The patterns of party polarization in East Asia.

Dalton, Russell J. ; Tanaka, Aiji

The alignment of parties within a party system shapes the nature of electoral competition, the process of representation, and potentially the legitimacy of the system. This article describes the distribution of parties and the levels of party polarization in the party systems of East Asian democracies. We examine the public's perceptions of party positions on a left-right scale to map the pattern of party competition. The evidence is based on two waves of surveys from the Comparative Study of Electoral Systems. We describe considerable variation in the polarization of Asian party systems, which has direct implications for the clarity of party choice and the behavior of voters. We conclude with a discussion of the implications of our findings.

KEYWORDS: polarization, left-right, political parties, spatial models

**********

For a party system to offer reasonable and meaningful political choices to the voters, and for democratic elections to function, the parties must present significant policy choices and distinct governing programs. Party systems should reflect the enduring structure of democratic competition and thus should represent the significant political and social cleavages in the nation. For instance, research on the formation of democratic party systems in Europe argued that parties provided an opportunity for existing social divisions to find political expression--and resolution--through electoral competition. (1) Without such content and a clear programmatic structure, elections could become ephemeral expressions of opinions or personality contests, rather than instrumental acts of governance. This logic is embedded in the discussions of electoral system reform that Benjamin Reilly described in the previous article.

Furthermore, the relative positions of parties in terms of cleavage or ideological dimensions have fundamental implications for electoral politics. For instance, Giovanni Sartori maintained that the degree of party polarization affected the patterns of governing and the legitimacy of the democratic process. (2) Extremely polarized systems typically experience higher levels of political conflict and popular protest. (3) In contrast, converging party positions may reflect the ability of government to build consensus among competing political factions.

Yet, the nature of party alignments in East Asian party systems remains uncertain. These party systems are incredibly diverse in their organizational structure, the number of parties, and the types of parties that compete (also Reilly, this volume). As others in this collection have noted, most of the party systems in this region are characterized by fluidity and volatility. But does this fluidity reflect shifts in voter preferences for parties with distinct policy profiles or the lack of clear programs for voters to choose among? In other words, the clarity of party positions remains uncertain, and much of our present evidence is based on the observations of political analysts rather than the perceptions of democratic electorates. Sartori's observations about the potential impact of party system polarization on the democratic process may be especially relevant as East Asia party systems attempt to consolidate their democratic systems.

This article examines the structure of party systems in East Asian democracies. We first describe the distribution of the public along the left-fight dimension. Like other spatial modeling studies using the left-right scale, we presume that left-right positions summarize citizen and party positions on the political issues of relevance in a nation, even though the term left-right can have different meanings for different voters. (4) We next describe how these electorates position the major parties on the left-right scale, which maps the pattern of party competition. Third, we use these electorate and party positions to describe these party systems in terms of their polarization and representation. Finally, we conclude with a discussion of the implications of our findings.

We analyze data from the Comparative Study of Electoral Systems that surveyed citizens across several Asian democracies. (5) This is the first comparative election survey of East Asian democracies and now includes two waves of surveys. The findings systematically compare how East Asian publics perceive the structure of party system competition and describe the party positions within these systems.

The Left-Right Framework

While each election comprises a unique set of issues, candidates, and activities, party systems should provide a stable structure for discussing and addressing the broader social conflicts in a nation. For example, Seymour Lipset noted that the conflict between the economic "haves" and "have-nots" seemed a fundamental element of most democratic party systems; Lipset and Stein Rokkan developed a more comprehensive framework that described the evolution of Western party systems in terms of the social cleavages existing in these societies. (6) New quality of life issues or issues of nationality provide other potential sources of social and party division. As new party systems have formed during the Third Wave of democratization, this raises the important question of whether the parties offer distinct policy choices in these new democracies. (7)

Clearly, there are many different frameworks (ideological or social group competition) that might structure party competition. Often these multiple frameworks are combined (or compressed) into a unidimensional structure of party competition. This approach is most clearly identified with Anthony Downs's spatial modeling of party competition and research that builds on his framework. (8) In established Western democracies, party alignments are often described in terms of the left-right dimension. The logic of this Downsian spatial model is that parties and voters can be aligned along this dimension to map the pattern of party competition. Even if this is an oversimplification of political reality, it provides a good first approximation of the nature of party competition. (9) Knowing whether a party is "left" or "right" helps citizens to locate themselves in relation to the party. Left-right position thus provides a cue to orient individuals to political issues and personalities. When citizens face a choice at election time, the Downsian logic implies that they will select the party that is closest to their own position along this continuum. A leftist will select a party nearby on the left end of the continuum, while a rightist will select a party near his/her position. Moreover, the distances between voters and parties along the dimension can predict the likelihood of voting and satisfaction with government policies. Thus, this is a theoretically powerful model of party competition.

