Source: Reproduced p. 9.

et al., Democracy,

Accountability

and

Representation^

The classic work

on this issue isMiller

and Stokes' "Constituency Influence in the relation in 116 between congres Congress," public preferences sional districts and the roll-call votes and issue preferences of the congressmen rep resenting them.8 Their main finding was a high correlation (with variations across issue areas) between the two, indicating considerable responsiveness. Critics, howev er, raised doubts about the direction of causality. Perhaps congressmen were influ encing their districts rather than the other way around. Page and Shapiro tried to solve some of these problems by using time-series analysis.9 They looked at whether shifts in public opinion were followed by appropri ate changes in policy and found such correlation in roughly two-thirds of cases. As that Page noted, these studies, including his own, did not eliminate the possibility earlier changes in policy had affected public opinion or that other variables?for or world events?might the media be responsible for these policy example, which

examines

In short, most of the studies in this tradition have found a high degree of but they are mainly confined to the United States (limiting their responsiveness, and seem to suffer from potentially fatal ability to find the cause of responsiveness) changes.10

and Stimson's book attempts to rein problems.11 Erikson, Mackuen, methodological this research tradition. vigorate are more diverse. Again, most early work Studies of mandate responsiveness effort was spent tracking down the elu focused on the United States. Considerable sive presidential mandate?whether presidents were given a mandate from voters. cast The classic works in this genre, particularly Kelley's Interpreting Elections, doubt on whether such mandates can be studied scientifically.12 Indeed, it is remark ably difficult to interpret what voters want in any given election.13 One solution to this difficulty is to focus not on what voters are voting for, but on what

360

politicians

campaign

for in elections.14

The advantage

of this formulation

is

Andrew Roberts that it is possible to measure more precisely the actions of politicians than the inten tions of voters. The Comparative Manifestos Project has had this goal. The organiz for ten advanced ers of this project have put together a database of party manifestos industrial democracies.15 They have combined assessments of these manifestos with follow in public spending to determine whether governments data on changes cases most on in is that The commitments. their finding through proclaimed policy governments

do follow extends

Democracy endemic problem

through this research

on their mandates.16 tradition

Stokes' Mandates

to Latin America

and

and the supposedly

there of politicians' ignoring their campaign promises. voters notice what politicians do and sanction Accountability?whether

them?

has attracted attention because it iswidely believed that citizens lack the sophistica tion to make the responsiveness or mandate conceptions work. The dominant vein of research looks at the effects of economic conditions on elections as a gauge of the extent to which voters punish politicians for negative outcomes. Most of this work has analyzed individual level data, and there is a wide consensus that economic con ditions are among the most important determinants of voting behavior.17 A smaller number of studies looks at cross-national variation in the relation between econom ics and election results, though their results are not as consistent.18 Powell's and the of accountability books add to knowledge by respectively measuring retro in and of advanced democracies whether investigating degree accountability spective economic voting prevails during market reforms in developing countries. In short, there has been a good deal of interest in issues of democratic quality. Stokes'

Research, however, has not proceeded under a unified banner. Much of it,moreover, has focused on the United States or advanced industrial democracies. This focus severely limits its ability to determine the causes of quality. In comparative politics, where such research is presumably most important because of the flaws of many thirdwave democ racies, it has taken a back seat to research on the causes of democracy.

The American

Baseline

focuses only on American politics, Though The Macropolity account of its of democratic quality. Using comprehensiveness

it is noteworthy for the an enormous amount of

empirical data and high-powered statistical techniques, Erikson, Mackuen, and Stimson investigate not only whether public opinion influences elections and policy outcomes, but also whether policies influence public opinion.19 Because of the work's scope, it can provide a set of standards against which to judge other democracies. The authors set their argument against the widely held belief that voters are ill informed about politics. Indeed, this conclusion is one of the main results of public and opinion research since the 1950s. If true, then the idea of responsiveness

361

Comparative

Politics

April 2005

accountability have to be discarded. To what can politicians respond if citizens do not give them clear signals about what they want? How can they be held accountable if voters do not know what they have done? Interestingly, Erikson, Mackuen, and Stimson accept this argument at the individ ual level but argue that at the collective level public opinion is real and surprisingly to them, the collective public manifests not just rational well-informed.20 According of rational expecta ity in its ordinary sense, but also the stronger characteristic tions.21 That is, it not only uses all available information about the past, but also looks well into the future. It does so not because all citizens are well-informed, but because a small number of informed citizens is enough to determine aggregate trends. In essence, the uninformed cancel each other out. In the first half of the book, Erikson, Mackuen, and Stimson investigate how govern ment performance affects public opinion and voting, the issue of accountability. They find that governments can not hide from the public. Presidential approval and election results, for example, appear to be most strongly a function of the economy. Moreover, citizens rely less on retrospective evaluations of the economy than on forward-looking predictions that are remarkably accurate. In the authors' terms, citizens are more like bankers, constantly thinking about the future, than like peasants, fixated on the past. In the second half of the book, the authors focus on the issue of responsiveness, whether policy responds to public opinion. The difficulties with this sort of research are legion. One has to do with the supposed instability of public preferences. The authors claim to have solved this problem by looking at time-series data across a variety of different issues. Using all nationwide opinion surveys over the past fifty years, they find that a single latent dimension explains thirty-eight percent of the in responses. This dimension, policy mood, captures citizens' beliefs over the proper scope of government: whether government should be expanded or con over time across a wide tracted. This national policy mood moves systematically

