25
Conclusion: Are Mixed-Member Systems the Best of Both Worlds?
Matthew Soberg Shugart and Martin P. Wattenberg
This book began with the question of whether mixed-member electoral systems might prove to be the electoral reform of the twenty-first century. Whether or not they will depends to a large degree on how their performance is viewed in the numerous countries that have adopted them in recent years. In particular, it will be crucial to see if their promise of delivering the best of both the majoritarian and proportional worlds of electoral systems is realized. The prospects for the spread of mixed-member systems also depend on the presence elsewhere of the basic conditions that brought about the recent proliferation of mixed-member systems. Thus, this final chapter reviews the experiences of the several existing mixed-member systems in an effort to assess the likely prospects for the continued spread of this mode of electoral reform.
INHERENT AND CONTINGENT FACTORS IN ELECTORAL REFORM
Shugart suggests in Chapter 2 that there are both inherent and contingent factors that lead to electoral reform. Inherent factors include the perceived pathologies of pre-existing extreme systems, and the systemic failures that result; contingent factors are political events and interests that lead politicians to pass an electoral reform. Within the category of contingencies, Reed and Thies (Chapter 7) define two classes of contingencies. Outcome-contingent. factors spur incumbents to vote for reform if they believe they will be better off under new rules, such as when PR was adopted by parties suffering electoral declines as new working-class parties emerged in Europe around the turn of the twentieth century (Rokkan 1970; Taagepera and Shugart 1989: 148-53). Act-contingent reasons are present when politicians do not actually
572
Conclusion
Áre MM Systems the Best of Both Worlds?
573
prefer the proposed new system, but fear electoral retribution if they are seen as blocking reform. We now review the inherent factors that led to electoral reform in each of the live cases of reform in established democracies, and then review the contingent factors that led to mixed-member systems being adopted in both established and new democracies.
Inherent Factors; Extreme Electoral Systems and Systemic Failure
Chapter 2 of this volume identified four ideal-type "extreme" systems, and suggested that they contain conditions that can, under certain circumstances engender pressures for reform. The four types are: pluralitarian (New Zealand), hyper-representative (Israel and Italy), hyper-personalistic (Japan), and hyper-centralized (Venezuela). In each of the five cases of reform in established democracies that are covered in the chapters of this volume, a systemic failure grounded in the pathologies of an extreme electoral system was a precondition for electoral reform.1
In New Zealand, as Denemark notes, much of the voter anger that ultimately was channeled toward electoral reform stemmed from a pervasive feeling that governments had become unaccountable. Two consecutive governments from different parties had unleashed, and then deepened, neoliberal economic reforms which radically transformed the state's relationship to the economy and the provision of public services. Perhaps most galling to New Zealanders was that none of these reforms had appeared as manifesto commitments prior to the election. To make matters worse, the government that began these reforms had received only 43% of the vote in 1984, but won 60% of the seats owing to the extreme pluralitarian nature of the electoral system. Incredibly, this was one of the better performances of the electoral system in this period. In the 1978 and 1981 elections, the party that came in second place in votes had been handed full unchecked governmental power. Minor parties had grown in vote shares, yet were being virtually shut out of parliament by the electoral system. With this record of systemic unresponsiveness in their recent past, New Zealand voters opted for MMP when
1 This view is not universally shared by students of recent electoral reforms. For instance, Dunleavy and Margetts (1995) claim that these reform episodes challenge an "orthodoxy"—of which one of us (Taagepera and Shugart 1989) is said to be a primary perpetrator. The alleged orthodoxy holds that electoral systems do (and should) change only in extraordinary situations such as revolutions or after wars. Dunleavy and Margetts counter that recent reforms have occurred as part of what they call "normal politics." They then proceed to speak of corruption and a "citizens' revolt" in Italy and note that electoral systems are hard to change, notwithstanding recent events. We would argue thai some sort of systemic failure has probably occurred well beyond the realm of normal politics when language like "citizens' revolt" is used to describe a process of change in an established democracy. Moreover, we would caution that, in spite of the recent movement toward MM systems, the great majority of the world's established democracies have not adopted electoral reform.
given the chance because it appeared to be a means of restraining political parties by forcing them to enter into coalitions in the event that no party received a majority of the popular vote. Electoral reform was thus an extra-constitutional means of placing a check on previously unrestrained governments.
To understand reform, it is also important to consider similar cases where reform has not (yet) occurred. Accordingly, we also review the experiences of Britain and Canada, which are cases that bear some similarities to New Zealand, but thus far have not changed their electoral systems. The British experience parallels that of New Zealand in that a government undertook radical economic reform after assuming power with far less than a majority of the popular vote. Thatcher's Conservatives reached their high point in votes in their first election, when they obtained about 44% of the vote in 1979. Despite declining vote shares in each of the next three elections, they continued to be re-elected. The radical behavior of these Conservative governments and their seemingly entrenched manufactured majority helped generate a constituency for proportional representation within the Labour Party, as well as among a segment of the public.
The fact that electoral reform remains merely on the agenda in Britain can be attributed to a key difference from the New Zealand case. Despite some similarities with New Zealand, Britain's experience with radical governments in the 1980s cannot be seen as emblematic of a systemic failure. Unlike in New-Zealand, where two consecutive elections had given seat majorities to the party that came second in votes, no such reversal had occurred in Britain since 1951 (though in 1974 a short-lived minority government was formed by the party that was second in votes). In addition, minor parties, while severely under-represented, have not fared nearly as badly as in New Zealand. Perhaps most significantly, British voters came to view the radical policy innovations of the Conservative years as largely beneficial (Norpoth 1992), and they decisively rejected a clear alternative platform offered by Labour in 1983 in what became known as "the longest suicide note ever written". Whereas New Zealanders saw both major parties leading the country in a radically new direction that had not been foreshadowed in either party's campaign promises, British voters had clear choices and made a relatively clear decision. Although the Conservatives were re-elected with declining votes shares (42.4% and 42.3%, respectively, compared with 43.9% in 1979), Labour was far behind (27.6% in 1983, and 30.8% in 1987).
When the Conservatives finally were voted out of office in 1997, the alternative was a party offering to perfect Britain's new direction, as symbolized by their slogan of "Britain Can Do Better". Transformed into what Tony Blair called "New Labour", the Labour Party moved closer to the median voter, just as the theory of party competition in plurality systems would predict. In short, there was no systemic failure of the party system in Britain.
574
Conclusion
Are MM Systems the Best of Both Worlds?
575
Consistent with this interpretation, Chapter 2 showed that Britain was not an extreme case of phiralitarianism as was New Zealand, in part because the winning party has frequently come closer to a majority of the vote and the seat bonus for the plurality party has tended to be less. Thus, the inherent problems of pluralitarian systems have not been present in Britain to the same degree as in New Zealand.
Nevertheless, electoral reform is on the British political agenda, in no small part because the electoral system itself engendered the conditions for reform-fostering radical policy swings, entrenching one party in power even as its voter support declined, and under-representing smaller parties. Most important among the under-represented parties are the Liberal Democrats, which grew originally as a more moderate opposition party than Labour in the 1980s, yet was severely punished by the electoral system. Also noteworthy is the regional role of parties like the Scottish Nationals and the Plaid Cymru of Wales, which may become more important now that regional assemblies have been elected (under MMP rules). Thus, national electoral reform is on the agenda, although radical reform such as a shift to full proportionality is not contemplated. The system proposed by the Jenkins Commission is only a moderately proportional variant of MMP, an incremental rather than profound change.