Both the Downsian spatial analyses and survey researchers using the left-right framework do not presume that citizens have a deep understanding of ideological concepts embedded in these concepts of left and right. Instead, such a simple structure provides a shorthand for summarizing the enduring issues of political debate in a nation. (10) And the left-right framework is not limited to traditional socioeconomic issues. For instance, to a Japanese blue-collar worker, left may mean social welfare policies; to a young Japanese college student it may mean environmental protection and other quality of life issues. Along with other public opinion researchers, Ronald Inglehart finds that most citizens in most nations can locate themselves on a left-right scale, and he describes the scale as representing the "major conflicts that are present in the political system." (11) Even if the specific definitions of left and right may vary across individuals and even nations, the simple structure of a general left-right scale provides a basis of comparison. (12)

When one turns to East Asian democracies, however, the existence of a simple left-fight framework of party competition is uncertain. Previous research suggests that parties in several Asian democracies have not developed strong institutional ties to a social base or a firm ideological identity. Most citizens in East Asia can position themselves on a left-fight scale, which implies they can relate to these terms, but the content of this framework is less clear. For example, the most extensive evidence exists for Japan. Several dimensional analyses have identified what appears to be a left-right framework to the Japanese party system; (13) the Japanese Communist Party (JCP) and other leftist parties are at one end of this continuum, and the Liberal Democratic Party (LDP) at the other. But at the same time, few analysts maintain that a Marxist-capitalist debate or other social cleavages are strongly related to electoral choice in Japan. (14)

The evidence from the newly consolidated democracies of Korea and Taiwan is also ambiguous. Doh Chull Shin and Byong-Kuen Jhee show that the left-right terminology is widely accepted and these orientations are primarily linked to political and economic reform attitudes. (15) A complicating factor is the turnover in Korean parties; new parties constantly entering the system make it more difficult to position parties along a left-right dimension. Most Taiwanese can also position themselves on a left-right scale, yet there is less consensus on the meaning of this dimension in Taiwanese politics. National identity issues are strongly related to Kuomintang (KMT) and Democratic Progressive Party (DPP) support, while polarization on economic and social welfare issues appears modest. (16) The relative weight of distinct issues is uncertain, and perceptions of party differences may have moderated over the past two decades. (17) For the Philippines, there is limited evidence that social cleavages (and the policies they reflect) actually structure party positions. (18)

However, systematic cross-national comparisons have been limited because national studies use different methods to assess the structure of party competition and the meaning of this structure. Even if a clear description of party alignments is available for one nation, the lack of similar analyses for other East Asian nations has impeded cross-national comparisons--and limited our understanding of how party systems may be similar or different. Therefore, we will systematically compare the framework of party competition across East Asian party systems using the left-right dimension. We realize the meaningfulness of the left-right dimension in East Asia is still an open research question, but this can be evaluated by assessing its ability to increase our understanding of these party systems.

Measuring Left-Right Positions

Until recently, it was difficult to compare party systems in the relative position of parties and the overall level of polarization because this required measuring the ideological position of parties as well as their vote shares. Typically, researchers estimated polarization from indirect indicators, such as the number of parties in an electoral system, the size of extremist parties, or the vote share for governing parties. Some previous studies coded parties into ideological families (e.g., communist, socialist, centrist, fascist) and used this to categorize the diversity of party systems. (19) These methods provide broad approximations of the actual ideological position of parties, but treat all parties of a family as identical and differences between families as equal interval differences. Another option is the use of party manifestos or expert opinion to estimate parties' left-right positions. (20) However, the comparative manifesto project focused on the salience of issues rather than party positions, and thus there is debate about the validity of this methodology. (21) The party manifesto study also included only two East Asian nations in their project, and few expert studies include more than one or two East Asian nations.

We therefore turn to another source to measure the positions of voters and parties: the perceptions of the electorate in the nation. The Comparative Study of Electoral Systems (CSES) is a cooperative international project that asks a common questionnaire in the national election studies of many contemporary democracies. The CSES includes surveys of Japan (1996, 2004), South Korea (2000, 2004), Taiwan (1996, 2001), and the Philippines (1998, 2004). (22) We also examine Australia (1996, 2004) and New Zealand (1996, 2002). These two nations provide a reference standard for comparing East Asian party systems with those of two established Western democracies. (23)

The CSES asks respondents to position themselves along a left-right scale using the following question: (24)

In politics people sometimes talk of left and right. Where would
you place yourself on a scale from 0 to 10, where 0 means the
left and 10 means the right?

0 1 2 3 4 5 6 7 8 9 10
Left Right

The ability to position oneself on the left-right scale does not imply that citizens possess a sophisticated conceptual framework or theoretical understanding of liberal-conservative philosophy. We simply expect that positions on this scale summarize the issues and cleavages that primarily structure political competition in a nation. The meaning of this dimension can, and indeed likely does, vary across nations. To one person, political identities may be shaped by economic concerns; for another it may involve social issues, or other matters. The left-right dimension thus provides the metric for our cross-national comparisons.