variance

variety of specific policy areas, thus contradicting expectations of instability. Their next step is to find a way of measuring policy. Indeed, the problem of mea surement has hindered research on what many would agree is the most important dependent variable in political science. To a large extent, institutions, parties, public opinion, and voting are important not for their own sake, but because they have an influence on policy. Yet the literature on policymaking is surprisingly weak largely because of the difficulty of dealing with this dependent variable. and Stimson find a number of ways of defining policy out Erikson, Mackuen, comes for each of the four major institutions of U.S. government: the House of the Senate, the Presidency, and the Supreme Court. They look first Representatives, at policy activity, the sum total of all the actions taken by these branches. But because policy activity deals mainly with minor matters, they also create a measure of important laws. Here

362

they borrow Mayhew's

dataset of important legislation

and

Andrew Roberts of policy innovation that is the difference between the number of laws passed in a given year.22 They conclude that policy is liberal and conservative to policy mood. Policy adjusts extremely rapidly to shifts in remarkably responsive an almost one-to-one relation between opinion changes find public opinion. They and policy changes, and most of this change occurs within a year or two. This rela tion applies to all four branches of government, though they respond at somewhat create a measure

speeds. For the House and Senate, the authors further break down respon into two pathways: anticipation (policy changes because representatives turnover and their (policy changes because out-of-touch represen positions) change in the House, where elections are always tatives are replaced). Responsiveness in the Senate, around the corner, is driven by anticipation, while responsiveness where elections are more competitive, is driven by turnover. these findings are not the last word. Page While powerful and comprehensive, different siveness

points out that, despite the enormous amount of data utilized, much is still left out.23 The aggregation involved in building up measures of public opinion and policy lim its the analysis to a simple left-right dimension purged of issues like foreign policy, abortion, and crime, which do not follow the national mood. Control variables, like actual events in the world, that might explain some of the variation in policy are omitted. More tellingly, the actions of elites, interest groups, and politicians them selves who might influence both public opinion and policy are not captured. Just as significant is the bottom line of policy outcomes. While policy may move does not always mean that in the same direction as public opinion, this movement does what the public wants. Many policies that enjoy widespread example, expanded health insurance?get caught in the veto-prone support?for over time does not necessari American of the gears political system. Responsiveness even for in substance thatmatter). (or representation ly imply responsiveness a take and Stimson However, Erikson, Mackuen, large step forward in trying to a not account show of only how public opinion and politics. They provide systemic elections influence policy, but also how policy feeds back on elections and public opinion. They find that when policy moves in one direction, the public mood moves the government

is important because, ifmood did not react in the opposite direction. This movement to policy, then policy might reflect the work of ideologues shaping public dialogue. They liken macro public opinion to a thermostat for government policy. When policy is too liberal (conservative), the public demands more conservative (liberal) policies, and the government responds to these demands. In the end, the authors compare a weak version of democracy, where inattentive citizens keep government responsive through occasional replacement of politicians, and a strong version, where watchful and attentive citizens force politicians to stay responsive at every moment. They argue that recent American history supports the strong version of democracy. Of course, the problem is that the authors have consid is in the U.S. rel ered only one case. It is not known either how high responsiveness

363

Comparative

Politics

April 2005

are. The U.S. may ative to other countries or what the causes of responsiveness as a baseline, but only once more is known about responsiveness elsewhere.

Advanced

serve

Democracies

InElections as Instruments of Democracy, Powell takes amajor step toward comparing the quality of advanced democracies. He finds that comparison is not simple. He con cludes that there are two visions of democracy which aim toward different goals. The majoritarian vision believes that themajority should rule directly by identifying and vot ing for a definite government and holding that government accountable at the next elec tion. The proportional vision, by contrast, believes that elections represent all opinions and that government institutions should be designed to give all views a say in policy.24 This distinction may appear similar to Lijphart's classic differentiation of majoritari an and consensus democracies.25 Indeed, the two agree that the size of the governing majority is a clear distinction between the two types of democracy. But while Lijphart looks mainly at institutions and theway power is exercised within government, Powell is more focused on the representation of citizens' opinions. The difference between them, however, ismore one of emphasis than of genuine disagreement. Powell argues that it is problematic to compare these two visions according to a sin should be able gle measuring stick. The majoritarian vision aims for amandate?voters to identify and select a future government?and should be able accountability?voters to clearly pronounce judgment on a government and replace it if necessary.26 The pro portional vision, by contrast, does not aim for either of these goals and be judged by whether it achieves them. Rather, it aims for what Powell representation. Itwants voters to be able to choose representatives with to their own who will then act on their behalf. The proportional vision

thus should not calls authorized opinions close allows for this

representation because it isworried thatminority opinions will be ignored. These visions are reflected in institutional designs. The majoritarian vision

is facili

tated by plurality electoral systems and institutions that concentrate power in the gov ernment. The proportional vision, by contrast, is helped by electoral laws ensuring pro portional representation, strong legislative committees, bicameralism, and federalism, all institutions that give minorities policy influence. Most countries can be classified as one type or the other, though a few, including the U.S., Spain, Japan, and Italy, do not fit into either one of these categories and are called mixed systems. The major portion of Powell's work consists of testing whether twenty advanced democracies actually fulfill their self-professed goals. Do majoritarian systems deliver mandates and accountability? Do proportional systems deliver authorized representa tion? The answer is that each system does well judged by its own standards but poorly well

by the standards of the other.