In the Canadian case, the seeds for systemic failure are certainly present, as Weaver notes. The plurality system clearly exacerbates Canada's regional: problems. Some parties are far more dominant in the seats elected from specific regions than they are in votes, thereby encouraging regional movements such as the Quebec separatists. Furthermore, sometimes a governing party in Ottawa has ended up with virtually no seats in the House of Commons from particular regions. Despite these problems, electoral reform has yet to gain serious consideration on the Canadian national agenda. A mixed-member system would be a logical choice, as it would retain SSDs while at the same time permitting national parties to win at least some seats in regions where they are currently weak. Interestingly, the lack of interest in electoral reform can largely be explained by the same regional patterns that threaten national unity. A regional concentration of the vote provides parties with rewards in the SSD electoral system, and these local strongholds—including control of provincial governments—could be threatened by a mixed-member system that increased proportionality. This salient feature of the Canadian political scene has been only a minor factor in Britain, and was not a factor at all in pre-reform New Zealand, where all the parties were of national scope and there are no regional governments. Like Britain, Canada lacks some ot the inherent conditions that generated reform in New Zealand. Indeed, Chapter 2 showed Canada to be considerably less pluralitarian than New Zealand, in part because the regional strengths of third parties have generated minority governments in some cases instead of the regular manufactured majorities and back-to-back reversed pluralities of New Zealand.
In contrast to New Zealand and Britain, where electoral reform made it onto the agenda partly because governments did too much, in other cases it could be said that reform became an issue because governments did too little. Examples include the hyper-representative cases of Italy and Israel and the hyper-personalistic system of Japan. Corruption became a critical issue in both Italy and Japan, with governments seen as unable to reform themselves. In Italy, citizens of the industrialized and strongly pro-European north were particularly frustrated, as they increasingly blamed the political parties centered in Rome for tunneling resources to fuel their clientelistic networks in the poorer south. A separatist party, the Lega Nord, hence grew spectacularly in the north. Still, the Christian Democratic Party held on to its position as the predominant parly, partly through use of clientelism and patronage in the south. Through this system, successive Italian governments were blamed for being unable to address Italy's stark regional variations in development through provision of collective social programs (Bull and Newell 1993: 203). As Katz notes, the "clean hands" investigation by Italian prosecutors probing political corruption effectively delegitimized the existing system, revealing the corruption underlying the entire system. The highly proportional electoral system that had maintained the same Italian parties in power for decades came to be blamed for the inability to overcome corruption, for threats to national unity, and for mounting social and economic problems. As a hyper-representative system with a significant candidate-centered component (open-list PR), the Italian system was inherently incapable of generating governments with a mandate to weed out the pervasive corruption and clientelism and to make the difficult policy tradeoffs needed to assure the country's successful integration into the single European currency.
In the Japanese case, as described by Reed and Thies, corruption flourished alongside a regulatory process that favored the ruling Liberal Democratic Party's business constituencies. Dependent on business contributions to finance intraparty competition under the existing hyper-personalistic system, the LDP repaid its business contributors with policy favors in the form of cartels and other regulatory concessions and was unresponsive to consumer interests (Cox and Thics 1998). A series of high-profile corruption scandals were clearly linked to the needs of candidates to finance their individual campaigns (see Chapter 7). Corruption became a salient issue, and electoral reform was promoted by some LDP leaders as a solution.
In sum, reform made its way onto the Japanese and Italian political agendas in large part because it appeared to be the most likely way to shock the system. A primary goal in both cases was to weed out clientelism and corruption by enhancing the accountability of governments and eliminating intraparty competition. Aside from being an end in itself, reducing the incentives that led to corrupt behavior was believed likely to enhance the capacity of
576 Conclusion
governments to accomplish policy changes lhat had been stalled despite be:»a desired by broad constituencies.
The failure of government to accomplish goals demanded by the broade electorate was also at the root of reform in Israel. As discussed by Rahat it).-, grand coalition ("Unity") governments of the 1980s established a system in which the two major parties checked each other and blocked progress on critical issue of peace with the Palestinians. When the coalition broke dov\ the strange-bedfellows deals on policy cut between Labor and the ultra-Orthodox parties—deals of a sort than can be expected to be inherent to hyper-representative systems—catalyzed the reform movement.2
Perhaps the clearest case of systemic failure is that of Venezuela, where a previously stable democracy was in serious decay by the late 1980s. As in and Japan, pervasive corruption had become a major issue, even leading to the impeachment of a president in 1992. Parties were failing to respond to new citizens' movements which demanded policy changes, such as environmeiiu;! protection and better (i.e. non-clientelistic) public service provision (Chapter 8 above). Above all, there was a deepening sense that politicians were unconcerned with voters' needs, as the major decisions were being made by a small clique of national leaders who had only a tenuous accountability to the i. ■ , and file (Martz 1992; Coppedge 1994; Crisp 1998). As argued in Chapter 2, such lack of legislator-constituent links is inherent to hyper-centralized .systems. Political reform was an issue in the 1983 and 1988 elections, but it was only after massive rioting in response to President Perez's neoliberal decrees that reform finally passed. By then it was, arguably, too late. The failures of successive Venezuelan governments to address the growing crisis led to two coup attempts in 1992, before the MMP system was even implemented, then to the presidential impeachment, and finally to the election of one of the coup
2 In the case of the majoritarian/pluralitarian systems, we have case studies in this volume on countries that have not adopted a mixed-member system thus far. Therefore, we have been able to consider what conditions were present uniquely in New Zealand thai brought about reform. For hyper-representative systems, unfortunately, we do not have any cases of no reform represented in this volume, notwithstanding the fact that Chapter 2 showed that there are two cases Belgium and Finland that are actually more extreme than Italy. In Belgium there were severe threats to the country's unity from the 1960s on and there was a resulting political reform. However, that reform took the form of a new federal constitution in 1993, rather than reform of the electoral law. As a society divided by a deep linguistic-cultural cleavage as well as other issue dimensions (Lijphart 1998), Belgium probably could not afford the bipolar party system that a MM system would be likely to bring about. Thus, a MM system would be less likely to be placed on the reform agenda. Finland, like Belgium, has an ethnic dimension of high salience (Lijphart 1998: 80), though there is no obvious case of systemic failure. Nonetheless, a political reform has been adopted in Finland, in the form of direct election of the president since 1996. This reform may make the system "feci" less hypcr-rcprcscntative, even though it does not appear thus far to have had a noticeable effect on the legislative party system. See Shugart and Carey (1992: 266-8) on the prospects for a directly elected presidency to enhance identifiability of government choice in Finland.
Are MM Systems the Best of Both Worlds? 577
plotters as president in 1998. In 1999 Venezuela adopted a new constitution 'jn an effort by the new president to replace the old failed order with a new one. Tellingly, however, the MMP system was not blamed for the crisis; and indeed, Venezuela's 1999 constitution may be the first in the world that can he read as requiring a mixed-member system. Its chapter on the congress states that it shall be elected by a method that is both proportional and "personalized."
Act-Contingencies: Electoral Competition
Inherent conditions leading to systemic failure are not enough to bring about reform. A key contingency common to reform in established democracies is the act-contingent explanation of pressures to "do something" to address systemic failure. Parties in each case found themselves impelled to act by pressures arising out of interparty competition. The starkest example comes from New Zealand, where it is quite clear lhat most major party politicians did not favor electoral reform when the issue first came on to the agenda. Yet, as Denemark notes, in a campaign debate the Labour leader made an unscripted comment in favor of holding a referendum on electoral reform—most likely as an attempt to catch his opponent off guard. After his party shelved the idea once it came to power, the opposing National Party picked it up in the next campaign as an example of Labour's unresponsiveness, even though National was no more interested in actually adopting MMP. Given the rising level of public anger at politicians at the time owing to the unpopularity of drastic economic reforms, the issue became one that neither party dared be seen as squelching.
Similarly, Reed and Thies note that even groups that were in fact opposed lo the idea of a MM system in Japan called their far more timid proposals "True Reform", which suggests that public support was so strong that even reform opponents felt compelled to portray themselves as in fact being reformers of a different stripe. In the Venezuelan case, all major presidential candidates in 1988 signed a document committing themselves to political reforms. As in Japan, there was no public outcry for a mixed-member electoral system in Venezuela, but election campaigns were driven by the quest for some sort of fix for a political system that was widely perceived to be broken.