Initial evidence on the validity of the left-fight dimension comes from the percentage of the public who can position themselves on this dimension. Across most East Asian systems, large majorities in the second CSES wave can locate themselves in left-fight terms: Korea (83 percent), Japan (91 percent), and the Philippines (97 percent). These levels compare favorably to left-fight understanding in both Australia (82 percent) and New Zealand (81 percent), where this terminology is widely used by citizens and elites. Only the Taiwanese public is hesitant to place themselves on the left-fight scale; but even here, almost half have a left-right position (48.5 percent). (25) The generally high levels of left-fight self-placements across Asian party systems are consistent with evidence from the World Values Survey that asked the same question in many of these same nations. (26)

The distribution of left-right attitudes among each national sample is presented in Figure 1. The horizontal bars in the figure represent a two standard deviation range of individuals along the left-right scale--that is, an estimate of the ideological range that encompasses approximately two-thirds of the public. The vertical hash mark on each bar shows the mean score for respondents who place themselves on the scale.

[FIGURE 1 OMITTED]

The figure shows that most publics tend to locate themselves near the center of the scale, with most national mean scores ranging between 5.0 and 6.0. The Korean public falls outside of this range to the left (mean = 4.89) and becomes more leftist between 2000 and 2004. The Philippine public, in contrast, is the most conservative (7.16), with 29 percent placing themselves at the rightmost category in 1998. The range of left-fight opinions (i.e., the length of the horizontal bar) tends to be fairly similar across nations. The Korean and Philippine publics are slightly more polarized than average, and the Taiwanese public is slightly less polarized--but these differences are fairly modest. In addition, longitudinal comparisons for the Asian nations find that the changes between surveys are fairly modest.

The ability of most East Asian citizens to locate themselves along a left-fight dimension is a first indication of the validity of this scale as a method of political orientation in East Asia. This provides a common metric. However, we also recognize that the content of these self-identities varies across nations. (27) In Australia and New Zealand, left right is strongly linked to economic issues and traditional/modern values. In contrast, economics plays a significantly weaker role in structuring left-right identities in East Asia, and cultural factors seem more influential. However, what is important for our analyses is that the left-right dimension summarizes important aspects of political cleavage in a political system, even if they vary across nations.

Mapping Party Positions

The next step of our analyses is to map and compare the position of political parties along the left-right dimension. Besides asking individuals to locate themselves on the left-right scale, the CSES asked them to position the major parties in the just-completed election. Since most articles in this collection study citizen political behavior, the public's perceptions of the parties' positions should be the framework that guides their voting choices and other political judgments, more so than alternative estimates of the party space from experts or party manifestos.

Figure 2 displays the mean location of the voters as the thick black arrows. In addition, it plots the citizens' placement of the major political parties in both elections, with the thickness of the arrow indicating the party's approximate voter share in the election.

[FIGURE 2 OMITTED]

One measure of the validity of the left-right framework is the ability of the public to use this scale to position the parties. Even in established democracies of Australia and New Zealand, approximately 20 percent of the public cannot locate the two major parties on the left-right scale, which is typically cited as evidence of limited political engagement or sophistication by some individuals. By comparison, the levels of party recognition in most East Asian democracies are quite similar, with 80-90 percent locating the major parties on the left-right scale. (28) Taiwan again lags in the recognition of left-right terminology, but half the public still locates the DPP and KMT on this scale.

In terms of the actual placement of parties, in the 1996 Japanese election the public perceived the conservative LDP toward the right end of the scale, while the JCP was located at the very left end of the dimension. The average voter (5.38) is slightly to the right of center. This is an example of a polarized system, because the major party competitors were seen as quite distant from the medium voter. Although the NFP (New Frontier Party) and the DPJ (Democratic Party of Japan) were closer to the average voter, the vote shares of these two parties combined do not match the votes gained by the conservative LDE

In 2004, the same general pattern persists. The NFP was dissolved and some of its members merged into the DPJ and others went back to their original body of Komei, now called the NKP (New Komei Party). The LDP and the NKP formed a coalition to govern, but the public perceives their government as still far to the right of the average voter. And while the LDP and JCP were seen as moving slightly toward the center by the 2004 election, the small parties in the center moved toward the ideological poles. Consequently, the overall polarization of the Japanese party system persisted from 1996 to 2004.

Perhaps the most dramatic restructuring of the party system has occurred in Korea. In the 2000 election, the Korean system represented an anomaly in terms of Downsian party competition. The second panel in Figure 2 shows that the Koreans on average located themselves just to the left of center of the left-right scale (4.89). However, Koreans saw virtually all the significant parties as positioned to their right. The Millennium Democratic Party (MDP) was the major liberal party, and the Grand National Party (GNP) was the major conservative party. Both of these parties were positioned half a scale point or more to the right of the average Korean, along with the United Liberal Democrats (ULD), the Democratic People's Party (DPP), and the small New Korean Party of Hope (NKP). Thus, rather than a balanced choice between left and right parties, in 2000 the voter saw only degrees of conservative choice. (29) The election was fought over evaluations of the presidency of Kim Dae-jung, who led the MDP and was opposed by the GNE