364

Andrew Roberts

Powell first looks at election results. As expected, the proportional vision represents most groups proportionally. Not only do all parties receiving significant shares of the vote obtain seats in parliament, but everyone gets a say in policymaking either by par ticipating in large coalitions or through strong parliamentary committees. All parties chosen by voters thus have a say in policymaking, whether in the government or out of it.

The majoritarian system is slightly more problematic. These systems give voters identifiable choices of future governments and usually give the chosen government the power to enact its policies over a relatively powerless opposition. In this sense, there is what Powell calls a mandate, but itmight be better termed identifiability. The problem is that thewinning party usually does not win amajority of votes, but only a plurality. A party winning forty percent of votes may control all the levers of power. Even more dis turbing are the approximately twenty percent of cases where the second place party wins the majority of seats. Accountability works better. Incumbent parties usually lose votes, and governments are decisively replaced when they fail.27 Powell repeats this analysis with voter preferences rather than election results. His reasoning is that elections artificially limit the number of options that voters have to choose from, and thus actual preferences should be considered as well. He uses both public opinion polls and expert placements of party ideology. His aim is to determine how close the government (that is, the average ideology of the parties in the parliament weighted by their influence) is to themedian voter. This approach measures responsive ness if it is assumed that the policy outputs of a government reflect its ideology. Here the proportional vision again outperforms the majoritarian one. The pivotal vote in proportional systems is closer to themedian. This conclusion is surprising. One would expect Downsian competition inmajoritarian systems to push parties toward the center and reward the more centrist party, but the opposite occurs just as frequently. Majoritarian systems fail on this score by not weeding out third parties.28 When there aremore than two parties, competition is no longer centripetal, and winning parties may be far from the center. Proportional systems can compensate for this distortion through power-sharing at the level of government, but majoritarian systems can not. In essence, majoritarian systems put all their chips on election outcomes, while proportional sys tems allow postelection bargaining to take some of the pressure off of elections. While persuasive, these results leave out the final step in the causal chain. There is no question that the ideological color of governments is important, but what really mat ters is how ideology is translated into policy. While majoritarian governments may rep resent only aminority of voters and be far from themedian voter, do they produce poli cy at variance with voters' desires? Moreover, the worry about proportional systems is not that they will stray from the median, but that they will not stray anywhere. That is, they will produce far less significant legislation because they are hamstrung by too many parties and too much opposition influence.29 Indeed, one might prefer amajoritar ian system despite its flaws because

it actually does something.30

365

Comparative

Politics

April 2005

In short, Powell shows that elections and institutions affect who governs. But it is not clear whether they affect how well a system is governed. The problem is to create a measure of policy choices that is comparable across countries, something no one a powerful message emerges from Powell's work. There has yet done.31 Nevertheless, is a trade-off between types of quality. While majoritarian systems perform fairly by their own standards, they perform poorly by the standards of proportional and authorized representation are pre systems, and vice versa. If proportionality

well

ferred, some accountability and identifiability not all good things go together.

Quality

Failures

will have to be sacrificed. On quality,

in Latin America?

and Stimson and Powell find a substantial degree of Erikson, Mackuen, in there is much more doubt about whether advanced democracies, responsiveness newer democracies for example, has argued that function equally well. O'Donnell, as in Latin America govern they wish, ignoring campaign promises, pub presidents lic opinion, and legislatures. He believes this situation is partially a result of the

While

uncongenial economic environment these countries have faced. InMandates Stokes analyzes one aspect of this supposed failure and Democracy, of democracy: the tendency of presidents to ignore the platform they were elected on, what she calls mandate violations. Her research questions are why politicians sometimes go back on their promises and what effect this action has on their soci eties. She investigates this question through both statistical analysis of forty-four inArgentina, Fujimori in elections and case studies of mandate violations by Menem Peru,

and

Perez

in Venezuela.

The nature of politics in Latin America allows Stokes to sidestep many of the dif ficulties of research on mandates. Politics in these countries in the years she ana to 1995?centered mainly on the issue of market reforms. Stokes is lyzes?1982 that is, in favor thus able to classify election campaigns as either efficiency-oriented, that is, against them.32 Most campaigns fit of market reforms, or welfare-oriented, in one or the other category. She then looks to see whether the elected candidate through on his pledge. From a dataset of forty-four elections, she finds that in twelve cases (27 percent) of these the candidate promised one thing, but delivered another. Interestingly, same to switchers all switched in the direction, from welfare promises efficiency policies, and all did so within days or weeks of taking office. In short, the data sug finds. The high per in Latin America is as bad as O'Donnell gest that democracy

well

followed

centage of switches on the key issue of the campaign and the rapidness of the switches seem to suggest that politicians cared little about voters and instead pur sued their own personal interests. Yet this explanation is riot the only one and is not

366

Andrew Roberts the one for which Stokes finds evidence. Instead, she argues that inmost cases presi act in the best interests of the dents switched out of a desire both to represent?to citizens?and

to win

reelection.

