Giới thiệu (Nam)

INTRODUCTION

Larry Diamond and Marc F. Plattner

Among the many factors that shape the character and viability of de-mocracy around the world, electoral-system design holds a special fas-cination for democratic scholars and practitioners. Most of the facilitating factors for democratic development change only slowly, and not entirely by design. It may take many years (at least) to build a democratic political culture, an active and pluralistic civil society, and the administrative infrastructure of an effective state. And generating a relatively developed economy, with a sizeable middle class and only moderate levels of inequality, is a challenge that for many poor coun-tries today must be measured in decades. Yet the institutional architec-ture of a country’s political system can be changed relatively quickly. With some amendments to the constitution or the writing of a new one, a country may go from being unitary to federal or presidential to parlia-mentary. Often, the electoral system may be changed by legislation alone, even without a constitutional amendment. Among the many struc-tural and historical variables that affect democracy, few are more open to rapid and intentionally designed change than the electoral system.

Adding to the fascination is the powerful accumulation of theory and evidence indicating that the design of the electoral system does matter quite a lot for the nature of the party system and the character of politics and public policy. Particularly in combination with a few other key variables, such as the structure of the executive (presidential or parlia-mentary) and the vertical distribution of power (unitary or federal), the electoral system can shape the coherence of party control of govern-ment, the stability of elected governments, the breadth and legitimacy of representation, the capacity of the system to manage conflict, the extent of public participation, and the overall responsiveness of the system. These dimensions of democratic character and quality, in turn, may well determine whether democracy survives or fails. A badly de-signed electoral system may polarize electoral politics or may fragment the parliament so severely that nothing can be accomplished. In either scenario, it may unwittingly empower extremist political forces whose commitment to democracy is ambiguous at best. Alternatively, if the electoral system tries to correct too decisively for the danger of frag-mentation by manufacturing a majority, it may leave important social and political groups excluded or underrepresented, trading one problem for another that may be even more debilitating. Thus, while some theo-rists strongly recommend a particular type of electoral system as generally best for all countries, others warn that the challenge of appropriate de-sign very much depends on the particular country context and on which danger (for example, the lack of a decisive majority or the exclusion of important minorities) represents the most serious threat to democracy.

In the opening essay of this collection, one of the leading theorists of democratic institutional design, Donald L. Horowitz, emphasizes how powerful the electoral system can be in shaping the character of a de-mocracy and how vexing the choices can be. There are myriad ways that voter preferences can be aggregated in order to determine which parties get how many seats in parliament. “Every electoral system has biases,” Horowitz warns, and no system merely passively translates “individual wishes into a collective choice.” System designers—and the broader publics in whose interest they are acting—must therefore be aware of these biases and make conscious choices between competing values and imperatives.

Horowitz identifies six possible aims of an electoral system, some of which directly conflict with one another. First is the increasingly com-mon normative goal of proportionality—trying to achieve a distribution of seats among parties in parliament that matches as closely as possible the relative party shares of the vote. It is in this respect that observers most often speak of “fairness.” Yet, Horowitz stresses, fairness in this sense “is only one among several goals, and it may not be the most important goal.” The most proportional outcomes are produced by one or another form of proportional representation (PR), and generally the fewer and larger the districts (up to the maximum point of the single nationwide district, as in Israel and the Netherlands), the more propor-tional the results. A second goal, however, is accountability of elected representatives to constituents, and the larger the districts, the weaker the direct ties between representatives and any kind of clear territorial constituency. Moreover, as Horowitz notes, PR (especially when using national lists) often “reposes great power in party leaders” to determine the rank order of the candidates on the list, further undermining ac-countability to voters. Horowitz’s third goal is durable governments, ones that can persist in office for the elected term of parliament (if there is a parliamentary system). This goal, too, may come in conflict with proportionality, as the more purely proportional an electoral system is, the more fragmented the parliament (into multiple parties) and thus the more difficult it is to forge and sustain a ruling coalition.

A fourth goal is facilitating the victory of the “Condorcet winner” (the one candidate who would win every paired contest). This objective is poorly served by PR, but well advanced by the alternative vote (AV), which requires voters to rank the candidates in single-member districts, and which then redistributes the second (and if necessary, third and then lower-order preferences) of the weakest candidates until one can-didate receives a majority. By giving politicians incentives to craft wider appeals beyond their political (e.g., ethnic or ideological) base, AV favors moderates. Thus, as Horowitz has written elsewhere, and as Ben Reilly discusses at length in his contribution to this volume, AV has intriguing possibilities for serving a fifth objective, promoting in-terethnic (or interreligious) conciliation. It can do so by inducing can-didates and parties to moderate their appeals and reach out for the second-place preferences of other parties and ethnic groups. This kind of “vote pooling” logic can only work if there are a large number of districts in which no one party or group has a majority, and in any case, AV is a majoritarian system which can yield quite disproportional out-comes. Other mechanisms, such as the Lebanese system of ethnically reserved seats (but with all voters voting for all seats), can similarly or even more powerfully induce crossethnic alliances, but with other draw-backs. The sixth (related) goal is minority officeholding, which can be served not only by PR but also by more explicit and extreme devices, though at the cost of “providing incentives for candidates to take ex-treme positions in order to heighten” their appeal to their identity base.