Moo-hyun Roh's election to president in 2003 as a representative of the MDP sowed the seeds of a rapid restructuring of the Korean party system. Roh's policies as president and the parliament's attempted impeachment of Roh in 2003 dramatically polarized the party system and shifted Korea's center of gravity to the left. Between 2000 and 2004, the average voter moved three-tenths of a scale point to the left. The MDP sustained large losses in the backlash following its leading role in the impeachment of Roh. The GNP, which also favored impeachment, saw its vote share drop slightly. Moreover, the public perceived both of these parties as moving dramatically to the fight in 2004; the MDP was seen as shifting 0.81 point and the GNP by 1.67 points (a dramatic shift by a major party). Counterbalancing these trends, the newly formed Our Party (UD) supported Roh and emerged with the largest vote share and the largest number of parliamentary seats. The UD's support for closer ties with North Korea and its identification with Roh led voters to place it well to the left (3.72) on the scale.

In Taiwan, the right means a position closer to the traditional authority (the KMT), whereas a leftist orientation means a position farther away from the traditional authority. Therefore, the left-right dimension does not fully capture party positions on policy orientations. The Democratic Progressive Party (DPP) is the main contender to the KMT. The TSU (Taiwan Solidarity Union) is very strongly oriented toward independence from China, while the PFP (People First Party) and the CNP (Chinese New Party) are strongly oriented toward unification with China. Indeed, analysts of Taiwanese politics often frame political competition in terms of the nationalist issue, which presumably shapes left-right identities. Therefore, the left-right dimension of the party system in Taiwan coincides with the feelings of anti-(left) and pro-(right) traditional authority, plus pro-independence vs. pro-unification.

The two major parties, the KMT and the DPP, are seen as converging slightly between 1996 and 2001, but this was already a centrist party system in ideological terms. A set of new parties emerged after the 1996 election, and they contributed to a slight increase in party diversity. (Subsequent to the 2001 survey, there was an apparent polarization of the Taiwanese parties as a result of the Chen administration and reelection.)

One of the most surprising party landscapes exists in the Philippines. Philippine democracy has experienced a continuing struggle since Ferdinand Marcos was deposed in 1986. Even though several national elections had occurred by the 1998 survey, the party system remained fluid and weakly structured. (30) In 1998, the LAMMP (Struggle of the Patriotic Filipino Masses) party received the most seats, and Joseph Estrada was elected president as head of the party. Lakas-NUCD (Lakas--Christian Muslim Democrats) was the previous governing party, and other parties represented liberal (LIB, AD) or regional (PROMDI) interests. However, the distinctive feature of the Philippine system is the location of the average voter to the right of all the significant political parties! This is an anomaly that Downs would not have predicted possible, at least as a sustained element of party competition.

Yet, the 2004 election also displays an un-Downsian party space. The party space had shifted to the left, possibly in reaction to Estrada's impeachment and his replacement in office by Gloria Arroyo. The Nationalist People's Coalition (NPC) included elements from the LAMMP as the major opposition party, and Filipinos saw this party to the left of the LAMMP position in 1998. The net result is that there is now an interpretable left-fight ordering between the NPC and Lakas. However, there is substantial overlap between the Philippine parties, with citizens seeing little distinction between their electoral choices. Moreover, the average Philippine voter is still positioned to the fight of all the established parties.

We suspect that this pattern reflects a combination of several factors. The Philippine voters themselves may embrace "the right" as a normative statement or a reflection of their religious beliefs more than an ideological position. (It appears that the only other nation in the CSES that follows a similar pattern is Mexico, where the self-placement of the average citizen is also to the fight of the major parties.) In both CSES surveys and other cross-national surveys, the Philippine public expresses a rightist tendency, so this seems to be a persistent trait and not a passing pattern in a single public opinion survey. (31) Rightist orientations in the Philippines also may contain populist appeals that are typically identified as leftist in cross-national comparisons or in elite discourse. For instance, expert observers would place the LAMMP to the left of Lakas in 1998, but the Philippine public reversed the ordering of these two parties. If the mass of the Philippine public saw themselves as located on the right, and they perceived the LAMMP as closest to their position in 1998 because of its populist rhetoric to represent the masses, then this is a consistent ordering. Finally, elite competition in developing democracies can sometimes exist with weak ties to voter interests. Philippine parties appear more centrist than the voters at large and the public does not see them as offering very distinct policy alternatives. This fits other interpretations of Philippine elections as candidate-centered contests with little firm programmatic content. Overall, however, citizens see a substantial gap between themselves and the parties, which implies that parties are not presenting alternatives that reflect the public's political preferences.