In Italy, the formal process of electoral reform began with a citizen-initiated referendum, which passed because of voter disgust with corruption. Thus, Italy presents the clearest example of the voters setting the agenda. In the other cases reviewed above citizens set the agenda indirectly, as politicians lost control of the issue and were pressured to respond to voter demands for reform. As Katz notes, Italian politicians were legally forced to respond because the referendum deleted certain phrases from the pre-existing electoral
578
Conclusion
Are MM Systems the Best of Both Worlds?
579
!0ral law deliberations. Why have so many recent reforms produced a mixed-• member system as the specific class of systems to be adopted?
A key reason that mixed-member systems so often have broad appeal is because they almost inevitably preserve some features of the pre-existing system" while grafting on new features preferred by some parties. In a country starting with an all-SSD system, such as New Zealand, MMP is a convenient wav to retain the familiar features of nominal voting in geographically defined constituencies while nonetheless adopting PR. In a PR system where reformers desire introducing more majoritarian characteristics, such as Italy, gMM is a logical way to retain some proportional element to appease minority parties while still crafting a fundamentally majoritarian system. If the preexisting system entails intraparty competition that reformers seek to curb, all mixed-member systems by definition have nominal voting, which continues to allow for some degree of personal vote seeking; for example, in Japan a personalistic system was replaced by a system that is not very personalistic, but nonetheless docs not entirely abandon nominal voting. Finally, in a closed-list PR system in which it is seen as desirable to introduce nominal voting, as in Venezuela, MMP permits this while retaining both proportionality and the selection of some legislators via closed lists. In sum, a general feature of all the cases in which MM systems were adopted in established democracies is that the system emerged as a compromise which retained some familiar features of the old system while nonetheless altering the basic principle by which seats are allocated.
law for one house and essentially left the two houses with incompatible eiec toral systems. Thus, one key contingency in each episode of reform was the act-contingent motivation of fear of the voters' wrath, thereby making p0|,.. cians come to terms with an issue most of them would have preferred to let ii-. dormant.
Outcome-Contingencies: Political Compromise Leading to Mixed-Member Reform
Another theme running through nearly all the chapters in Part II is that mixed-member systems typically emerge as a product of political compromise. The bargaining that produces a mixed-member system as the specific political reform is an outcome-contingent factor because, once parties have (perhaps reluctantly) recognized that their political survival requires electoral reform, each party may have a different conception of the system that best suits it. Thus, parties tend to favor different specific electoral rules accent ■■■> to what systemic outcomes are most favorable to them. As systems that balance the tendencies of PR and majoritarian electoral systems, mixed-member (MM) systems in their myriad variations offer especially fertile terrain lor political bargaining. Although voters may have been clamoring for reform, in no case was there a clamor specifically for a mixed-member system. The resulting MM systems were the product of cross-party compromise.
Political compromise in established democracies
Most MM systems did not emerge as the experts' pick, but rather as a piud-uct of negotiation among parties with diverse preferences. An exception i> New Zealand, where a Royal Commission weighed several options and ^el-tied on MMP because it came closest to meeting the Commission's criteria lor what a good electoral system for New Zealand should look like (chapter 4 above). New Zealand is the only case thus far in which a proposal drafte.l i>> an independent commission was submitted to a referendum. Britain in.;> soon follow a similar path, given the Labour Party's manifesto commitment to holding a referendum on electoral reform, but we suspect that there will first be some bargaining between Labour and the Liberal Democrats concerning the proposal made by the Jenkins Commission. In Italy, the referendum only triggered bargaining in parliament. In Israel, Japan, and Venezuela, there never was a public vote on the electoral system.
Thus, apart from New Zealand (and perhaps Britain in the future), we have mixed-member systems emerging as compromises among party leaders and legislators. Some parties preferred majoritarian systems, some PR; some preferred party-centered voting, others candidate-centered. MM systems, as noted in Chapter 2, offer the promise of balance between the pure types. However, the presence of mixed preferences is no doubt present in all dec-
Political compromise in new democracies
New democracies do not present electoral system designers with a tabula rasa. There are usually parties—and always interests—that carry over from the pre-existing authoritarian regime. Some of these regimes will have staged elections—even if they lacked meaningful choice—and the pre-existing electoral system may even wind up serving as a point of departure for reformers, much as it docs in established democracies.3 Mixed-member systems have been adopted in those cases where a former "official" party remained strong enough to play a role in brokering the transition to democracy. For instance, among the former communist countries, Albania, Armenia, Bulgaria, Croatia, Georgia, and Hungary were all characterized by communist parties (or their successor organizations under a new name) that remained strong enough to bargain over their gradual exit from power. In the Russian Federation, the initial electoral law was adopted by presidential decree, but the incumbent legislators had all been elected in single-seat districts in the last election of the Soviet Union. As Moser and Thames show, the designers of the
3 Even where a non-democratic regime did not hold elections of any sort, as with some military governments, there is usually a prior democratic experience that might serve as a Point of departure for a new system.
580
Conclusion
Are MM Systems the Best of Both Worlds?
581
Russian system consciously sought to retain personal representation inberen in SSDs even as they sought to introduce a party-building feature through^; PR. In most of the other former communist systems as well, incumbent legi^ lators had been elected in SSDs. Thus, as in some ongoing democracies a tier of SSDs established a familiar aspect of the electoral system to be maintained while the system was opened up to new contenders via a proportional tier
All but one of the post-communist cases that adopted a mixed-member sys tern chose MMM. The far greater likelihood of MMM over MMP in p0st communist systems can be explained by the continued considerable strength of the communist party (or its successor), which generated a clear cleavage between the old and new forces. In Armenia, Bulgaria, Croatia, Georgia, and Hungary former communist parties retained bargaining power over the new laws and presumably expected a majoritarian system to be in their interests as a major party. In Russia, where the President controlled the writing of the new law, it was a desire to defeat the former communists that led his pro-market reformers to favor the more majoritarian system.4 The one post-commimist state that adopted MMP was Albania, where an initial election had been held under a pure SSD majority system, but the ex-communists were aware of their plummeting voter support and chose MMP for the second election.5 The ascendant former opposition party, confident of its new-found majority status, then shifted the electoral system to MMM.6
In contrast to countries that began their democratic transitions with relatively strong holdover parties from the pre-existing regime are a set of cases in which the power of the communists quickly disintegrated. In each of these cases, including Czechoslovakia, Poland, and Romania, variants of purely list PR rather than mixed-member systems were adopted. Continuing the theme, in Bulgaria—perhaps the only case where a pure MM system was adopted and then abandoned7—the Communist Party's fortunes declined so precipitously after it won the 1990 election under MMM that it sought refuge in a pure PR system, apparently fearing that even MMP would prove an
,. enlbarrassment because the party might not win any SSDs.s As for the nonparty authoritarian regimes, it is noteworthy that not a single transition to democracy succeeding a purely military regime has yet resulted in a mixed-member system.9 MM in new democracies is thus a product of transitions
.: that feature a declining but still powerful ruling party and a rising opposition. In some other cases of "new" democracies, MM was adopted after other
.' electoral systems were tried. A series of often unstable or allegedly fraudulent elections preceded the adoption of a MM system in Bolivia, Mexico, and Thailand. Thus, these cases are a sort of hybrid: they are not really "established" democracies, but neither are they brand-new regimes. Such reforms nl0re closely resembled those of the established democracies in that an exist-
' jng legislature representing multiple parties was a central actor in the adop-
•'• tion of MM. Indeed, each of these three cases prior to reform may be seen as what was labeled in Chapter 2 an "extreme" system. Bolivia, and especially Mexico, were hyper-centralized systems, while Thailand was a hyper-
;   personalistic system.