These are difficult choices, encompassing powerful normative and political considerations, and yet the consequences of adopting a par-ticular electoral system are not always straightforward, immediate, and predictable. The legacy of preexisting cleavages and incentives per-sists, at least for a while. Thus, Horowitz warns, “the full effects of a new system can only be gauged after two or three elections, when politicians and voters have adapted to its incentives.”

As Richard Soudriette and Andrew Ellis show in Chapter 2, more than a few countries have changed their electoral systems in recent years. About one in eight have changed the basic nature of the system, and many others have made adjustments within the same system. Ap-parently, when they choose or redesign an electoral system, more and more countries are concerned these days with fairness and inclusive-ness, including the representation of women and minorities. The most common type of major shift is from a majoritarian rule, such as the single-member-district, “first-past-the-post” (FPTP) system like that in the U.K. or the United States, either to PR or to a mixed system combin-ing some single-member districts with some seats filled by PR lists. As they show in Table 2 (p. 21), since 1993 a dozen countries have made precisely that shift, while three have shifted from PR to a mixed system. Party-list PR is still used by more countries than any other type of system (over a third of countries that hold competitive elections employ it). But the mixed system is growing in appeal, in part because it provides for some degree of proportionality while tempering PR’s ten-dency to party fragmentation, and in part because it enables countries to get some of the fairness and inclusion of PR along with some of the direct accountability to territorial constituencies that comes most natu-rally with single-member districts. Mixed systems vary considerably, however, particularly in terms of whether they follow a truly mixed logic or instead distribute seats to parties using the proportional rule. Most mixed systems are truly mixed or “parallel systems,” in that one block of seats (in many countries it is half or sixty percent) is filled strictly through FPTP and the other through PR. A few countries, how-ever, including Italy, New Zealand, Mexico, and Bolivia, have more or less emulated the German “mixed-member proportional” model. Under this system, the party list vote determines proportionally the overall share of seats for each party (subject to whatever threshold may be im-posed, such as the 5 percent minimum in Germany). On top of seats it has won in the individual districts, each party then wins a quota of seats from its PR lists sufficient to give it a national allocation of seats pro-portional to its vote. Soudriette and Ellis provide in Table 1 a compre-hensive inventory of the all the electoral systems used in all the world’s countries, including for directly elected presidents, most of whom are chosen by the two-round, run-off system when no candidate receives a majority of the vote in the first round.

In chapter 3, the Australian political scientist, Ben Reilly, builds on the seminal work of Donald Horowitz to examine the utility of preferen-tial electoral systems as a tool of ethnic conflict management. Like Horowitz, Reilly’s concern is to examine to what extent electoral sys-tems can “encourage cooperation and accommodation among rival groups, and therefore work to reduce the salience of ethnicity.” Like Horowitz as well, Reilly favors vote-pooling mechanisms, which “make politicians reciprocally dependent on votes from groups other than their own.” Such systems give an advantage to moderate candidates who reach across the divides of party and ethnicity to appeal for the second and lower-order preferences of the voters, and who thereby exhibit a “capacity to represent groups other than their own.” In Reilly’s termi-nology, these are “centripetal systems” in that they press parties and candidates toward the accommodating center. Both Horowitz and Reilly are drawn to the alternative vote as a potentially powerful mechanism in this regard. Each author also recognizes the value of the single trans-ferable vote (STV), in which voters rank a larger number of candidates in multimember districts. STV functions like a proportional system. Thus, candidates do not need to win a majority of the vote, but rather a minimum quota (which, as Horowitz notes, would be one-fifth plus one in a four-seat district). The advantage of STV is that it is a fairer, more proportional system than AV, better able to represent minorities. The disadvantage, Horowitz stresses, is that the low threshold for election in a district “provides few incentives to interparty agreements to transfer votes,” and thus generates “weaker incentives to compromise” than AV, under which a candidate must ultimately gather enough lower-order preferences to win a majority of votes in the district.

Weighing the experiences with preferential voting in Estonia, Aus-tralia, Fiji, and Papua New Guinea, Reilly concludes that these vote-pooling systems can foster ethnic accommodation and political movement toward the center. There are several conditions that must hold, however, for these effects to materialize. First, there must exist among both political leaders and the general electorate a “core group of moderates” ready to respond to the incentives generated by preferential voting. Second, because it takes time before political leaders and sup-porters grasp and adapt to the new incentives, system continuity is important; once a new electoral system is put in place, it has to be allowed to operate consistently through several elections. And third, groups must be structured and distributed across electoral districts so that there is real competition and uncertainty, and thus the need for different groups to pool votes in a number of different districts. If, as in Fiji, districts “are drawn in such a way as to become the exclusive pre-serve of one ethnic group or the other,” then the crossethnic logic of preferential voting will never take hold.