Her argument begins with the observation that politicians must support welfare oriented policies in order to win elections. There is simply not enough mass support in the region as evidenced by the small number of parties that have for neoliberalism won elections with an efficiency-oriented campaign. No matter their real beliefs, campaigns. office-seeking politicians will thus run welfare-oriented But some politicians believe that voters are mistaken. They understand that, even if voters prefer them, welfare-oriented policies will be economically disastrous. They in which case they bait and may come to this realization during their campaigns, switch, or soon after they take office. In either case, they decide that they should not follow through on their campaign promises. This decision does not mean that they sacrifice themselves for the good of the country. In fact, they do not give up hope of staying in power. Their hope rests in voters' uncertainty about their beliefs. Citizens may oppose neoliberalism, but they can be persuaded to change their minds if it brings good results. Politicians thus haz ard that the superiority of their choices, seen in higher growth rates, will make it clear to voters that switching was the correct decision. To prove this point, Stokes shows that the patterns of switching are not consistent with an interpretation of politicians as rent-seekers who switch to line their own pockets. For example, competitive elections make switches more likely. Politicians dissemble only when they have no other choice. Similarly, countries with term limits do not produce more switchers, though they ought to if politicians were only out to enrich themselves. She also finds evidence that voters are genuinely uncertain about the effects of economic policies. Their ex post judgments on politicians who switch are much more sensitive on nonswitchers. to outcomes than their judgments Fortunately for politicians, switching usually yields good economic growth. of politicians do not support an interpretation that Furthermore, the motivations were rent-seekers. Presidents do not appear to enrich themselves, gain they merely additional campaign contributions, or shirk their responsibilities by switching. More to the point, they appear to have genuinely believed that following through on their campaign promises would be disastrous for the country. Switchers, moreover, were all other things electorally at the next election relative to nonswitchers, Voters lies and them. Politicians thus pay a cost for recognize punish being equal. not and do receive obvious benefits. switching personal However, even if Stokes' argument that mandate violations occur in the name of

punished

representation is accepted, all is not dates are not credible, then voters these switches are likely Moreover, distrust politicians and the political

right in Latin America. If the platforms of candi have little basis for choosing between parties. to have second-order effects as citizens begin to process. While a case can be made for the man

367

Comparative

Politics

date violations

April 2005

that have occurred

in Latin America,

the case for continued violations

is far weaker.

The important question, from this problem. It is hard Europe or Asia. Stokes finds violations than older ones?a

in particular has suffered then, is why Latin America to think of similar cases in new democracies in eastern that young political parties are more likely to commit

strong party system ensures consistency. But while the some Latin American countries from western Europe, of parties distinguishes youth not it does distinguish them from other new democracies. The difference has some thing to do with the character of mass publics in Latin America. There is a remark able disconnection between the economic realities as seen by politicians and voters.

to elect efficiency-oriented In contrast, in eastern Europe voters were willing one are roots The of this disconnection of the parties.33 key issues in explaining eco nomic reform, but answers are so far thin. Voters, of course, may not be entirely at fault here. As Stokes points out, politi cians need tomake a greater effort to explain what is happening and why they violat to do so. Eastern To date politicians ed their mandates. have been unwilling have had the advantage of being able to link reforms with politicians to Communism and a return to Europe.34 However, judging by the suc opposition cess that switchers have had in getting reelected in Latin America, there may be

European

some potential

Economic

to create mass

Reform

support for free markets.

and Democratic

Quality

is not the end of democratic quality. If voters can not select The failure of mandates ex still catch them ex post. Dominguez ante, argues that, as long may politicians they as democracy survives, accountability can substitute for mandates.35 Voters can keep In fact, if politicians on a leash by sanctioning incumbents for poor performance. voters additionally punish politicians for breaking their promises, politicians would have a strong incentive to keep their promises. Stokes provides some evidence for in that the consequences of efficiency-oriented this conclusion policies matter for election outcomes and mandate violators are punished at the polls relative to nonvio lators,

ceteris

paribus.

is whether The standard of accountability that is usually applied in democracies affects election outcomes. The idea that voters reward or economic performance punish incumbents based on their economic performance has become a standard in political science.36 Less commonly remarked on is how it relates to the quality of is wanted, then ret It is easy to see that, if good economic performance a to incentive take care of the economic strong voting gives politicians rospective are as of wrinkles studies the Of there course, here, economy. political business

democracy.

368

Andrew Roberts cycle make clear, but if voters have rational expectations, as Erikson, Mackuen, and Stimson argue, economic voting is probably a good thing.37 There is one place, however, where retrospective economic voting might lead to poor outcomes. In developing countries in need of drastic reforms, economic voting may be counterproductive. The reason is what has come to be called the J-curve.38 Major reforms make things worse before they make them better.39 If citizens vote according to the state of the economy, then incumbents who try to represent and undertake necessary but painful reforms will be punished. Forward-looking politi cians will thus avoid reforms for fear of electoral retribution. This logic has ledmany to economic to the conclusion is not conducive that democracy reform and that executives will produce better economic results. However, incumbents have often managed both to undertake painful reforms and to get themselves reelected. In a clever introduction to Public Support for Market Reform, Stokes lays out a framework for analyzing public support for market reforms. She hypothesizes that rational voters can take other positions than retro For economic voting. example, faced with economic decline as a result of spective unconstrained