At least briefly, we want to compare the East Asian democracies with the established Western democracies of Australia and New Zealand. Both of these party systems have been traditionally polarized along traditional class lines (and other policy cleavages), and left-fight terminology is part of common political parlance. Thus we would expect both systems to display a fairly wide span of party choices. The Australian Labor Party and Liberals were fairly polarized in the 1996 election, and the long-serving Labor government was forced from office because of the conflicts over economic reform. This began a period of heightened competition on economic reforms and then further polarization over cultural issues such as immigration and terrorism. Thus, by 2004 the two major Australian parties are even more clearly divided, and new contenders such as the Greens and One Nation have entered the party fray. Similarly, in the wake of the shift from a majoritarian to a proportional representation (PR) system, New Zealand party differences increased as the major parties sought to differentiate themselves. (32) The PR system has also allowed new parties to compete for parliamentary seats, such as the Greens and the populist New Zealand First party. Thus, the 2004 election offered New Zealanders a wide array of party choices, and eight parties won seats in parliament. Indeed, of all the party systems in the figure, the New Zealand system appears to offer the voters the widest range of choice, spanning both traditional economic cleavages and new cultural issues.

In summary, most Asian party systems follow a Downsian model of party competition. The average voter is typically located near the center of the left-right dimension, with the parties aligned to the left and right along this continuum. There are, however, two notable exceptions. In 2000, the Korean party space defied Downsian logic. Voters perceived the two major parties in similar positions on the left-right continuum, and both were located to the right of the average Korean. This presumably reflects Kim Dae-jung's decision to ally with the conservative ULD. By the 2004 election, however, a Downsian space had reestablished itself, with the new UD perceived as the leftist alternative and the GNP on the right. Even more un-Downsian is the Philippine system, where voters locate themselves on the right, but see the parties as located to their left. Such a pattern seems unsustainable, since eventually a party should move to offer a clear choice to voters. Such patterns have emerged in other new democracies and it potentially indicates the importance of personality or nonideological factors in the structuring of party competition. Such lack of ideological choice also signifies a democracy deficit in the party system.

Party System Characteristics

We have described the location of the average voter and political parties in this set of nations. In addition, when taken together these data also describe the characteristics of the party system as the combination of these elements. Party research has stressed that the degree of polarization in a system is important for several reasons. Sartori, for instance, argued that the stability and legitimacy of a democracy were related to its degree of polarization; G. Bingham Powell claimed that polarization affects the stability of a democracy. (33) A highly polarized system produces clearer party choices but presumably also is more fractious and has more intense partisan competition. In a highly polarized system, the ideological gap between winners and losers is generally greater, and the policy implications of government control are more substantial because the parties are offering more distinct programs. Conversely, a centrist party system should reflect greater consensus within the electoral process--at least in left-right terms--and less interparty conflict.

Another central element of the dimensional model of party competition is the calculation of the representation gap; that is, the gap between the average voter and the government on the left-right dimension. The democratic ideal holds that this difference should be small, so that government is close to the average voter. But the party choices presented to voters can systematically affect this representation gap. For instance, highly polarized party systems may systematically generate a larger representation gap, since alternative governments tend toward the extremes.

Typically these party system characteristics are intuited from indirect indicators, such as the number of parties in an electoral system, the size of extremist parties, or the vote share for governing parties. However, we can directly measure both characteristics. The polarization of a party system can be conceptualized as the distribution of parties along the left-right scale, weighted by each party's vote share. (34) A few large parties near the center of the scale should thus generate a low polarization score, with a number of large parties at the left-right extremes producing a high score. Similarly, the representation gap can be easily measured by the absolute value of the left-right difference between individual citizens and the governing party.

Table 1 presents estimates of polarization and the representation gap for the Asian nations in the CSES. (35) The statistics on polarization restate what is apparent from the party locations in Figure 1. In broad cross-national terms, East Asian publics perceive their party systems as less polarized than do the citizens in the established democracies and the new democracies of Eastern Europe that are included in the CSES project. (36) To the extent that clear party positions reflect and reinforce voter and social group alignments, the limited dispersion of parties in the Philippines and Taiwan may explain why Aie-Rie Lee (in this issue) finds that a set of political attitudes weakly predicts voting choice in these nations. At the same time, the greater party polarization in Japan facilitates greater voter differentiation (and a higher percentage of explained variance).

For the East Asian party systems, the most dramatic pattern is the increase in party system polarization in South Korea between 2000 and 2004 as we discussed above. The Philippine party system displays convergence between waves, which means a further narrowing of the distinctions between parties that citizens can observe. Beyond these cases, there is a general continuity in the level of polarization across CSES waves, implying that we are seeing fairly stable traits of these party systems. Moreover, the cross-national pattern does not seem clearly linked to aggregate features of party systems that are typically used to imply levels of polarization. For instance, the nature of electoral systems, old/new democracies, and the number of parties are not distinctly related to the level of polarization for this set of nations. (37)

A second characteristic of party systems is the distance between the average voter and the government--what we call the representation gap. The right side of Table 1 presents the average absolute difference between the voter and the left-right placement of the prime minister's party. The greatest representation gaps emerge in the consolidated democracies of Japan, Australia, and New Zealand, where party systems are relatively polarized and the governing parties are perceived as located toward the outer range of party choices. By comparison, the representation gap is smaller in the newer democracies of South Korea, Taiwan, and the Philippines. The 2000-2004 Korea trend is again especially striking. The anomalous party pattern in 2000 produced exceptionally low party polarization of a non-Downsian form. Even though polarization increased dramatically in 2004, the development of clearer party alternatives actually decreased the representation gap slightly.