The vote-pooling model of using the electoral system to encourage crossethnic alliances and appeals stands in pointed contrast to Arend Lijphart’s consociational or “power-sharing” model of democracy in divided societies. The logic of this system, as Lijphart reprises it in chapter 4 of this volume, is to accept ethnic divisions as a given and then devise mechanisms to enable ethnic political parties to share power and resources in proportion to their votes and their numbers in the popu-lation. The two foundations of the model, Lijphart stresses, are power sharing—“the participation of representatives of all significant commu-nal groups in political decision making, especially at the executive level”—and group autonomy, which gives different groups (either on a territorial or cultural basis) control over their own social practices, edu-cation, use of language, and so on. Proportional representation is the cardinal mechanism for fostering power sharing but in the full consocia-tional version, there are three other mechanisms: 1) a grand coalition of all or virtually all the parties, who divide up ministries and other execu-tive roles in rough proportion to their political strength in parliament and society; 2) a mutual veto over key government policies (which must therefore be reached by consensus); and 3) group autonomy through federalism or nonterritorial means.1 Horowitz and others criticize the consociational model for putting so much faith in the willingness and ability of leaders to reach accommodation after elections have entrenched and polarized ethnic solidarities, instead of seeking to bridge divisions and forge conciliation in advance, through preelectoral coalitions.2 In this essay, Lijphart dismisses critics of his consociationalism, including Horowitz, as having no realistic alternative to the power-sharing model. The alternative vote, he cautions, is a majoritarian system that can leave ethnic minorities severely underrepresented. Using the example of con-temporary Iraq, Lijphart warns that any electoral system that does not ensure the fair and full representation of each group risks the group’s alienation and rejection of the system; thus he strongly favors PR.

Power sharing, Lijphart insists, is not a “one-size-fits-all” model, but rather can be adapted in a variety of ways to different country circum-stances. Still, he feels some institutional designs are generally right for all divided societies (indeed, for all countries). The best electoral sys-tem, he insists, is a fully proportional one (either PR or the German-style mixed-member proportional system), using closed or nearly closed party lists; this encourages the formation of strong, coherent parties. Some break on pure proportionality can be warranted, through the cre-ation of multimember districts that are not too large in size or through the establishment of a modest electoral threshold. Lijphart particularly likes two-tiered PR systems such as the Danish one, where the majority of seats are elected from moderate-sized multimember districts, and then a smaller number of compensatory seats are allocated to ensure greater overall national proportionality between the parties’ shares of votes and parliamentary seats. As in his voluminous other writings, Lijphart urges here that PR be joined to a parliamentary system of government, which facilitates coalition formation with broad power sharing and avoids the zero-sum character of presidential elections. Moreover, he urges that in parliamentary systems, the president should be a purely ceremo-nial head of state, and selected by parliament (ideally through some requirement for broad consensus) and not by direct vote of the people. Where groups are territorially concentrated, federalism is “an excellent way” to provide for their autonomy. In this respect, he recommends a weak upper house of parliament and a system of relatively small and homogeneous units to ensure autonomy and avoid domination by a few big states.

The role of electoral rules under federalism forms the concern of R. Kent Weaver’s analysis in chapter 5. How, Weaver asks, do federalism and the electoral system interact with one another? Are federal systems somehow distinctive in what they require from an electoral system? Not very distinctive, he concludes: “Federalism is compatible with a vari-ety of electoral rules at both the national and territorial levels.” And thus federal systems vary widely in the number and character of parties that predominate. Some federal systems use PR and some the single-member-district plurality (or first-past-the-post) system. One, Australia, uses AV for the lower house and STV for the upper. Most use the same electoral system for subnational legislatures that they do for the national one. Where there are strong and distinct subnational identities, majoritarian electoral systems like FPTP tend to produce much more homogeneous territorial concentrations of party strength. Thus, in marked contrast to PR, FPTP electoral rules “appear significantly more likely to create a high risk of territorial exclusion from the governing-party caucus at the national level in parliamentary systems.” This, in turn, may feed a region’s sense of political alienation. Majoritarian electoral rules may keep small secessionist parties from winning many seats, but once such “antisystem” parties reach a critical plurality of 35 percent or more, FPTP may in fact strengthen them. These mixed impli-cations leave Weaver unconvinced that there is any one best type of electoral system for governing democratic federations.

The Debate Over Proportional Representation

For several decades now, debates about institutional design in de-mocracies have heavily revolved around the fundamental choice between majoritarian and power-sharing logics, and thus between FPTP and PR. In 1991, the Journal of Democracy published a seminal piece by Arend Lijphart arguing for the general superiority of proportional representation, especially in new democracies and deeply divided soci-eties. Lijphart’s bold essay inspired a vigorous debate between him and several critics of PR. In Section II we reproduce these articles, which continue to be widely cited. In chapter 6, Lijphart rejects majoritarian democracy in favor of the consensus model, which “tries to limit, di-vide, separate, and share power,” and features multiple parties, coalition governments, and more equal executive-legislative power relations. By virtually ensuring a multiparty system, proportional representation (com-bined with parliamentarism) is a key device for structuring democracy in this consensual way. But in addition to ensuring minority representa-tion and so better managing ethnic conflict, PR is preferable, Lijphart maintains, because it is intrinsically more democratic. Moreover, his data from the Western democracies suggest that parliamentary-PR sys-tems have a higher quality of democracy, including higher rates of voter turnout, without the diminished governability and economic perfor-mance that many critics allege to be the “cost” of PR systems. Moderate PR systems, like those in Germany and Sweden, that give rise to a mod-erate number of parliamentary parties seem to offer the best combination of power sharing and governability for new democracies.