;    In the Bolivian case, as Mayorga shows, the adoption of a MM system was
■ linked with the process of administrative decentralization as a means to infuse local accountability in a system of highly centralized parties. In Thailand, a multi-seat plurality system with exclusively nominal votes was characterized by highly factionalized parties that avoided campaigning on national policy themes. As in Japan and Russia, adopting a MM system was a means by
• which incentives for greater party cohesiveness could be injected (Hicken n.d.). In the case of Mexico's party-based authoritarian system, forms of mixed-member systems were already in use long before a transition to democracy was much more than a remote possibility (see Chapter 10). Although the Mexican system remains MMM, recent reforms have introduced greater proportionality, which is consistent with what we expect of democratization
. in which a declining, but still strong, party retains veto power over the elec-
' toral law.
4 A related consideration in Russia and also in Ukraine is that the parallel tiers of MMM are more favorable than the linked tiers of MMP to the survival of nonpartisan candidates,' whom the presidents of both countries had found could be dealt with by trading patronage for votes (personal communication with Sarah Birch and Robert Moscr).
5 These observations on Albania—and, in the next paragraph, on Bulgaria are basedi on Shugart's visits to these countries on consulting missions dealing with electoral-law drafting in 1991.
6 Subsequently the former communists won again and were the party benefiting from the now more majoritarian features of the system. Such are the risks one takes when manipulating electoral systems for short-run advantage.
7 Denmark in 1918 had a system that was very close to MMP, but it did not fully meet; the defining criteria given in Chapter 1 because Copenhagen contained no SSDs. In theresf: of the country, MMP rules prevailed (see Elklit 1992).
THE BEST OF BOTH WORLDS?
The persistence and further spread of mixed-member systems will ultimately depend on whether this type of electoral system is perceived as having delivered on its promise of offering the best of both worlds. How then do we define
s The recent experience of Hungary, where the Socialists won only one SSD. was very keenly on their minds.
9 South Korea had a military-dominated regime prior to its adoption of a mixed-member majority-assuring system in its transition to democracy in 1987. However, there were both elections (under SNTV) and an official party, so it was not a pure military regime on the order of those in Argentina or Chile.
582
Conclusion
Are MM Systems the Best of Both Worlds?
583
the best of both worlds? The answer must be broken down into the interpatt and intraparty dimensions, as each pure type of system theoretically'^ advantages on each dimension. A balance between the pure types on eithe dimension maximizes electoral efficiency, as defined in Chapter 2. T On the interparty dimension, each of the basic types of electoral system • perceived as offering certain distinct advantages. The features of purely majoritarian systems that are most often positively regarded are:
1. enhanced stability of governments (in parliamentary systems), as a result of their relative disproportionality; and
2. clear choices for voters (identifiability), as a result of a two-party or two-bloc pattern of competition.
The interparty advantages of proportional systems are generally taken to be-
1. their fair treatment of smaller parties, such that diverse views of the electorate are represented; and
2. their encouragement of coalitions, such that a single party that fails to capture a majority (or very nearly so) of votes is not permitted to govern alone.
Therefore, the best of both worlds on the interparty dimension would be a multiparty system, but one in which most parties are aligned with one of two broad blocs such that voters can choose between potential center-left or center-right governments.
On the intraparty dimension, again, each of the basic types of electoral system is perceived as offering certain distinct advantages. Majoritarian systems are based on nominal voting and are said to offer the advantages of:
1. representation of local interests: and
2. personal accountability of individual members.
Proportional systems are based on party-list voting and are admired for their ability to produce:
1. parties of national scope; and
2. coherent and disciplined parties that offer programmatic representation.
Thus, the best of both worlds could be summarized as disciplined national parties whose individual legislators can be held accountable for their articulation of local interests.
Criteria for Evaluation: The Interparty Dimension
On the interparty dimension, we assess whether each mixed-member electoral system has provided:
1, a multiparty system aggregated into two blocs; and
2. Eeffective representation of small parties, including parties that may not fit well into either principal bloc.
{n addition to an impressionistic assessment of these two criteria, based on the case studies of this book, wc also employ the method of estimating interparty efficiency that was introduced in Chapters to compare mixed-member systems to other electoral systems.
Two blocs
Germany, the oldest mixed-member system, has provided a two-bloc system throughout much of its existence. The only clear exception would be during the three-year period of the Grand Coalition, when the two biggest parties uoverned together. At present, however, Germany clearly has two government alternatives: Christian Democrat-Free Democrat vs. Social Democrat-Greens. As Klingemann and Wessels note in Chapter 13, the German two-vote MMP system increases the likelihood that the junior coalition partner will remain large enough to maintain parliamentary representation while the larger party benefits from strong performance in the SSDs.10
Among the nine other mixed-member systems discussed in this book, five have developed what are essentially two-bloc party systems: Hungary, Israel, Italy, Japan, and New Zealand. Of course, the development of such systems cannot be fully attributed to the mixed-member system, but the electoral system may help bring about a two-bloc party system, or maintain one where it has already arisen. The achievement of such a two-bloc format and the possibility of alternation in government was an explicit goal of reformers in Italy and Japan. In several of the new MM systems, it is especially noteworthy that a two-bloc party system was unlikely to emerge under the pre-existing rules. Italy and Israel had fragmented multipolar party systems, Japan had a dominant-party system, and Hungary's first election exhibited a three-bloc format.11 In each case, the mixed-member system may have contributed to the consolidation of the party system.
More specifically, where the pre-existing party system was fragmented and multidimensional, a mixed-member majoritarian system appears to have been particularly effective in facilitating two blocs of parties. In all probability, MMP would not have been as successful at bringing about consolidation under such conditions. The amalgamation that occurred—often taking the form of pre-election coalitions, such as in Italy and Hungary—would have been less likely under a mixed-member format that offered small parties the
10 Ii does so by facilitating coalition-based strategic voting, whereby many voters favor the larger party in SSDs and its coalition partner on the list, and by the provision for overhang seats for a party that outperforms its proportional vote in the SSDs.
11 On Hungary's evolving parly system and its policy consequences, see Haggard ct al. (21)00).
584
Conclusion
Are MM Systems the Best of Both Worlds?
585
prospect of full proportional representation. Thus, we would posit that MMM is more likely than MMP to correct a pre-existing problem of excessive party system fragmentation. Bolivia is a case in point: Its pre-existing party system was extremley fragmented, and Bolivia remains a hypei~ rcscntativc system even under MMP.
Two of the MMP systems have nevertheless performed surprisingly well on this criterion. Germany has a two-bloc format, as noted, and New Zealand developed one by the time of its second MMP election (see Chapter 14) Bolivia does not appear as likely to develop such a party system, though its unique hybrid constitutional form may be a factor in retarding the growth of a two-bloc party system (as we discuss below). Venezuela's current political instability makes it impossible to predict.
Proportional representation of small parties
Most of the mixed-member systems covered in this book perform reasonablv well on this dimension. Naturally, all the MMP systems do so given that thev are proportional systems. Even among the MMM systems, however, the PR tier has permitted smaller parties to carve out space for themselves if they are unwilling to line up behind one of the dominant blocs. Owing to the list tier, and especially where voters cast two votes, minor parties are able to survive without necessarily undermining the two-bloc party system. For example, the directly elected Israeli prime minister is freed from dependence on minor parties, though parties that represent significant social interests retain influence. Similarly, the Party of Democratic Socialism in Germany currently coexk -within the two-bloc format of the larger party system.
Inter party efficiency
As defined in detail in Chapter 2, the index of interparty efficiency is built out of three basic components:
1. Pre-election identifiability (ID), which is the tendency of elections to revolve around two competing blocs of parties, such that voters can determine the likely post-election government options.