Conclusion

Party competition in free and fair elections remains the prerequisite for a meaningful democratic order. This essay has provided a first comparative mapping of the party alignments in several East Asian democracies and comparisons to the patterns in established democracies. Our findings add both to our understanding of party competition in East Asia and to broader theories of partisan politics.

Our central analyses focused on the distribution of citizens and their perceptions of political parties along the left-right scale. By asking voters to locate themselves and the major parties along the left-right continuum, we can describe the structure of the party system as perceived by the public. There is considerable variation in these patterns across Asian party systems. This is not unusual since variation in polarization exists across European party systems as well. But it is noteworthy that the greatest party polarization exists in the consolidated democracies of Japan and New Zealand, not in the new democracies of Taiwan, the Philippines, and South Korea. This pattern is exceptional, because political conflicts in recent elections in these three latter nations imply a highly divisive political environment.

One possible explanation for the lower polarization in East Asia's new democracies is that conflicts in these new party systems are not ideologically based (but linked to personal or nonideological alignments), and thus are not reflected in left-right positions. The Lee contribution to this volume, which demonstrates the weak policy base of party support in these nations, reaffirms this point. Alternatively, new party systems typically, especially in Third Wave democracies, may begin as nonideological movements in left-right terms, and the process of democratization involves the development of clear programmatic choices in the party system. Thus, ideologically based partisan cleavages may actually increase in the early stages of democratization, as seem apparent among East European party systems.

In more specific terms, there are problematic examples of party competition that apparently violate the basic logic of Downsian party competition. The clearest case is the Philippines, where the public sees all the parties as located to the left of their own position. The Korean election of 2000 is another example of a non-Downsian party space be cause of the opportunistic strategizing of Kim Dae-jung. Such anomalies raise doubts about the ability of elections to make authoritative policy choices and for voters to hold governments accountable for their policy actions in such cases. A distinct and institutionalized system of party competition seems essential to meaningful democratic elections.

In broader theoretical terms, the degree of party polarization is often indirectly measured by the number of parties in a political system, and impressionistic evidence of their ideological differences. Our findings suggest that left-right party polarization can be quite distinct from the number of parties existing in a party system. We also expect that the institutional reforms to produce party cohesion that Benjamin Reilly described in his article will be less influential than the parties' strategic electoral choices in how they present themselves to voters. Parties may choose to diverge in majoritarian electoral systems, and they may choose to converge in proportional systems. The impact of electoral system rules appears secondary to parties' electoral calculations. (38)

Finally, our cross-national comparisons yield the surprising finding that the gap between the average voter's position and the perceived position of the governing party is greatest in established democracies. The distance from the voter to the government in the most recent Australian, Japanese, and New Zealand elections is significantly greater than in South Korea, Taiwan, and the Philippines. This, too, seems to be a consequence of greater left-right polarization in these party systems--so that with a bipolar pattern of competition, either selected government represents a more distinct ideological position that distances it from the opposition's supporters.

In summary, beyond the substantial diversity in the pattern of party competition across specific Asian party systems, these data suggest a general pattern of democratic development. Clear and enduring party choices often evolve as part of the democratization process, until voters, parties, and elites find a stable structure for electoral competition. Where this structure exists--as in Japan, Australia, and New Zealand-voters can more effectively translate their policy preferences into government programs, and clear party positions probably structure policy preferences as well. But where parties do not offer clear options, the potential for the democratic process to function is severely limited.

Notes

(1.) Seymour Martin Lipset and Stein Rokkan, eds., Party Systems and Voter Alignments (New York: Free Press, 1967); also see the article by Ian McAllister in this collection.

(2.) Giovanni Sartori, Parties and Party Systems (New York: Cambridge University Press, 1976).

(3.) Arend Lijphart, Democracies (New Haven: Yale University Press, 1984); G. Bingham Powell, Contemporary Democracies (Cambridge: Harvard University Press, 1992).

(4.) Dieter Fuchs and Hans-Dieter Klingemann, "The Left-Right Schema." In M. Kent Jennings and Jan van Deth, eds., Continuities in Political Action (Berlin: de Gruyter, 1989); Russell Dalton, "The Quantity and the Quality of Party Systems," Comparative Political Studies (forthcoming, 2008); Ronald Inglehart, Culture Shift in Advanced Industrial Society (Princeton: Princeton University Press, 1990); Oddbjorn Knutsen, "Left-Right Party Polarization Among the Mass Publics." In H. Narud and T. Aalberg, eds., Challenges to Representative Democracy (Bergen: Fagbokforlaget, 1990).

(5.) We are using modules 1 and 2 from the CSES. These data were downloaded from the CSES website (www.cses.org), which includes the questionnaires and other documentation.