neoliberal reforms, voters might instead look to the future and believe that current pain means future prosperity. Bad conditions will then be greeted with both opti mism and support for the government. Stokes terms this intertemporal voting, and it iswhat market reformers hope for. There are two other ways that voters might respond to economic decline. They about the economy but still support the incumbent, because might be pessimistic believe the government is not responsible for the decline but is simply cleaning they mess the left up by the previous regime. She calls this stance exonerating. And vot ers might

greet decline with optimism, because they know things will be better in the vote against the incumbent either because they see their own group as but future, hurt unfairly or because they are committed to the opposition. This stance is called distributional or oppositionist. The remainder of the book uses public opinion data from six states?Spain, East see which of these stances best Poland, Mexico, Peru, and Argentina?to the public response tomarket reform. In each case, the authors regress two dependent variables, support for the incumbent and optimism about the economy, against a variety of measures of economic performance. The results show that citi zens adopt a variety of postures toward economic decline. While rising unemploy

Germany, describes

ment

to the government, usually has the expected effect of increasing opposition are more to reactions inflation and stagnant growth diverse. The intertemporal economic news producing optimism and support?is response?negative fairly com

mon. There is also evidence

that some citizens exonerate the government. It isworth in many of these countries encouraged citizens to take such noting that politicians that there would be no gain without pain or postures through public pronouncements that the economic decline was the previous regime's fault.

369

Comparative

Politics

April 2005

Stokes proposes a number of hypotheses to explain where retrospective economic voting can be expected and where not, but the nonrandomly chosen data set of six coun tries does not allow her to test these hypotheses. She reasonably suggests that economic voting will be more common where the ruling party is clearly in control, where the opposition is not clearly associated with the old regime, and where party identification is weak, among other conditions. Future research may be able to test some of these inter esting hypotheses and thus contribute to explanations of when reforms are sustainable. about the volume is its challenge to the recommendation of concen Noteworthy trated executive power during economic reforms. Citizens do not have to be duped or ignored to pass reforms. It thus provides some measure of hope that market reforms can be undertaken without sacrificing too much in democratic quality. It thus chal can not see that voters in new democracies lenges the widespread assumption beyond their next paycheck. In fact, the evidence shows a considerable degree sophistication in voters who are experiencing democracy for the first time.

Normative

of

Standards

Many people favor democracy because it creates incentives for politicians to represent the best interest of the public. Yet much research needs to be done to explain how differ ent democracies succeed in this task and why some are more successful than others. Moreover, study of the normative issues of democratic quality is still at a fairly rudi mentary stage. Should politicians fulfill their mandates? Should they respond to public opinion? For what should voters hold them accountable? If these standards conflict, which should be preferred? It is noteworthy that both Powell and Stokes spend consider able time considering the suitability of their normative standards. deal with these Several chapters of Democracy, Accountability, Representation seems to function well if issues.40 For example, concerning mandates democracy follow through on their promises. As Stokes notes, voters have clear politicians increases. information on election day about what to expect, and trust in politicians But what if conditions change, and it is no longer in the public's best interest that politicians follow through on their promises? Similarly, if deliberation among legis lators helps to produce better policy, then the desirability of mandates might need to be reconsidered.41 Indeed, for these reasons no country has an imperative mandate. Its value can best be seen by considering its The same is true for responsiveness. converse: a government that does the opposite of what the public wants. But there are situations where responsiveness breaks down. There are technical matters the public is not well-informed. There are situations where public opinion is unclear, contradictory, or evenly divided. Public opinion may be unstable when stabil are better informed than citizens ity is needed. Any situation where governments a implies breakdown of responsiveness.

numerous

about which

370

Andrew Roberts

the least problematic of the standards, has its Even accountability, seemingly flaws. If voters choose a specific standard to judge the government, two problems follow. First, governments that put forth less than their best effort but for fortuitous reasons achieve good outcomes may be reelected. Second, governments that perform well but for reasons outside their control did not achieve the desired outcomes may not be reelected. While accountability may give politicians an incentive to avoid seri ous failures,

itmay not induce them to put forward their best effort. it is not possible to achieve all these standards at the same time. If Moreover, to stick to their mandates, have they will have to sacrifice some respon politicians siveness. Indeed, accountability and mandates may themselves be contradictory, as they require voters to send two separate and distinct messages with a single vote: an evaluation of the incumbent's performance and a choice among future policy direc tions. This contradiction is one of the reasons mandates are hard to discern. On the other hand, the two links may also substitute for or reinforce one another. The accountability standard is often proclaimed as the most effective and practi cable one. Citing the impossibility theorems of social choice theory, scholars like have argued that anything more than throwing the bums out is infeasible. Schmitter and Karl similarly believe that a workable definition of democracy should