Lijphart’s arguments are disputed from several perspectives. As our focus here is on electoral systems, we do not include the debate on parliamentary versus presidential government, but it is nevertheless worth noting here Donald Horowitz’s argument that presidentialism can be a tool of ethnic conflict management by providing a transethnic symbol for the country and by generating incentives to ethnic modera-tion and compromise to the extent that vote-pooling rules are utilized. (These include the preferential vote and Nigeria’s requirement that a presidential candidate must win at least 25 percent of the vote in at least two-thirds of the states to be elected on the first ballot).3

In chapters 7 and 8 Guy Lardeyret and Quentin Quade vigorously challenge Lijphart’s case for PR and argue strongly for the advantages of majoritarian democracy. Indeed, Lardeyret favors the most majoritarian system possible, which (as Lijphart indicates) is not presidential but rather parliamentary government combined with the single-member-district plurality method of election—the Westminster system. Because such plurality elections for the legislature tend to give rise to a two-party or two-party-dominant system, they create strong parliamentary governments, fusing executive and legislative power, and free of the need for coalitions. Lardeyret believes the greatest danger lies in the fragmentation associated with parliamentary-PR systems, like the Fourth Republic that faltered in his native France. Quentin Quade also cites this example, as well as pre-Mussolini Italy and Weimar Germany, to show how PR can foster not conciliation and compromise but fragmen-tation, extremism, and governmental paralysis. Like most critics of the consensus model, Quade focuses attention on the intrinsically greater fragility of coalition government.

Precisely because it gives such wide scope to the representation of “minorities,” and thereby encourages polarization and party fragmen-tation, PR, Lardeyret maintains, “is dangerous for countries faced with ethnic or cultural divisions.” By contrast, “Parties in plurality systems tend to be moderate because most votes are to be gained among the undecided voters of the center.” Ethnic moderation will be greatest when members of the same ethnic group must compete against one an-other in single-member districts, along crosscutting political and ideological lines of cleavage. Lardeyret prefers plurality electoral sys-tems not only for the stronger, more stable and decisive governments they produce. He (like Quade) also believes that plurality electoral sys-tems are “more democratic as well as more efficient,” because, unlike PR, they lock out extremist parties, marginalize small parties, and give the choice of who will govern to the voters, rather than to elites of numerous parties negotiating in secret after the election, sometimes for weeks or even months. Quade writes: “Plurality voting encourages the competing parties to adopt a majority-forming attitude . . . to be moder-ate, to seek conciliation . . . in short, to do before the election, in the public view, the very tasks that Lijphart applauds PR systems for doing after the election.”

The institutional choice of PR (with parliamentary government) or plurality elections (whether in a presidential or parliamentary system) involves more than the empirical and analytical debate between Lijphart and his critics. At play here as well is a tension between competing values. There is a certain intrinsic contradiction in democracies between representativeness and governability. PR systems risk some sacrifice in decisive governance and clear alternation of majorities in order to maxi-mize representativeness. Opponents of PR put a higher priority on governability than on directly representing in the legislature the many divisions in society.4 To be fair however, few prominent students of democracy would advocate the kind of extreme PR—with virtually no minimum threshold of the vote necessary for entry into parliament—that existed for some time in Italy and Israel. In fact, during the 1990s Italy switched to a much more majoritarian electoral system, while Israel has gradually raised its electoral threshold to 2 percent (in its most recent, March 2006 elections). As Lijphart emphasizes, there are many types and degrees of PR, and moderate PR, with thresholds like the 5 percent minimum in Germany, tend to give rise to only a moderate number of parties. In responding to Lardeyret and Quade in Chapter 9, Lijphart not only defends his comparative analysis of the performance of PR and majoritarian systems. He also reiterates the normative case for PR—“that disproportional election results are inherently unfair and undemo-cratic”—noting that none of Britain’s postwar governing parties carried a majority of the vote (a fact that has held true for all of India’s one-party governments as well).

Is there, then, one type of electoral system that is right for all coun-tries and historical moments? In his reflection on the debate over PR, in chapter 10, Ken Gladdish insists not. There is in practice a wide range of electoral-system formulas. Each one, Gladdish stresses, has costs as well as advantages, and the type of system that will best serve democracy in a given country depends on the nature of its social cleavages and pat-terns of political mobilization. How a country resolves the tension between representativeness and governability is a value choice that must be informed by the country’s own particular history and culture. PR not only facilitates legislative fragmentation, and thus heightens the difficulty of forming and maintaining governments, it may also produce bizarre coalitions of expedience that cannot govern coherently or minority cabinets that cannot govern with any confidence. Further, it has ambiguous consequences for representation, since party-list sys-tems with large multi-member districts (or in the extreme, as in Israel and the Netherlands, no districts at all) sacrifice direct linkage between voters and representatives. This lack of accountability of MPs to spe-cific geographical constituencies became a major issue in the debate over a permanent electoral system for South Africa.5 Yet, particularly when there are more than two significant parties, as in Britain and India, plurality systems may give rise to severely disproportional results (the Labor and Conservative parties in Britain have repeatedly won thump-ing majorities in parliament with barely 40 percent of the vote, and often less). Moreover, Gladdish notes, plurality systems also tend to breed voter apathy in districts (often the vast majority) where the con-test for the seat is a foregone conclusion. Crafting an electoral system thus involves hard choices that cannot be made theoretically or in the abstract, but must be done on the basis “of practicality and aptness relative to national circumstance.”