(6.) Seymour Martin Lipset, Political Man: The Social Bases of Politics (Baltimore: Johns Hopkins University Press, 1981), p. 230; Lipset and Rokkan, Party Systems and Voter Alignments.

(7.) Herbert Kitschelt et al., Post-communist Party Systems (New York: Cambridge University Press, 1999); Scott Mainwaring, Rethinking Party Systems in the Third Wave of Democratization (Palo Alto: Stanford University Press, 1999); Alejandro Moreno, Political Cleavages: Issues, Parties and the Consolidation of Democracy (Boulder: Westview Press, 1999); Shaheen Mozaffar, J. Scarritt, and G. Galaich, "Electoral Institutions, Ethnopolitical Cleavages, and Party Systems in Africa's Emerging Democracies," American Political Science Review 97 (2003): 379-390.

(8.) Anthony Downs, An Economic Theory of Democracy (New York: Wiley, 1957).

(9.) Gary Cox, "Centripetal and Centrifugal Incentives in Electoral Systems," American Journal of Political Science 34 (1990): 903-935; Knutsen, "Left-Right Party Polarization."

(10.) Other research describes the different correlates of left-right self-placement across nations, and thus the different interpretation of this scale: Ronald Inglehart and Hans-Dieter Klingemann, "Party Identification, Ideological Preference and the Left-Right Dimension Among Western Mass Publics." In Ian Budge, Ivor Crewe, and Dennis Farlie, eds., Party Identification and Beyond (New York: Wiley, 1976); Inglehart, Culture Shift; Russell Dalton, "Social Modernization and the End of Ideology Debate: Patterns of Ideological Polarization," Japanese Journal of Political Science 7, no. 1 (2006): 1-22. We also conducted more detailed analyses of the correlates of left-right for the six nations in this article, and this methodological appendix is available from the authors.

(11.) Inglehart, Culture Shift, p. 273.

(12.) Inglehart and Klingemann, "Party Identification, Ideological Preference"; Inglehart, Culture Shift; Dalton, "Social Modernization"; Oddbjorn Knutsen, "Left-Right Materialist Value Orientations." In Jan van Deth and Elinor Scarbrough, eds., The Impact of Values (Oxford: Oxford University Press, 1995).

(13.) Scott Flanagan, "Electoral Change in Japan." In Russell Dalton, Scott Flanagan, and Paul Beck, eds., Electoral Change in Advanced Industrial Democracies (Princeton: Princeton University Press, 1984); Herbert Weisberg and Aiji Tanaka, "Change in the Spatial Dimensions of Party Conflict: The Case of Japan in the 1990s," Political Behavior 23 (2001): 75-101.

(14.) Scott Flanagan et al., The Japanese Voter (New Haven: Yale University Press, 1991).

(15.) Doh Chull Shin and Byong-Kuen Jhee, "How Does Democratic Regime Change Affect Mass Political Ideology?" International Political Science Review 26, no. 4 (2005): 381-396. Kang surveyed members of the National Assembly in 2002 and 2004, and argued that the Roh administration was increasing the ideological polarization of the Korean political parties. Won Taek Kang, "Ideological Clash of Progressives and Conservatives in Korea," Korea Focus 13, no. 5 (2005).

(16.) John Fuh-Sheng Hsieh, "Continuity and Change in Taiwan's Electoral Politics." In John Fuh-Sheng Hsieh and David Newman, eds., How Asia Votes (New York: Chatham House, 2002); Yun-han Chu, "Taiwan's Year of Stress," Journal of Democracy 16 (April 2005): 43-57.

(17.) Hsieh, "Continuity and Change"; Dafydd Fell, "Measurement of Party Position and Party Competition in Taiwan," Issues & Studies 40, nos. 3/4 (September/December 2004): 101-136.

(18.) Steven Rood, "Elections as Complicated and Important Events in the Philippines." In Fuh-Sheng Hsieh and Newman, How Asia Votes; also see article by McAllister in this issue.

(19.) Lee Sigelman and S. Yough, "Left-Right Polarization in National Party Systems: A Cross-National Analysis," Comparative Political Studies 11, no. 3 (1978): 355-379; Donald Gross and Lee Sigelman, "Comparing Party Systems: A Multidimensional Approach," Comparative Politics 16, no. 4 (1984): 463-479.

(20.) Hans-Dieter Klingemann, "Political Parties and Party Systems." In J. Thomassen, ed., The European Voter: A Comparative Study of Modern Democracies (Oxford: Oxford University Press, 2005); Miki Caul and Mark Gray, "From Platform Declarations to Policy Outcomes." In R. Dalton and M. Wattenberg, eds., Parties Without Partisans (Oxford: Oxford University Press, 2000).

(21.) Michael Laver and J. Garry, "Estimating Policy Positions from Political Texts," American Journal of Political Science 44, no. 3 (2000): 619-634; Matt Gabel and John Huber, "Putting Parties in Their Place: Inferring Party Left-Right Ideological Positions from Party Manifestos Data," American Journal of Political Science 44 (2000): 94-103.