2. Post-election majority approximation (MA), which is based on the seat share obtained by the largest party or pre-election coalition forming part of the initial post-election government; its share is divided by 0.5V<; to indicate how close it came to a majority of seats.
3. Plurality enhancement (P), which indicates the degree to which the electoral system over-represents the largest party or pre-election coalition.
The first two components arc averaged to arrive at a value of electoral linkage (L), and then interparty efficiency (Ei„ter) is calculated as:
Pinter     P ~™ 1 P-
f ,\$l£ 25.1. Indicators uj'Electoral Efficiency in Mixed-Member Systems
Post-election
Pre-election     majority Electoral   Plurality Interparty
identifiability   approximation linkage     enhancement efficiency Country (ID) (MA) (L) (P) (Eirll„) '
MMM
Hungary	0.83	0.90	0.86	0.34	0.20
Italy	1.00	1.00	1.00	0.22	0.21
Japan	1.00	0.96	0.98	0.18	0.16
Russian Federation	0.50	0.28	0.39	0.02	0.59
MMP					
Bolivia	0	0.49	0.25	0.02	-0.74
Germany	0.67	0.97	0.82	0.03	-0.16
New Zealand	0.87	0.86	0.87	0.06	-0.07
Venezuela	0.75	0.60	0.68	0.05	-0.28
Other					
Israel	0.75	0.48	0.62	0.03	-0.35
Mexico	0.88	0.96	0.92	0.16	0.08
Values of all these components and the resulting efficiency index for twenty-one democracies that use electoral rules other than mixed-member systems were displayed in Chapter 2. Table 25.1 shows the values for the ten mixed-member systems covered in this book. Of course, these values should be viewed with caution, given that we have at most two elections under the MM format except for Germany and Hungary. Nonetheless, a consideration of how widely divergent countries have performed thus far under MM systems is valuable in assessing the promise of this class of electoral system.
In order to make comparisons between mixed-member and other types of electoral systems, we report mean values for several classes of electoral systems in Table 25.2. The first row of this table reports the indicators of all
Table 25.2. Mean Values of Indicators of Interparty Efficiency, by Category of Electoral System
Type of system	TD	MA	L	P		Absolute valuea
Non-MM systems	0.70	0.88	0.79	0.12	-0.09	0.25
Exclusively SSD systems	0.98	0.98	0.98	0.22	0.21	0.23
Exclusively list systems	0.54	0.81	0.68	0.08	-0.24	0.31
MM systems'3	0.86	0.89	0.88	0.15	0.03	0.17
MMM (Hung., Italy, Japan)	0.94	0.95	0.95	0.25	0.19	0.19
MMP (Germany, NZ, Ven.)	0.77	0.81	0.79	0.05	-0.17	0.17
"Mean of the absolute values of the scores on Einler.
bExcfuding the unusual hybrid constitutional systems found in Bolivia (parlianientarized presidentialism), Israel (presidentializcd parliamentarism), and the Russian Federation (president-parliamentary).
586
Conclusion
Are MM Systems the Best of Both Worlds'/
587
non-MM systems covered in Chapter 2, whereas the second and third rows show those for the pure-SSD and pure-list PR systems, respectively. The fourth row shows mean values for seven of the ten MM systems. The reason for the exclusion of three of the MM systems—Bolivia, Israel, and the Russian Federation—is that each of these employs a unique hybrid constitu-tional form not found among our set of non-MM cases. Bolivia is a system 0f parliamentarized presidentialism (see Chapters 9 and 19 by Mayorga), where congress selects the president if no candidate receives a majority of the popular vote. Israel has the opposite hybrid, presidentialized parliamentarism (see Chapter 6 by Rahat and Chapter 16 by Kazan), where the head of government is directly elected yet is dependent on parliamentary confidence. Both of these formats may undercut the two-bloc tendencies of the typical MM system because neither a direct vote nor elite bargaining among parliamentary parties is decisive in empowering governments.12 The Russian system likewise does not establish either the presidential or parliamentary vote as decisive in the formation of governments, being what Shugart and Carey (1992) call a president-parliamentary system.11 Moreover, Russia has a notoriously unin-stitutionalized party system which makes it hard to analyze using conventional tools of electoral-system analysis (sec Chapter 22 above).
The seven MM systems that are employed in straightforward parliamentary or presidential systems average 0.86 on idcntifiability (ID), which is an intermediate level compared with the pure-SSD (0.98) and pure-list (0.54) systems, but is much closer to the SSD systems. An ID score of 0.75 or better implies a two-bloc party system with some minor parties that can swing to either major bloc; the score can be 1.00 only when one of two major parties is likely to form a single-party government, or when every significant party is encompassed by one of two blocs. The considerably higher score for MM -terns than for list PR systems reflects their greater tendency to generate clear choices of competing governments. The lower score relative to pure SSI > -\>-tems reflects their tendency to have significant minor parties. In stim, this
12 The incentives that these systems provide for parties are different, but in the end neither is likely to be as conducive to the formation of a two-bloc party system as a pure parliamentary or presidential system with MM. The Bolivian format may increase the incentive ol parties not to align with a major bloc before the election so as to play a role in the selection h> the congress. Yet it does not make parties in parliament responsible for govei.nng (via no-conlidence votes) as does parliamentarism. The Israeli format has encouraged a bipohinly in the prime ministerial election, but this election is not decisive for government formation, ami there has been a resulting centrifugal movement on subsidiary dimensions of partisan conflict and a great increase in the dispersion of the parliamentary vote (see Chapter 10 by Hazan).
13 In such a system, the president may appoint cabinet ministers and may also dianis* them at will without regard for whether they enjoy parliamentary confidence. However, parliament also has the power to cast no-confidence votes. This system is different ,roin other so-called hybrid systems, such as those of Finland, France, and Portugal, in w:iie:i t .t parliamentary majority effectively determines the composition of the cabinet. Shugart i-.nc Carey (1992) consider these latter three cases to be "premier-presidential" systems.
, intermediate placement on ID suggests that the best of both worlds is indeed being realized by MM systems on this performance indicator.
On majority approximation (MA), MM systems (0.89) have a slightly higher score than the PR systems (0.81), though not as high as the SSD sys-tems-(0.98). On plurality enhancement (P) the MM systems are intermediate again, as expected, with a score of 0.15. They do not enhance pluralities
\ nearly as much as SSD systems (0.22), but as a class they do so more than PR systems (0.08).
Finally, on interparty efficiency, MM systems indeed fall in between the pure types, and indeed, at 0.03, score as almost ideally efficient as a class. If this sounds too good to be true, in a sense, of course, it is. This highly efficient mean score masks deviations from efficiency by individual systems. Thus, the last column of Table 25.2 indicates the absolute value of the efficiency scores. The typical MM system deviates, one way or the other, by 0.17, which is still less than found for either SSD or PR systems. Thus, MM systems are indeed between the two pure types, as expected.
The final two rows of Table 25.2 break the data for MM systems into \1MM and MMP subtypes. Although doing so risks increasing the error inherent in a very small sample, it is worth doing this in order to assess the performance of the more specific sub-categories that we have oriented this book around. Each of the three MMM systems does indeed lean toward the majoritarian side of the scale, as indicated by the positive values on the efficiency index, and also by values on all the other indicators that are similar to the SSD systems. Each of the three MMP systems leans toward the "representative" side of the scale, as indicated by the negative values on the efficiency index.14 The analysis of this section provides considerable support for the .-.otion that mixed-member systems offer the best of both worlds on the interparty dimension.