Learning from Country and Regional Experience

If Gladdish is right that the best type of electoral system for a country depends on its social cleavages and political and historical realities, then we need to pay particular attention to the wide variation in country and regional experiences with electoral systems. This we do in Section III. We begin this review with a debate between two leading political science scholars of Africa, Andrew Reynolds and Joel Barkan, on whether PR is the best electoral system for African democracies. Reynolds (who wrote his Ph.D. dissertation under Arend Lijphart) argues enthusiasti-cally for the combination of PR and parliamentary rule in the African context, echoing Lijphart’s concern with promoting broad inclusion and power sharing and avoiding zero-sum politics. Reviewing the political experience of Malawi, Zambia, Namibia, and South Africa in the early 1990s, following the wave of transitions to democracy, he shows that none of Lardeyret’s criticisms of PR hold up very well in southern Af-rica. Rather than foster extremism by including fringe parties in parlia-ment, PR tends “to coopt extremists, giving them an incentive to press their case with ballots rather than bullets.” This was a particularly strik-ing consequence of the low electoral threshold that allowed white reac-tionary and black militant parties, as well as small liberal parties, to enter parliament in South Africa. Rather than promoting ethnically ex-clusive parties, PR has enabled South African parties to assemble dis-tinctly multiracial lists of candidates, to a degree that would simply have been impossible under FPTP. Of course, given the electoral domi-nance of the African National Congress (ANC), it is easy to use the South African experience to rebut Lardeyret’s warnings about the instability and weakness of governments under PR and the undue leverage of ex-tremist parties. What worries Reynolds, however, is the PR system’s fail-ure to enable voters to hold representatives accountable (or even to know who their representatives are) when districts are as large as they are in South Africa. This is one reason why he strongly recommends PR in small to medium-sized districts (of, say, 5 to 12 members), and with open lists so that voters can choose individual candidates from the parties.

For ethnic conflict management, there is an even more compelling reason to favor PR in Africa, Reynolds argues. Like Weaver, he is con-cerned about the polarizing consequences under FPTP of electoral outcomes that leave individual parties totally dominant in particular regions. A classic example is the first multiparty elections in Malawi in 1994, where the three regionally based parties each won 75 to 100 percent of the seats in their territorial strongholds, despite the presence of significant opposition. In each case, the majoritarian character of FPTP left minority sentiment in each region virtually unrepresented, whereas with PR each region would have had some political presence of parties that were dominant in the other regions. Such a geographic mix-ing of representation helps to promote ethnic inclusion and cooperation within parties as well as between them, and thus to inhibit polarization. In addition, Reynolds finds that PR has produced a more diverse and representative South African parliament, in terms of race, ethnicity, and gender, than would likely have been achieved under FPTP.

These arguments leave Joel Barkan unmoved. In chapter 12, Barkan argues that FPTP—with its single-member districts and the close, or-ganic ties that they forge between legislators and territorial constituen-cies—much better fits the circumstances and needs of a still predomi-nantly rural Africa. In these countries, most people expect their representatives to help deliver on their basic needs for clean water, schools, roads, and health care, and voters evaluate parties and repre-sentatives “in terms of their potential for, or past record of, constituency service.” PR, with its multimember (and often very large multimember) districts, is intrinsically inferior in this regard, to the point where (as in Namibia) members of parliament may “rarely make an appearance in rural areas.” Indeed, this alarming representational gap prompted the ANC to assign its MPs arbitrarily to “represent” designated areas (with dubious results). Moreover, Barkan maintains, PR does not yield nota-bly fairer (more proportional) outcomes in agrarian Africa. Rather, his analysis of specific elections suggests that in rural Africa, where people cluster territorially by ethnicity, FPTP produces allocations of seats to parties that are about as proportional as under PR, especially when all districts contain the same number of voters.

In rebutting Barkan, Reynolds makes two essential points in chapter 13. First, when one compares PR and FPTP systematically in Africa (rather than focusing on anomalous individual cases like Malawi), the claim to equal fairness collapses. Both within Southern Africa and in Africa as a whole, PR electoral systems yield significantly more propor-tional outcomes; “when it comes to translating seats into votes, PR is not just marginally superior [to FPTP] but substantially so.” Second, irrespective of overall proportionality, one must worry about the geo-graphic patterning of party strength. One key to ethnic conflict management and democratic consolidation in Africa, Reynolds reiter-ates, is to avoid the emergence of one-party states at the subnational (regional) level and to ensure a critical mass of competitive seats. Even where FPTP produces similar outcomes nationally, it does so by making particular parties nearly invulnerable within their areas of ethnic strength. Reynolds wants to use PR in moderate-sized multimember districts to generate crosscutting cleavages, so that a party from one ethnic base will still have an incentive to campaign in the “other” areas, and oppo-sition parties subnationally will be kept in the electoral game, rather than permanently frozen out and alienated from the democratic process. In these respects, Reynolds uses a variation of Lijphart’s method (PR) to achieve one of Horowitz’s goals (inducing political parties to appeal across ethnic lines). As he has done in other published work, Reynolds recommends for South Africa the Scandinavian model of a two-tiered PR system, in which the disproportionality that may creep in from hav-ing relatively small multimember districts can be corrected with a second distribution of seats that brings each party’s share of parliamentary seats into balance with its share of the national vote.