Riker

be founded on accountability. However, Fearon, in another contribution to this book, argues that much of the rhetoric about politics indicates that elections are seen more as selecting good politicians than as sanctioning bad ones. There is a widespread dis like of office seeking behavior even though accountability functions best when politicians want to be reelected. The same goes for the premium that is put on princi ples and consistency. Further, term limits are popular despite the fact that they make accountability much more difficult. This line of reasoning is not watertight. The accountability conception still has much to recommend it. But it does suggest how little is known about what iswanted from democracy. These reflections do not exhaust all of the normative issues at play. However, they suggest that much more thinking is needed about the standards to which govern ments should be held. In fact, these areas seem to be tailor-made for productive dis empirical political scientists and political theorists, two groups that had little to say to each other.42 Indeed, it is unfortunate that political theorists have not devoted much energy to questions of democratic quality in recent

cussions between have heretofore

years. It is hard to think of a worthy

successor

to Hanna Pitkin's classic, The Concept

of Representation.^

Conclusions As

democracy

and Avenues expands,

for Future Research variation

among

existing

democracies

will

inevitably

371

Comparative

Politics

April 2005

not automatically function in the same way as old as democracies. Most troubling, they may not link their citizens with policymakers well as democratic governments in the advanced industrial states. Poorly functioning are better than nondemocracies, democracies but they are still troubling both for increase. New

democracies

will

and for citizens of these countries. More systematic study of the performance of new democracies (and even old on a of of indicators democratic is The needed. five books ones) quality variety under review both revive attention to the subject of democratic quality and provide a set of sophisticated techniques to analyze it. There are several areas where additional democrats

work might be done. to extend these studies to new First, to make causal statements it is necessary a at states and therefore do not cap of limited group regions. All of these books look ture all of the variation in democratic quality. To understand why certain democra and accountability, it is neces cies foster responsiveness, mandate responsiveness, sary to repeat these studies in a wider sample of democracies. The applicability of these concepts should be kept inmind. O'Donnell argues that inmany less advanced informal institutions trump formal ones.44 Laws may be passed, but democracies they are not applied or enforced. Informal institutions of particularism may be better predictors of behavior than legal norms. The influence of informal institutions before focusing entirely on formal institutions, as several of the studies here do. Variations across policy areas should also be considered. All of the studies here use aggregate measures of outcomes. For Erikson, Mackuen, and Stimson it is policy liberalism; for Powell, distance of the government from the median voter; and for between the tenor of campaigns and policy choices. Stokes, the correspondence Much is hidden under these aggregates. Governments make policy in a large variety should be considered

of areas, and it would be surprising if responsiveness, accountability, and mandate did not differ across different policy areas. Social policy, for exam responsiveness ple, likely exhibits different dynamics than fiscal or regulatory policy. Scholars have already done a great deal of research on policymaking within indi vidual policy areas in both developed and developing countries. These studies need to be placed in a larger framework. The quality of democracy may provide just such a framework. One way of synthesizing studies of privatization, pension reforms, and fiscal policy, for example, may be in terms of the extent to which citizens' prefer are so diverse that ences are expressed in policy decisions. Studies of policymaking some such unifying theme would yield considerable gains. Hard thinking about concepts like path dependence and blame avoidance that are indi prominent in work on policymaking would be required. Does path dependence cate that present preferences are being ignored or that both past and present prefer ences are being taken into account? Indeed, it is not widely recognized that path

372

Andrew Roberts are not dependence may be at variance with democracy. Answers to these questions obvious, but they are important in assessing how well democracy works. Moreover, several of the books under review suggest that institutions play an impor tant role in the quality of democracy. Much of the recent work relating institutions to policy outcomes could be reinterpreted in light of quality considerations. Inwhat sense, for example, does a system with a large number of veto players help or harm democrat and accountability may be weakened, ic quality? Mandates though responsiveness might be improved. It is interesting, too, that several recent veins of work postulate that nonaccountable institutions produce better outcomes than accountable ones. This argu ment has been stressed most strongly inwork on central bank independence.45 Zakaria likewise worries that institutions in the United States have been made too responsive to the citizenry.46 Under what conditions should institutions be made accountable and responsive? How well the public is informed obviously plays a role. issues are key to democratic quality.47 For mandates, Finally, informational to it is necessary that and outcomes, accountability produce good responsiveness, of policy citizens be informed of what politicians say and do and the consequences a and Stimson show that the macro public possesses choices. Erikson, Mackuen, as While rational that characterize information of expectations. they high degree countries might doubt whether ordinary citizens in their scholars of developing could achieve this level of informedness, it is worth noting that this result is driven by only a small number of well-informed citizens.48 At the macro level the uninformed tend to cancel each other out. and confirmation. On the one Stokes' evidence presents both disconfirmation seem to be of Latin America citizens hand, very poorly informed about the options to vote for efficiency-oriented their countries. Their reluctance campaigns facing what that do not will be forced to do while in office. know suggests politicians they Not only are people uninformed, but they are systematically misinformed.49 On the other hand, her analysis of public opinion in countries undergoing market reforms finds a range of sophisticated responses to complex reforms. Voters seem able to countries

understand both the short-term negative effects of reform and its long-term positive consequences. A capacity to learn may explain some of this contradiction, but clear ly more work is needed on how well informed citizens are about the behavior of politicians and the problems facing their country. The issues described here do not exhaust the problem of democratic quality. The quality of choices that citizens have in elections, the extent of their participation, the quality of the media, the extent of clientelism and corruption, the performance of subnational governments, the judiciary, and bureaucracy are all important parts of link the between citizens and policymakers. These books make clear that citizens matter. Politics is not a self-contained circle of elites. Elections and public opinion affect policies. The links are not always direct and clear, but citizens

select politicians,

express

their policy preferences,

and sanc

373

Comparative

Politics

April 2005

are a triumph for democracy. The tion incumbents. These actions in themselves degree, kind, and source of the public's influence need more study. Once they are better understood, the crucial question of representation?whether accountability, lead to policies in the best interest of the public?can mandates, and responsiveness be addressed. The classic debate between those who believe that citizens know best and should be given more control and those who think that important decisions should be left in the hands of knowledgeable politicians and experts is still without a answer. The dependent variables of these studies will independent variables of the next generation of work.