Criteria for Evaluation: The Intraparty Dimension
On the intraparty dimension, we can assess the effects of the mixed-member systems on two criteria:
1. local accountability of individual legislators; and
2. national, and presumably, programmatic representation.
v*!.iMeX1C0 'S excluded here' as lts limited MMM makes it an intermediate type between ■viMM and MMP; indeed it has the most efficient score of all the cases analyzed here. This |-suit may be as much a product of the specific party-system context in which the system »-ts adopted as of the electoral system itself. However, the limited MMM system in Mexico '» unlikely to be as majoritarian as the parallel MMM systems or as proportional as the -■mpensatory MMP systems, so it docs seem as though Mexico has found the best of both '•ends—at least on the interparty dimension, and at least under current political condi-
588
Conclusion
Are MM Systems the Best of Both Worlds?
589
Local accountability
The existence of an SSD tier should engender the personal accountability 0f members elected from the nominal tier to local interests and issues. Germany is the most advanced example of this local representation in a mixed-member system (see Chapter 13 by Klingemann and Wessels). The only system reviewed here where such an outcome is not expected is Israel, as only the prime minister is elected nominally. Nonetheless, even in Israel we could say that the direct election of the PM has promoted his personal accountability (as he is a member of the Knesset as well as head of the government) to his constituency—the nation as a whole. All the remaining systems have achieved this criterion to some degree, with the possible exception of Italy.
In the New Zealand and Japanese cases, achievement of this criterion is hardly surprising, given that the previous system was one of nominal voting in geographically defined districts. Therefore, a tradition of constituent responsiveness was already well established—indeed, to a pathological extreme in the Japanese case. What is most striking in these cases is that some degree of personalization has also carried over into the list tier. In Japan this phenomenon is encouraged by electoral rules whereby the SSDs of the nominal tier effectively function simultaneously as nominating districts for the list tier. Not all Japanese parties or alliances take advantage of this option, because some seek to obscure the personal accountability of less popular candidates (see Chapter 17). However, to the extent that parties do take advantage of it, legislators elected from lists have an incentive to "shadow" the, legislators elected in their SSD and thereby to enhance their party's representation of local interests. In New Zealand and Germany such practices are more informal, in that parties have tended to assign their list-tier M Ps to one; or more SSDs.
The cases where it is most difficult to infuse local accountability are those; countries with no previous experience with nominal voting. Germany at the founding of the Federal Republic,15 as well as Bolivia and Venezuela prior to.: their recent reforms, were all such systems. Bolivia provides evidence of anj emerging local accountability, as Mayorga notes in Chapter 19 that legislators elected in SSDs have been asked to perform basic constituent services; and obtain support for local improvements. Additionally, large parties have? begun to select nominal-tier candidates from among leaders of prominent local organizations or locally based minor parties. Given the absence of any formal mechanism for local accountability in the pre-existing Bolivian closed-list PR system, this result is a clear testament to the role of nominal voting if facilitating the articulation of local interests. In Venezuela, Crisp and
15 Germany had a system of exclusively nominal voting before World War I (s$e. Scarrow, Ch. 3), but the interwar Weimar Republic used a pure list-PR system.
jCulisheck find that legislators elected from the nominal tier are more likely to view their role as representing their constituents than are list-tier representatives (Chapter 18). That these legislators have begun to undertake even minimal responsiveness to their districts in the face of highly centralized nomination processes of the major parties again shows the power of the incentives offered by nominal voting.
In Italy, D'Alimonte found that the quality of individual candidates and their ties to localities do not seem to matter in elections (Chapter 15). However, this result is not an indictment of the mixed-member system per se, as it stems from the nature of multiparty alliances that negotiate their SSD nominations prior to elections. Unlike the German and New Zealand party systems, Italy's two-bloc system does not currently consist of two large parties that win the bulk of the SSDs. Rather, it consists of alliances of parties that must decide which candidate receives their endorsement in each SSD. Thus, D'Alimonte notes, party leaders effectively exercise more control over members than under the old (open-list) system. It is too early to say for certain whether this will remain a permanent feature of the system, or whether these alliances will be transformed into parties themselves.
Although local accountability of legislators is a desirable feature, it is possible to carry this too far. For example, a legislator elected from a SSD might behave as nothing more than a local delegate, even at the expense of the national interest. Cases of such extreme "localness" are frequently found in situations where members are elected without a party affiliation. Nonpartisan candidates may take seats whenever an electoral system employs exclusively nominal votes, as is the case in SSDs.16 The drawback of nonpartisan or independent legislators is that they are likely to be more susceptible to trading votes with the executive or party leaders—offering their votes in exchange for patronage or pork that is distributed to their constituency. Such members are rarely held accountable for their votes on national policy issues; their local accountability typically revolves around their parochial defense of local interests and their distribution of goods to their districts (Cox and McCubbins 2000; also Carey and Shugart 1995).
The Russian case is particularly instructive on this point. As Moser notes, party identification is extremely weak in the Russian Federation. As a result, there has been a great potential for independent candidates in the nominal tier. Nonpartisan legislators in Russia have been more willing to engage in ad hoc deals with the president involving pork and patronage than have list deputies. Nonetheless, Moser notes that the number of nonpartisan candidates elected declined substantially from 1993 to 1995. Moreover, the prominence of nonpartisan candidates is more a product of the newness of
u> This is also the case in multi-seat (non-list) plurality and semi-proportional formulas, including SNTV. Indeed, there were always numerous independent candidates elected in each of Japan's postwar elections under SNTV.
590
Conclusion
Are MM Systems the Best of Both Worlds?
591
democracy in Russia than of the MMM system. The presence of the list tier has encouraged parties to be much less parochial than they could have been in a purely nominal voting system. At the same time, the presence of SSDs ha^ ensured against overly centralized parties that might have emerged in a setting of such new democratic experience under a pure list system.
In general, then, we can say that there is a high degree of personalization and local representation in mixed-member systems. With the possible exception of Italy, and allowing for the special case of Israel, local accountability exists via the nominal tier, as expected. Even more striking is that list-tier legislators in some cases—particularly Germany, Japan, and New Zealand-develop ties toward constituencies even though (closed-) list representation would normally work against such ties. The most likely reasons for this are that parties want to retain a high profile in the SSDs and thus encourage their list legislators to respond to local interests, and that the legislators themsel.^. hope to win an SSD race in the next election. The best-loser provision such as that in Japan would only enhance this tendency for representation of local interests to spill over from the nominal to the list tier.
National parlies
All else being equal, list voting is more likely to promote parties of national scope than purely nominal voting because parties pool their votes over a broad multiscat district. If the lists are closed, parties must develop a reputation—usually policy-based—throughout the entire district and cannot get by on the strength of popular candidates. Thus, list voting is an antidote to the potential for legislative parochialism in purely nominal electoral systems.
The adoption of mixed-member systems in countries that previously used exclusively nominal voting in SSDs, such as New Zealand and Russia, offer the clearest examples of the effects of adding a list tier. In the case of New Zealand, there has been a great increase in the number of parties physically present in parliament, from an average of 2.6 in the 1969-93 period to 6 after the first MMP election in 1996. This increase primarily reflects the effect ul' the MMP system on the mterparty dimension, but it has important consequences for the intraparty dimension as well. No longer must a party with national priorities carefully nominate its leaders in those few districts where a local plurality is within reach; it can now act as a genuinely national party by focusing its campaign on gaining votes throughout the country to maximize its list representation.17 In Russia, Moser argues that the list tier has served
17 There is, however, still the possibility, under the New Zealand law, of a party falling below the 5% threshold and yet gaining PR seats if it has managed to elect one candidate in an SSD. So targeting a favorable SSD remains one option for a small party in New Zealand. Our point is that a party that can attain 5% nationally (and thus be assured of representation) need not concern itself with finding a local constituency that particularly favors it. It can be represented exclusively as a minor national party, an option that was obviously not present before the reform.
its intended role of encouraging the development of national parties, prominent regional politicians in Russia have seen it in their interests to place resources into developing a national partisan identity in order to obtain scats from the list tier.