In chapter 14, the first of two essays focusing on Latin America, Jeffrey Cason assesses the impact of electoral reform on democratic stability in Uruguay. Uruguay has had over the decades one of the world’s most unique and dysfunctional electoral systems, the “double simulta-neous vote” (DSV). A form of proportional representation, DSV allows multiple party factions to contest for votes in the general election. In parliamentary elections, this has fostered a breathtaking fragmentation of party lists (in 1994, there were 641 lists from the three main parties, in a country of barely two million voters), because there has been no cost to the party for such fragmentation. Parties get a proportion of seats in parliament equivalent to the total vote share for all their factions (and factions in turn get a share of their party’s seats equal to their share of the party vote). The result has been extremely weak, divided parties and severe clientelism. In presidential elections, the effects were even more perverse, as the winner of the presidential election was the candi-date with the most votes from the party that received overall for its multiple candidates more votes than other party. This meant that presi-dents could be (and were) elected with less than a quarter of the vote personally, and without even being the most popular candidate. Inevi-tably, “These relatively thin mandates called into question the legiti-macy of the eventual presidential winner.”

Alarmed by the rise of the Left, the two established Uruguayan par-ties engineered a constitutional reform in 1996 to require a single presidential candidate for each party, chosen in a party primary, as well as a second-round runoff election in the event that no presidential can-didate won a majority on the first round. Although the reform only slightly modified and rationalized the DSV system for the parliamen-tary elections (forbidding party factions to combine their votes for different lists after the election to win seats), it did have an immediate and striking effect on the next presidential election in 1999. In particu-lar, the new majority rule for presidential election forced the Left alliance to move toward the political center and focused the second round of the presidential election much more on policy proposals. This “made the Left appear less threatening” and made Uruguayan democracy less “risk-prone.” Even the modest rationalization of DSV in parliamentary elections reduced the number of parliamentary lists and gave parties an incentive to unify. And by requiring broader coalitions for electoral success, the reforms made it easier for elected leaders to govern.

In chapter 15, Richard Snyder and David Samuels examine a distinc-tive feature of Latin American electoral systems, their exceptionally high levels of malapportionment—the “wide discrepancy between the share of legislative seats and the share of population held by electoral dis-tricts.” Federal systems (of which there are several in Latin America, in-cluding Mexico and Brazil) often deliberately overrepresent some na-tional units in the upper chamber of a parliament as part of the power-sharing federal “bargain.” What is unique about many Latin Ameri-can democracies, however, Snyder and Samuels show, is the degree of malapportionment of lower chamber seats—on average, this distortion is twice as severe in Latin American democracies as in the advanced indus-trial democracies. Moreover, they show, the overrepresentation of some districts and underrepresentation of others are by no means random with respect to social and political cleavages. Where malapportionment oc-curs in Latin America, it invariably gives national legislatures “a distinct rural and conservative bias.” This unfairness, which is “often hidden from public view,” not only handicaps urban and leftist constituencies in the battle over policies and resources, but also yields several other pre-dictable consequences across countries and over time. It tends to widen rifts between presidents, who are elected from the entire country with all votes weighted equally, and assemblies that are disproportionately be-holden to conservative and clientelistic rural bosses. It leaves presidents vulnerable to pressure from these rural interests, including the need to channel undue patronage to them (and thereby neglect wider and more substantively compelling urban needs). And it may compel national lead-ers to tolerate authoritarian enclaves in the countryside, because the presi-dents need the legislative support of these overrepresented rural bosses.

Snyder and Samuels review three possible innovations to address the problem. Judicial oversight of reapportionment, they argue, is less prom-ising than in the United States because Latin American judiciaries are often much less independent and capable than North American ones. Based on the success of Mexico’s independent Federal Electoral Insti-tute (IFE, created in 1996), they see more promise in the creation of a “neutral, nonpartisan electoral commission that is legally obligated to reduce lower-chamber malapportionment.” A third possible avenue is change in the electoral law itself, so that correcting malapportionment becomes part of a wider package of reforms to improve the quality of democracy. Of course, one possible reform is to do away with electoral districts altogether in favor of a single nationwide list (as in Peru), but this involves big costs in parliamentary accountability to the citizens.

Instead, they recommend a mixed electoral system, with enough seats allocated through PR (operating in a nationwide district) to temper any effects of malapportionment among single-member districts.

Political scientists are naturally drawn to the promise of electoral reform, but as Emanuele Ottolenghi warns in chapter 16, the cure can be worse than the original illness. Disgusted with the political paralysis and extortion that increasingly plagued the coalition-formation process in Israel’s highly fragmented party system in the 1980s and early 1990s, and unable to achieve a dramatic alteration in an electoral system that fostered fragmentation, reformers pressed for a radical change in the very nature of parliamentary government. Under the plan that was even-tually adopted in 1992, the prime minister was no longer to emerge from parliamentary bargaining (which gave small religious parties in between Labor and Likud extravagant power to make demands), but was to be directly elected by the people, in a separate ballot simultaneous with the traditional vote Israelis cast for a party list in their highly propor-tional system. This change, reformers confidently predicted, would not only give the directly elected prime minister the moral authority to form a government more quickly and to govern more decisively, it would also give him more seats, as he would sweep like-minded representa-tives into parliament on his “coattails.”

In reality, as Ottolenghi shows, precisely the reverse occurred. Freed from having to use their parliamentary vote to signal their choice for prime minister, Israeli voters engaged in ticket splitting with a ven-geance. While prime ministerial candidates were able to win majorities on the first ballot, the two major parties, Labor and Likud, hemorrhaged votes to smaller, more ideological, or single-issue parties, and the par-liament became still more fragmented (going from 10 parties in 1992 to 15 in 1999). As a result, forming and maintaining majority coalitions became even more difficult, and “the executive remained hamstrung in its ability to forge ahead with coherent policies on controversial is-sues.” So unsastisfying and indeed counterproductive were the conse-quences of the reform that the entire innovation was reversed in 2001 in favor of the traditional parliamentary system (but now with the valu-able innovation of the “constructive vote of no confidence”).6 The les-son, Ottolenghi underscores, is that such constitutional innovation cannot be a substitute for fundamental electoral reform. The latter pro-vides the only sure means of reducing parliamentary fragmentation, thereby strengthening the capacity of a prime minister to form and sus-tain coalitions and thus to govern effectively.