definitive

thus become

the

NOTES Iwould

like to thank Jamie Druckman,

Jeff Jenkins, Kathy Thelen,

and the editors of this journal

for help

ful comments. 1.

"What Do We Know about Democratization after Twenty Years," Annual Geddes, of Political Science, 2 (1999), 115-44. 2. Journal of Democracy, 5 (January 1994), 55-69. Guillermo O'Donnell, "Delegative Democracy," "Illusions about Consolidation," Journal 1 (April 1996), Also, Guillermo O'Donnell, of Democracy, inNew Democracies," "Horizontal Accountability Journal of Democracy, 9 34-51; Guillermo O'Donnell, 13 (July 1998), 112-26; Thomas Carothers, "The End of the Transition Paradigm," Journal of Democracy, See Barbara

Review

(January 2002), 3. Philippe

5-21.

Is Not," Journal and Terry Lynn Karl, "What Democracy Is...and C. Schmitter of 2 (Summer 1991), 75-89. Democracy, 4. I have avoided Jacobs and Shapiro call this linkage "substantive this term only democracy." I fear that itmay be confused with direct or participatory Lawrence R. Jacobs and because democracy. Y. Shapiro, "Studying Substantive Democracy," 1994), 9-17. Susan Stokes, and Bernard 5. Adam Przeworski, Press, (New York: Cambridge University Representation Robert

PS: Political

Science

Manin,

eds., Democracy, 1999) p. 9.

27

(March

Accountability,

and

and Politics,

6.

Ibid., p. 10. Cf. Alan Blinder, 115-26; Fareed Zakaria, 7.

"Is Government The Future

Too Political," Foreign Affairs, 76 (November-December 1997), at Home and Abroad (New York: W. Illiberal Democracy

of Freedom:

W. Norton, 2003). 8. Warren Miller Science

Review,

9.

Benjamin

Science

Review,

in Congress," Influence American and Donald Stokes, "Constituency 57 (March 1963), 45-56. on Policy," American Page and Robert Shapiro, "Effects of Public Opinion 11 (March 1983), 23^13.

Political Political

and 10. Benjamin the Links between Public Opinion Responsiveness? Page, "Democratic Untangling 27 (March 1994), 25-29. Policy," PS: Political Science and Politics, on responsiveness in other countries, mainly 11. Miller and Stokes tried to encourage data collection in western Europe but also including Brazil. For a summary of their efforts as well as several belated see Warren Miller, the data collected, in to come to terms with ed., Policy Representation attempts Western Democracies Oxford Press, 1999). (Oxford: University 12. Jr., Interpreting Elections Press, (Princeton: Princeton University 1983). Kelley Stanley Kelley, recommends Mandate,"

374

greater attention to public opinion data. See also Robert 105 (Autumn 1990), 355-72. Political Science Quarterly,

Dahl,

"The Myth

of the Presidential

Andrew Roberts 13.

Conley president will Presidential

that the important point is not the mandate itself, but in what situations an elected on a effect it has life. Cf. Patricia Heidotting mandate and what political Conley, proclaim How Elections the National of Chicago Mandates: Agenda Shape (Chicago: University argues

Press, 14.

2001). This approach that citizens vote for.

assumes

that the issues politicians

emphasize

are also

in their campaigns

the ones

I. Hofferbert, Richard and Ian Budge with Hans Kelman, Parties, Klingemann, and Democracy Press, 1994). (Boulder: Westview In another interesting work, Fishel finds that American tend to fulfill or partially fulfill 16. presidents most of their specific campaign promises. Jeff Fishel, Presidents and Promises: From Campaign Pledges 15.

Cf. Hans-Dieter

Policies,

to Presidential

(Washington, D.C.: Congressional Quarterly Press, 1985). Performance on Retrospective "A Retrospective and Douglas See, for example, D. Roderick Kiewiet Rivers, 6 (1984), 369-93. Voting" Political Behavior, "A Cross-National of Economic 18. G. Bingham Powell, Jr. and Guy D. Whitten, Analysis Voting: of Political Context," American Journal of Political Science, 37 (May 1993), 391-14. Taking Account to the definitions 19. the authors call this phenomenon of this arti according Though representation, 17.

cle, it is better termed responsiveness. I. Page and Robert Y. Shapiro, The Rational 20. See Benjamin Press of Chicago, Americans' Policy Preferences (Chicago: University

Public:

Fifty

Years

of Trends

in

1992).

21.

See James Surowiecki, The Wisdom of Crowds (London: Little Brown, 2004). in American database has been widely used in important works See, for example, politics. Keith Krehbiel, Pivotal Politics: A Theory of U.S. Lawmaking of Chicago Press, (Chicago: University to develop similar databases for their own countries. The for comparativists 1998). Itwould be worthwhile in David Mayhew, Divided We Govern: Party Control, Lawmaking, data were originally presented and 22.