A final noteworthy case is Italy. One of the challenges confronting Italian reformers was the growing regionalization of the party system. The new Italian MMM system increases the opportunities for parties to gain seats outside their principal region more so than a purely majoritarian system would have done. In other words, although a pure SSD system encourages the aggregation of multiple parties into two blocs—as in France after 1958 (Duverger 1986)—such a system only enhances the regionalization of a party system, compared with list PR.18 The list tier—especially in the Italian lower house, where seats are allocated nationwide—permits parties to make use of votes outside of the regions in which they dominate regional competition. The ability of the list tier to promote greater nationalization of the party system than would be feasible for a pure SSD system is also a desirable feature of Mexico's electoral law, and could make a MM system beneficial for Canada as well.
Assessing Mixed-Member Systems
In short, the material presented in this book suggests that mixed-member systems have been largely successful thus far. They appear to be more likely than most other electoral systems to generate two-bloc party systems, without in the process reducing minor parties to insignificance.19 In addition, they are more likely than any other class of electoral system to simultaneously generate local accountability as well as a nationally oriented party system. Other electoral systems may generate a mix of these attributes, but generally not without introducing features that may be seen as undesirable, such as intraparty competition and factionalism, which frequently characterize STV or open-list PR.
Mixed-member systems simultaneously encourage divergent incentives that lead party systems to exhibit many of the features of the "efficient" and desirable balance. On the interparty dimension, they permit some parties to specialize in SSDs and others to specialize in lists. On the intraparty dimension, they permit some legislators to specialize in appealing to local interests while others attend to their party's national priorities.
18 Indeed, despite France's vaunted political centralization, its party system is rather decentralized, with different party labels often being used in different parts of the country (and in different elections) within the center-right alliance of parties.
19 They are not the only electoral systems that can produce this balancing act—small-wagnitude PR as in Greece and Spain or STV as in Ireland can do so as well— but MM systems are probably, as a class, more likely to do so than other types.
592
Conclusion
Are MM Systems the Best of Both Worlds?
593
Achieving this balance of electoral features may well be more likely jri MMM than MMP systems, Under MMP, parties need not perform well in' nominal-tier voting to be assured their full proportional share of seats Jn contrast, a party that fails to have attractive candidates or to articulate local interests under MMM will see its seal share signilicantly trail its national vote share. An obvious disadvantage of MMM, however, is found on the inter-party dimension, where such systems are less proportional than MMP and hence under-represent smaller parties. This conundrum can be averted by the use of vote transfers between tiers. As explained in Chapter 1 and practiced in; Hungary and Italy, an MMM system can be made less majoritarian by transferring votes from the nominal tier to boost smaller parties' list-tier allocations. Alternatively, vote transfers for parties winning SSDs could be used in MMP systems to reward parties for success in the nominal tier without sacrificing a high degree of proportionality.20 Such issues take us into the realm of fine-tuning mixed-member systems, to which we now turn.
No electoral system is free of disadvantages. Even if mixed-member systems did indeed provide the best of both worlds—as we think they often do—-they nonetheless draw criticism on several accounts. Of all the criticisms of mixed-member electoral systems, perhaps the most serious is that they are too complex. Simplicity is usually held to be a virtue in electoral systems (Taagepera and Shugart 1989), and MM systems have been criticized as introducing unnecessary complexity. If the connection between voting behavior and seat outcomes is not readily apparent, the legitimacy of the electoral system may suffer. Some studies have noted that voters in Germany, the longest established MM system, do not really understand whether the nominal vote (erstimme) or the list vote (zweitstimme) is more important (Kaase 1984:163). At the same time, as Klingemann and Wessels show in Chapter 13, German voters have managed to make good strategic use of their two votes over the years, even if they do not fully understand the mechanics of their MMP system.
Although some individual mixed-member systems are quite complex (e.g. Hungary), in general these systems are not necessarily more complex than many other frequently employed systems. Explaining the details of the d'Hondt divisor method of seat allocation or the two-stage allocation of remainders in many European PR systems takes some patience on the part of the student, as well as the teacher. None of these systems has the simplicity of an SSD plurality system, yet the manufactured majorities—and especially the occasional reversed pluralities—of SSD systems arc themselves complex
20 Although no system currently practices vote transfers within MMP, one way of doing so would be to augment a party's list-tier votes when it wins an SSD with a candidate who outperforms his party's list vote in the district. For instance, if the parly wins 42% of the. list votes in an SSD in which its candidate wins 49%, the votes that put the candidate over 42% would be added to the party's list votes.
outcomes requiring explanation and justification. Against the experience of other countries with allegedly "simple" electoral systems, mixed-member systems are not so complex after all. Thus, we would argue that complexity is really a non-issue with respect to MM systems.
In addition to complexity, various other criticisms of these systems surface from time to time. Prominent among these are the absence of choice among candidates on the list tier, and the closely related problem of SSD losers being admitted into the legislature. Both criticisms are most likely to arise in countries with no prior history of closed-list voting, where it is taken for granted that only popular candidates enter parliament. We discuss each of these criticisms in turn, and then potential solutions.
There arc currently no MM systems that provide voters a choice among candidates within their preferred party.21 If the Jenkins Commission proposal for Britain were adopted, we would have a case of a MM system with a preference vote on the list. While intraparty competition has the drawback of encouraging factionalism and thus can undermine party cohesiveness, closed-list voting may be politically unpalatable in some countries. For example, it is clear that the Jenkins Commission proposed a preferential vote in the list because of criticisms from the Conservative Party and some of the media about the use of a closed list to select British members of the European Parliament. It is also likely that in countries with a history of relatively fluid parties—a feature found particularly in some presidential systems, such as Brazil, Colombia, or the United States—adopting closed lists would be politically difficult, even as part of a MM system that retained legislator-voter contact through the nominal tier.
Another factor of MM systems that is sometimes seen as undesirable is the likelihood that candidates who lose their SSD races can be "saved" by means of a favorable position on the party list. In Japan, for example, there were many complaints in the media about politicians who failed to win a constituency nevertheless ending up in the first Diet selected under MM electoral rules. Such results do not necessarily imply that the electoral system is flawed, as there are clearly benefits to keeping a party's leaders in parliament, even if they are not proven vote-getters. However, this feature of MM systems can be a public relations problem for party leaders who lose in their districts, and it may call into question the legitimacy of the electoral system in countries with no prior experience with list voting.
21 Israel has experimented with primary elections, but these have thus far been open only to party members (Hazan, Ch. 16). Thus, they offer a genuine choice over a party's candidates only to a subset of those voters who support the party's list in the general election. Some parties in Venezuela used primaries in connection with the MMP elections of 1993 and 1998 (Crisp and Kulisheck, Ch. 18). Primary elections are one potential way in which countries with mixed-member systems might enhance voter choice. However, further research is needed into how they would operate under different conditions than those found in the USA, such as under parliamentary government.
594 Conclusion
Fortunately, there are ways of dealing with the various criticisms of mixed member electoral systems. For example, there is no reason why the list tier of a mixed-member system cannot operate under an open list or, preferably from our point of view, a quasi-hst or flexible-list format.22 Such a set-up mpv impair party cohesion compared with closed lists, but where creating cohesiv» parties is not necessarily a goal shared by reformers, or is politically unfeasi-blc, an open-list MM system at least ensures proportionality beyond th" norm of purely nominal-vote systems. In pre-existing open-list PR systems on the other hand, adopting a MM system could actually increase party cohesion relative to the status quo, even if the open list is retained for the list tier This is because the SSDs would be likely to encourage factions of parties to agree to common candidates in the nominal tier.