As in Israel, so in Japan and Taiwan citizens and civic groups became increasingly frustrated with features of their electoral system that seemed to foster the most unsavory aspects of politics. Here the culprit was the system of the single nontransferable vote (SNTV), a rare system used mainly in those two countries (and for a briefer period in Korea). Under SNTV, candidates contend for seats in multimember districts, but the voter is allowed only a single vote, and “excess” votes may not be transferred to a party’s other candidates in these districts. The top vote getters are elected by the plurality rule with whatever number of votes they receive. The system thus places an intense premium on party orga-nization, which enables parties to properly estimate and distribute their voter support in a district. If a party nominates too many candidates, its vote may be spread too thinly, costing it seats. Yet if it nominates too few, it can also suffer. Money and organization become vitally impor-tant (to the point where political parties in Taiwan have even been known to instruct their voters to support different party candidates based on their birthdays, so that everyone born in January through March supports the party’s first candidate in the district, those born April to June vote for the second candidate, and so on). Rarely does such coordi-nation work flawlessly. A perverse feature of SNTV is that candidates of the same party are forced to compete against one another as well as against the opposition party candidates. Consequently, parties are highly factionalized and election campaigns are complex and hugely expen-sive, inducing politicians either to unleash huge flows of patronage to the district or to fashion extreme appeals to the ideological faithful.

In chapter 17, Jih-wen Lin explains how long-sought electoral re-form from SNTV to mixed but more majoritarian systems was finally achieved in Japan (1994) and then Taiwan (2005). In each case, the public became disillusioned with the corruption and inefficiency of the political system, but this broad disaffection was not sufficient to force change. Rather, Lin shows, change could only happen when three fac-tors converged. First, “public disappointment eroded the governing party’s electoral strength and finally deprived the party of its dominant status,” unfreezing a political situation that had seen insecure members of the ruling party blocking any reform. Second, a critical mass of influ-ential politicians discovered that they had a personal interest in reform. Specifically, ruling party legislators whose electoral success was owed “to personal image rather than particularistic connections” realized that their fortunes could be better served by FPTP. Third, electoral reform had to become part of a larger, log-rolling deal in order to attract the support of a majority of legislators (especially from smaller parties threat-ened by the switch to a more majoritarian system). Although these conditions are derived from the political experience of Japan and Tai-wan in switching from SNTV, they would seem to have relevance to other countries seeking electoral reform, and to explain why reform is so difficult to achieve: Even if society can be rallied behind the cause, legislators do not like to abandon the system that put them in parlia-ment. Yet Lin’s study underscores why electoral reform can matter. In the four elections following Japan’s change to a more majoritarian elec-toral system, the party system has shown striking signs of consolidation toward a two-party system, and a similar trend is likely to unfold even more rapidly in Taiwan. While the legacy of patron-client networks still remains potent in Japan, Lin predicts that “as the two-party system takes shape, the mode of campaigning will likely come to focus more on policy issues and debate.”

Ironically, just as Taiwan was abandoning SNTV, Afghanistan was adopting it for its 2005 legislative elections. Andrew Reynolds explains this “curious case” in chapter 18. As it happened, international experts like Reynolds himself had recommended that Afghanistan use closed-list PR in multimember districts based on the country’s 34 historic prov-inces. This would give voters a fairly simple and accessible choice (iden-tifiable by a symbol for each list) and would foster more proportional outcomes. Yet while there was a consensus against resurrecting the old FPTP system (inherited from British colonial rule but not used in 40 years), there was resistance to PR among Afghans who did not like or trust political parties, and from President Hamid Karzai himself (already elected the previous year), who wanted to allow Afghans to vote for individual candidates. International experts anticipated most of the prob-lems that resulted from the use of SNTV. While the system apparently produced a fairly proportional outcome in ethnic terms, it yielded a chaotic party situation, with 33 political parties or groups in the assem-bly, most lacking any real coherence or a critical mass of assembly mem-bers. With the incentives to personal campaigning, powerful individu-als were elected, including a substantial number of warlords, criminal gang leaders, religious fundamentalists, and drug traffickers. And given the large size of many of the districts, most candidates were elected with only small percentages of the vote. Voters were apparently confused by the profusion of candidates and the long list of names on the ballot, leading to a low voter turnout and a high incidence of spoiled ballots. Though President Karzai thought the system would work to his benefit, it instead weakened him, as experts had warned, leaving him with sup-port from less than a third of the assembly. The result is a fragmented legislature hostile to Karzai, featuring “clientelism and shifting alli-ances” and mortgaged to “men with tainted pasts.” Moreover, if other country experiences are any guide, the system will be difficult to change. Yet Reynolds warns, “If SNTV is used in subsequent elections, the frag-mentation and parochialism of the legislature will increase, and politics in general will remain detached from the masses.”