(22.) Additional information on the CSES surveys, fieldwork, and questionnaires is available from the project website (www.cses.org). Both CSES modules include Thailand, but the Thai survey did not include the left-right scale.

(23.) See Ian McAllister, Political Behaviour (Melbourne: Longman Cheshire, 1992).

(24.) The Japanese survey used a progressive/conservative scale as an equivalent to left-right. The left-right question is widely used in electoral research. Inglehart and Klingemann, "Party Identification, Ideological Preference"; Fuchs and Klingemann, "The Left-Right Schema"; Dalton, "The Quantity and the Quality of Party Systems."

(25.) However, in the World Values Survey, 97 percent of the Taiwanese respondents positioned themselves on the left-right scale.

(26.) Dalton, "Social Modernization." In addition, nineteen Western democracies were included in module II of CSES, and on average 89 percent of these publics positioned themselves on the left-right scale. Among the four East European democracies in this module, 88 percent positioned themselves on the scale.

(27.) See the regional comparisons in Dalton, "Social Modernization."

(28.) The percentage of the public able to locate two major parties is as follows: Australia, Liberals (83 percent) and Labour (80 percent); New Zealand, Labour (82 percent) and National (80 percent); Japan, LDP (85 percent) and JCP (80 percent); Korea, GNP (93 percent) and UD (90 percent); Philippines, Lakas (83 percent) and NPC (81 percent); Taiwan, DPP (49 percent) and KMT (49 percent).

(29.) In part, this reflects the unusually personalist nature of Korean party politics. The progressive, democratic reformer Kim Dae-jung ran for the presidency in 1998 as the head of a centrist/conservative coalition.

(30.) Rood, "Elections as Complicated and Important Events."

(31.) For instance, in the 2001 World Values Survey, the Philippine leftright mean is 6.44, which is between the CSES mean of 7.16 in 1998 and 6.01 in 2004.

(32.) Jeffrey A. Karp and Susan A. Banducci, "Issues and Party Competition Under Alternative Electoral Systems," Party Politics 8, no. 1: 123-141.

(33.) For example, Sartori, Parties and Party Systems; Powell, Contemporary Democracies.

(34.) Dalton, "The Quantity and the Quality of Party Systems."

(35.) We thank Rein Taagepera for his assistance in developing these staffstics. The polarization index is measured as P = [SIGMA] (party vote [share.sub.i]) * [(party L-R [score.sub.i]--party system average L-R score/5).sup.2] (where i represents individual parties). This P index has a value of 0 when all parties occupy the same position on the left-right scale, and 10 when all the parties are located at either 0 or 10 on the scale. The representation index is measured as R = [SIGMA] (governing party L-R score--voter L-R [score.sub.i])/n (where i represents individual voters).

The representation gap is measured as the absolute difference between the average voter and the perceived position of the governing party. The governing party is defined as the party of the prime minister.

(36.) These cross-national comparisons are based on a different set of nations in both timepoints, and thus changes over time may reflect a different mix of nations. There were fourteen Western democracies (Western Europe and North America) in module I and seventeen in module II. There were seven Eastern European nations in module I and four in module II. For additional analyses, see Dalton, "The Quantity and the Quality of Party Systems."

(37.) Ibid.

(38.) Ibid.

Russell J. Dalton is professor of political science and former director of the Center for the Study of Democracy at the University of California, Irvine. He has been a Fulbright Professor at the University of Mannheim, a German Marshall Research Fellow, and a POSCO Fellow. His recent publications include The Good Citizen (2007), Democratic Challenges, Democratic Choices (2004), Citizen Politics (2006); and he is coeditor of Oxford Handbook of Political Behavior (2007), Citizens, Democracy and Markets Around the Pacific Rim (2006), Democracy Transformed? (2003), and Parties Without Partisans (2001).

Aiji Tanaka is professor of political science at Waseda University and dean of Academic Studies. He has been a co-principal investigator of the Japanese election studies, and his research focuses on electoral behavior in Japan and comparatively. His recent publications include "Change in the Spatial Dimensions of Party Conflict: The Case of Japan in the 1990s," Political Behavior (2001) (with Herbert F. Weisberg); and he coauthored Seijigaku (Political Science: Theory and Scope) (2003).

Table 1 Party System Characteristics

 Party System Voter-Government
 Polarization Representation Gap

Nation CSES 1 CSES 2 CSES 1 CSES 2

Japan 3.30 3.30 2.68 2.44
South Korea 0.70 3.55 2.19 2.06
Taiwan 1.18 1.14 1.51 2.26
Philippines 1.33 0.46 1.94 2.34
Australia 2.04 1.96 1.92 2.42
New Zealand 3.81 3.35 2.91 2.45
Average of Western nations 3.30 3.18 -- --
Average of East European nations 3.67 5.14 -- --

Source: Calculated by authors from CSES, modules 1 and 2.

Note: See note 35 for the calculation of the polarization and
representation indices.