If there is concern about politicians who lose their district race getting into parliament, this can easily be dealt with by prohibiting double inclusion, i.e. simultaneous nominal- and list-tier nomination of a given candidate. Parties would thus have to decide which tier each candidate was most likely to win from, and candidates would not be able to hedge on a possible SSD loss through a high position on the parly list. As far as we know, only Thailand has ever banned double inclusion, and Mexico is the only other system to have imposed any legal limits on it (see Chapter 20 by Weldon). The rarity of limits on double inclusion is probably attributable to elite preferences for maximum opportunities for the success of individual candidates.23 Short of banning double inclusion, using a best-loser provision in the list tier (as in Japan and the Italian senate) ensures that the SSD losers who are elected are relatively more popular than other losers of their party. Such a provision has
22 Under an open list, voters have the option of casting a nominal (preference) vote, but only nominal votes count toward determining the order of election of a party's candidates. In a quasi-list system, the nominal vote is mandatory, and such votes agam exclusively determine the order of election. Under flexible lists, there is a party-ordered list, but voters have the option of casting a preference vote. Voters who cast a party-list vote are assisting the election of candidates in the order they appear on the ballot (unlike voters who cast a party vote in an open-list system). However, in a flexible-list system, candidates who obtain an established quota of nominal votes can move up ahead of copartisans whom the party ranked higher. The preference vote proposed by the Jenkins Commission is understood to be a flexible list (personal communication with David Farrell), even though the Commission's report uses the term "open list".
23 Another way of dealing with the criticisms launched against both closed lists and the "saving" of SSD losers via the list tier is a hybrid MM-quasi-list system. For instance, the ballot could list each candidate of a given list-tier district—save for the one who is nominated in the corresponding SSD and require voters to cast their list vole by indicating a preference for one candidate within their preferred party. Candidates who lost in a SSD would thus be ranked by their party's voters outside their own district, rather than by party leaders or on the basis of how well they performed in their losing SSD effort (as in the Italian senate). Such a hybrid SSD quasi-list system would not overcome the disadvantages of intraparty competition (e.g. factionalism and parochialism), but it might have appeal in settings where criticisms of both the closed list and the election of SSD losers prevent the adoption of a MM system with a closed-list tier.
Are MM Systems the Best of Both Worlds? 595
the further benefit of reducing the number of SSDs that will be "safe" seats, as candidates will have an incentive to cultivate ties to districts even if they are unlikely to win the district plurality (see Chapter 17 by Reed and Thies).
We would not recommend any specific variant of the mixed-member format as a one-size-fits-all solution. In fact, one of the appeals of MM systems is that they lend themselves to an even greater range of fine-tuning than the various "pure" electoral systems. In general, a system should avoid excessive complexity, but some complications are inevitable when the goals themselves are complex, as in the case of trying to have the best of both worlds. The pure jvlMM and MMP systems are the most straightforward variants of this class of electoral system. Employing vote transfers within either of the primary types may be a justifiable further complication to balance the desired degree of proportionality with an incentive for major parlies lo commit resources to both tiers. Moving beyond the closed list for the list tier may be both unnecessary and undesirable in most cases. Yet it may be a necessary complication to move beyond an all-SSD or an open-list system that is perceived as in need of repair. Our general point is that MM systems permit myriad variations that can suit a specific political context, while still holding out the promise of providing the best of both worlds—i.e. the best of both identifiable governing blocs and proportionality, and the best of both local accountability and cohesive and programmatic national parties.
PROSPECTS FOR CONTINUED SPREAD OF THE MIXED-MEMBER IDEA
Although mixed-member electoral systems offer ihe best of both worlds, typically they completely satisfy almost no one. As MM systems invariably reflect a set of political compromises, this is hardly unexpected or undesirable. Thus, we should not be alarmed that many of the countries that have adopted MM systems arc considering further change to their electoral law. Most of the discussion has assumed the nature of mid-course corrections rather than abandoning the mixed-member format entirely. This is true even in Venezuela, despite the drastic overhaul of the entire constitution that has taken place. Thus, if New Zealanders decide that their MMP system gives too much power to small parties, they can fairly easily reduce proportionality by-adopting vote transfers to boost the representation of parties that win many SSDs, or by switching to MMM.24 If Russians come to believe that too many
24 They could also increase their "loophole" threshold. For example, instead of allowing a single SSD victory to open up list representation for a party even if its list vote is below the national threshold, they could require two or more SSDs, or a combination of one SSD and some national vote total lower than the regular threshold. Or a party that wins one or more SSDs but not a sufficient list vote could simply be denied any seat other than its SSDs.
596
Conclusion
independents are being elected with small vote shares, they can reduce the ratio of members elected via SSDs or adopt a two-round majority system in the SSDs. In short, MM systems allow for continued flexibility in the design of electoral laws. We see this feature as an asset that countries ought to value: with an MM system the electoral laws of a country can be regularly updated to suit changing needs and political conditions.
It is our expectation that the advantages offered by mixed-member systems will lead more countries to follow this worldwide trend. Countries experiencing some sort of systemic failure should continue to be the most likely candidates. However, perhaps most significant will be whether there is a second wave of adoptions from countries not facing any particular crisis, but simply looking to improve their system of government. For example, in the fall of 1999 a new Portuguese administration came to power committed to political reform, pledging among other things to create single-seat constituencies to boost voter identification with politicians. We suspect that many Portuguese legislators will not want to totally abandon the list-based PR system that elected them, and hence a MM system should arise as the natural bargain. Britain is another country where a MM system might be adopted through the course of normal (i.e. non-crisis driven) political reform. As David Farrell suggests in Chapter 23, this process in Britain may well simply represent a desire to join the modern political world.
There comes a time in the diffusion of innovation, when various countries start to adopt new policies or laws because it is generally recognized that these represent an improvement over existing options. We may soon be at this juncture for the spread of MM systems. Regardless of whether mixed-member systems continue to proliferate, they have now joined majoritarian and proportional systems as basic options which must be considered whenever electoral systems are designed or redesigned. Such a development represents a fundamental change in thinking about electoral systems around the world.
Glossary
Items in italics indicate cross references.
Alternative vote. A form of majoritarian allocation in which voters may rank candidates in order of preference (see transferable votes). Uses sequential elimination to determine winners, usually in single-seat districts (SSDs). Also known as 'instant runoff. Used in most states for elections to the Australian House of Representatives. Best loser. A provision in some mixed-member systems by which seats in the list tier are allocated to candidates who have failed to win their seats in the nominal tier, but have obtained higher vote totals or run closer races than other losing candidates of the same party (or other political organization). Less common than the closed list. Closed list. A system in which the voter must cast a single vote for one list of candidates submitted by a party (or other political organization). No nominal or preference votes are permitted (contrast with open list). Candidates are elected in the order in which they are ranked on the list as submitted. Used in proportional representation systems in Argentina, Spain, and elsewhere, and for the list tier of most mixed-member systems.
Compensatory allocation. Distribution of seats in an upper tier of an electoral system so as to compensate parties for deviations from proportionality that result from the use of a small district magnitude or majoritarian allocation in a lower tier. Cumulative vote. An electoral system in which each voter may cast a number of nominal and nontransferable votes up to the number of seats at stake, and may give more than one vote to any given candidate. Sometimes erroneously classified as a system of proportional representation. Used in some local elections in the US. Deviation from proportionality (D). An index of the extern Lo which a given distribution of seats deviates from perfect proportionality to parties' vote shares. Calculated as:
D = (l/2)2|(AV-r,)|,
where s, is the seat share of the i'-th party, vt is its vote share, and £ (sigma) stands for the summation over all parties involved.
District magnitude (M). The number of seats allocated in a given electoral district. Infective number of parties (N). An index that indicates the number of hypothetical equal-si7ed parties that would have the same effect on the fractionalization of the party system as do the actual parties of varying sizes. Calculated as:
N=l/Ip,-2,
where pt is the fractional share (of votes or seats) of the /-th party and S (sigma) stands for the summation over all parties.
Htxible list. A form of list proportional representation in which a voter may cast either one or more preference votes or a single list vote. Candidates are ranked by the party