As Adeed Dawisha and Larry Diamond indicate in chapter 19, inter-national experts on electoral systems fared little better in Iraq, where (more or less as in Afghanistan) they had recommended a system of party-list PR in multimember districts, based on the country’s 18 prov-inces. The alternatives to PR were discarded one by one because of their substantive or technical unsuitability. FPTP was too majoritarian, and in any case drawing single-member districts was not possible in Iraq’s fragile transitional circumstances. AV was infeasible for the same rea-son, and probably too complicated for first-time voters. STV was also ruled out for its complexity, and SNTV for its known perversities (as articulated above). By contrast, PR fit with the powersharing logic of emergent political practice and institutional design in Iraq, appealed to party leaders who wanted to consolidate their control, and lent itself well to the implementation of a 25 percent quota for female representa-tion, as established in the interim constitution. Yet the question remained of how PR should be structured in Iraq. While several Iraqi and interna-tional officials had advanced proposals in 2004 for a two-tiered system on the Scandinavian model, with provincial lists and a compensatory adjustment from national lists to achieve greater overall proportional-ity, the UN electoral mission chose, for its simplicity and ease of administration, PR within a single nationwide district.

The difference between the two types of PR—district-based and na-tional-list—may seem subtle but in fact carried huge consequences. The district-based system offered some hope of empowering local ac-tors and identities, and thereby putting a break on the polarization of politics around national ethnic and sectarian identities. Then too, it might have generated ties between elected representatives and geo-graphical constituencies, “making for greater local accountability.” In the absence of any districts, or even any awareness of who the candi-dates were, “the January 2005 parliamentary elections became almost purely a national-identity referendum, untempered by any local com-ponent or flavor.” In addition, the lack of any minimum floor of representation for geographical areas contributed to the Sunni boycott of the elections. Each of these developments distorted and polarized the political landscape in lasting ways. Thus, although for the Decem-ber 2005 election Iraq implemented essentially the two-tiered, district-based PR system that had been recommended back in 2004, the polarization of party politics around ethnic and sectarian identities had by then become deeply entrenched. During the campaign “each alli-ance focused its energy on cementing support among its own base, while doggedly obstructing intrusions by other alliances into its home areas.” Violence intensified, Iraqis again voted overwhelmingly along ethnic and religious lines, and the only secular, nonethnic list was crushed. As Dawisha and Diamond recount it, the Iraqi case tells a sobering tale of the limits of electoral-system engineering, once ethnic polarization and violence take hold.

The Afghan and Iraqi cases no doubt appear discouraging to elec-toral-system experts who believe each country might have fared better in its transition if their advice had been taken. Yet the broad trajectory of electoral systems in the world is much more hopeful. Over the last few decades, knowledge about the consequences of different electoral sys-tems has accumulated and been disseminated. In general, the direction of change is away from the extremes of purely majoritarian and purely proportional systems, toward either mixed systems or proportional ones with at least some modest inducement to party aggregation. Preferential systems have yet to be tried on a large scale, but where they have been employed, their results have been, on the whole, promising. The more successful models, such as the German mixed-member proportional sys-tem and the Scandinavian two-tiered PR, have been gaining adherents, and the most dysfunctional systems, such as SNTV and the block vote (a hyper-majoritarian system based on multimember districts) have been losing out. Gradually, a kind of process of natural selection is unfold-ing, as democracies experiment and adapt, discarding what does not work, or at least what is not suitable for particular societies. Over time, we can expect to see more changes, in which countries seek to find a better balance between inclusion and governability, and between pro-portionality and accountability. There will no doubt be many more pain-ful false starts and lost opportunities, but democracies in general are moving toward fairer, more rational, and more viable electoral systems.

NOTES

1. The early articulations of the model are Arend Lijphart, “Consociational Democracy,” World Politics 21(January 1969): 207–25, and Democracy in Plural Societies: A Comparative Exploration (New Haven: Yale University Press, 1977). A more generic and comprehensive consideration of the model can be found in Arend Lijphart, Patterns of Democracy: Government Forms and Performance in 36 Countries (New Haven: Yale University Press, 1999).

2. Among the places where Donald Horowitz has critiqued the consociational model are Ethnic Groups in Conflict (Berkeley: University of California Press, 1985), 568–76; A Democratic South Africa? Constitutional Engineering in a Di-vided Society (Berkeley: University of California Press, 1991), 137–45, 167–71; and “Constitutional Design: Proposals versus Process,” in Andrew Reynolds, ed.,

The Architecture of Democracy: Constitutional Design, Conflict Management, and Democracy (Oxford: Oxford University Press, 2002), 20–23.

3. Donald L. Horowitz, “Comparing Democratic Systems,” Journal of Democ-racy 1 (Fall 1990): 73–79. The larger debate over parliamentary versus presidential government is reproduced in Larry Diamond and Marc F. Plattner, eds., The Glo-bal Resurgence of Democracy, 2nd ed. (Baltimore: Johns Hopkins University Press, 1996), 124–61.

4. Larry Diamond, “Three Paradoxes of Democracy,” Journal of Democracy 1 (July 1990): 48–60.

5. Vincent Maphai, “The New South Africa: A Season for Power-Sharing,” Journal of Democracy 7 (January 1996): 67–81.

6. This prohibits a vote of no confidence from bringing down a government unless it also puts forward an alternative coalition with a majority of votes to form a new government. The provision thus diminishes speculative efforts to bring down a government in order simply to get bargaining leverage over the formation of a new one.