This

1946-1990 Press, 1991). (New Haven: Yale University Investigations, in Jeff Manza, 23. Public," Benjamin Page, "The Semi-Sovereign Public Opinion: Polls, Policy, and the Future Page, eds., Navigating

Fay Lomax of American

and Benjamin

Cook,

Democracy

(Oxford:

Oxford

Press, 2002). University It is hard to identify works in political philosophy 24. has been slow to address normative questions philosophy racy. 25.

concerning

these two visions. the institutional

Indeed, political structure of democ

Cf. Arend

26. term

that represent

Powell's coined

Democracies: 1984), 27.

Press, 1999). (New Haven: Yale University Lijphart, Patterns of Democracy use of the term mandate does not correspond with the definitions in this paper. A better Kaare is identifiability. in Parliamentary Governments Strom, by Strom "Minority of Nonwinning Cabinet Solutions," Comparative The Rationality Political Studies, 17 (July

199-227. In proportional

replaced in full. 28. The United

systems,

States

by contrast,

is virtually

governments

the only country

are difficult in which

only

to identify

in advance

two parties

receive

and are rarely

almost

all of the

votes and seats. 29. for example, finds that systems with a high number of veto points, characteristic of pro Tsebelis, Veto Players: How portional systems, pass fewer laws than those with fewer veto points. George Tsebelis, Political Institutions Work (Princeton: Princeton University Press, 2002). 30. One might similarly argue that by muting systems are more prone to proportional accountability This hypothesis would have to be tested empirically. corruption and voter dissatisfaction. measures 31. Tsebelis has developed cross-national of the amount of policy change, but not of the direction

of change. to avoid the misun I use welfare-oriented Stokes, following Elster, uses the term security-oriented. that security might be associated with national security rather than economic derstanding security. 32.

375

33.

Bunce

Latin America.

April 2005

Politics

Comparative

thus finds a stronger correlation between She attributes it to the electoral mandate

reform

in eastern Europe than in and democracy in eastern Europe and the election of back in time. Why were liberal governments

for reform

liberal governments, but this argument pushes the problem to win mandates in eastern Europe but not in Latin America? Nationalism on the prior causes. Valerie authoritarian Bunce, past are her speculations Economic Reform," Annual Review of Political Science, 4 (2001), 43-65. able

34. might 35.

Since

some Latin American

dictators

undertook

neoliberal

be equated with welfare-oriented policies. "Free Politics and Free Markets Jorge Dominguez,

reforms,

in Latin America,"

and the nature

of the

"Democratization

to dictatorship

opposition Journal

and

9

of Democracy,

(October 1998), 70-84. In a contribution 36.

to Democracy, and Representation, Przeworski and Limongi find Accountability, that the probability of a ruler's surviving in office is not related to economic the conditions, contradicting common finding that the vote for incumbents depends on economic conditions. They suggest that certain Powell's clarity of responsibility?must institutional conditions?perhaps be present for this accountabili are not often replaced But if Indeed, in proportional ty to pertain. systems governments completely. Powell is correct, responsiveness is not thereby limited. 37. and Howard Rosenthal, On political business cycles, see Alberto Alesina Partisan Politics, Divided and the Economy Government, Press, 1995). (Cambridge: Cambridge University 38.

See Adam

1991). 39. Neoliberal

Przeworski, reforms

tributional

consequences It also contains

Democracy

and

the Market

(Cambridge:

are not always and unequivocally their many hazards.

good.

Poor

Cambridge

University

implementation

Press,

and serious

dis

are among

of Stokes' Mandates 40. summaries and Democracy and Erikson, chapter-length and Stimson's The Macropolity. Mackuen, 41. Jon Elster, ed., Deliberative Press, 1998). Democracy (Cambridge: Cambridge University 42. One of the coeditors of the book, Bernard Manin, is a political theorist, as are several contributors. 43. 44. 45.

Hanna

F. Pitkin, The Concept See note 2.

of Representation

(Berkeley:

Press,

1967).

151-62.

Alesina

Some

Banks," Foreign Affairs, 78 (March-April 1999), 2-8. 46. Zakaria, The Future of Freedom. 47. For many citations on research relating information see Philip Keefer, "What Does Political Economy countries, Vice Versa," Annual Review of Political Science, 48. Herbert Kitschelt, Zdenka Mansfeldova,

1 (2004), Radoslaw

failures Tell Us

to policy failures in less developed about Economic and Development

esp. 265-68.

and Gabor Toka, Post-Communist and Inter-Party Competition, Representation, Cooperation (Cambridge: Cambridge that the citizens of eastern Europe are very well informed about pro Press, 1999), demonstrate differences among political parties. Markowski,

Systems:

University

grammatic 49. Of course, they may fer different policies.

376

of California

and Lawrence "Central Bank and Macroeconomic Summers, Independence Journal 25 (May 1993), Credit, and Banking, Evidence," of Money, Comparative For a critique of this work, see Sheri Berman and Kathleen McNamara, and Central "Democracy

Alberto

Performance:

Party

University

be aware of what will

happen

yet still want

to send the message

that they pre