|
|
|
|
Electoral Law and Party System Formation in New Democracies: The Case of Poland
Institute on East Central Europe
March, 1997
Introduction
After the revolutions of 1989 the countries of Central and Eastern Europe confronted choices regarding their future political institutions. One of these institutions was the electoral system: the rules governing which parties and organizations may participate in elections, who can vote for them, and how the votes of the citizens are translated into seats in the parliament.
Political scientists have long examined the link between the electoral and party systems, showing that restrictive electoral laws can limit the degree of multipartyism. In the cases of the new Central and Eastern European democracies, with their choice of electoral regimes based on the proportional representation principle and with the explosion of new parties after the 1989 revolutions, the issue of the link between electoral laws and party systems has very practical implications. This is so because the kinds of party systems emerging in these countries will exert a strong influence on the success of democratic consolidation, and are an important factor shaping the content and consistency of social and economic policies. To a large extent, government stability in these countries does and will depend on the degree of fragmentation of the party system.
Studies previously conducted on Western democracies confirm the validity of these claims. In his sample of 20 established democracies, Arend Lijphart observed that there was a very strong negative correlation between the effective number of parties and cabinet durability. 2 The reason for this correlation is very simple: the more parties are needed to form a government, the more likely it is that they will disagree on a policy issue requiring a compromise. Moreover, in Central and Eastern Europe, where the continuity of economic reforms is of paramount importance, frequent government changes may subvert the overall direction of economic transformations, and will almost certainly undermine the govern-ment's credibility in the eyes of foreign investors and the emerging domestic business elite.
More importantly from the point of view of consolidation of democracy in the region, Harmel and Robertson have found a positive correlation between cabinet stability and legitimation, as measured by survey data. 3 Their study implies that strong fragmentation of the party systems in Central and Eastern Europe leading to cabinet instability may indeed undermine democratic consolidation in these countries. Similarly, in a time-series study of six Western democracies, Weil found that the structure of the opposition, measured by the polarization and fractionalization of the party system, has a strong impact on political confidence and on popular support for democracy. 4 His conclusion that higher polarization and fractionalization of the party system depress support for democracy has crucial implications for the study of electoral and party systems in the Central and Eastern European countries undergoing transitions to democracy.
It is therefore appropriate to begin by asking to what extent the design of electoral institutions can help or hinder the process of transformations. Taagepera and Shugart have asserted that electoral systems "can make or break a party -- or even a country... They are also easier to change than most other features of a political system and therefore offer a promising field for `political engineering.'" 5 Has the experience of Central and Eastern European countries in the formative years of their democracies confirmed the view that political engineering can help structure the party system in a way that facilitates a successful democratic consolidation?
Since its first semi-free elections in 1989, Poland has undergone two revisions of its electoral law. In the 1991 elections the Poles voted under an almost pure proportional representation system, which resulted in a parliament representing 29 parties and organizations. Between 1991 and 1993, such a parliament produced three different coalition governments, and was dissolved by the president after a vote of no-confidence.
In 1993 the same parliament changed the electoral law, establishing a five percent national vote threshold for parties to gain representation in the parliament, revising the number of electoral districts, lowering the average number of seats per district, and using a more restrictive electoral formula that privileged the larger parties. The 1993 elections subsequently resulted in a much less fragmented parliament, with only seven parties represented. The same coalition (though with three different prime ministers) has been ruling Poland since then and most likely will continue to do so until the next elections in September 1997.
The Polish case seems to be an example of successful electoral engineering, in which extreme multipartyism under a pure proportional representation regime, which was causing government instability and obstructing consistent implementation of economic reforms, was then successfully limited by the change of the electoral regime. However, the precise effects of this change in the electoral regime need to be examined more closely. This essay will analyze the electoral laws and the 1991 and 1993 election results in Poland in view of the theory of electoral laws and their political consequences.
Electoral Laws and Their Political Consequences
A substantial literature on the link between electoral regimes and party systems exists in political science. This literature originated from the pioneering work of Maurice Duverger, who formulated his famous law regarding plurality electoral systems: "the simple-majority single-ballot system favors the two party system." 6 The counterpart for the proportional representation system, later dubbed Duverger's hypothesis, was stated in similar terms: "the simple majority system with second ballot and proportional representation favor multipartyism." 7 William Riker later commented that Duverger's reasoning represented an early use of rational choice analysis to understand the consequences of particular political institutions. 8 Duverger analyzed electoral laws with regard to two different types of effects: the mechanical (direct) effect of translating votes into seats, and the psychological (indirect) effect of altering the perceptions and choices of party leaders and voters. An electoral system based on plurality voting has the mechanical effect of underrepresenting third parties, and the psychological effect of inducing strategic voting, because rational voters are not willing to "waste" their votes by supporting a party which is unlikely to gain power. These two effects jointly provide for the maintenance of a two-party system: the system provides a disincentive for voters and leaders to support third parties, which get eliminated as a result, and also deters the two main parties from splitting into smaller ones.
Duverger's analysis of the effects of proportional representation (hereafter PR), however, is somewhat less assertive. He noted that there is an association between PR and multipartyism, but did not claim that PR was the main culprit in causing multipartyism. Duverger suggested that multipartyism arises "from the mutual independence of sets of antitheses," 9 meaning a number of different cross-cutting cleavages within the society. As far as political effects of PR go, Duverger notices that under PR there is no incentive for parties with similar ideological stances to merge, and no disincentive for party splits. However, he also says on the basis of empirical data that the multiplicative effect of PR is "not as strong as sometimes suggested," 10 and that it is limited mostly to the creation of new, but small, parties. On the whole, Duverger claims, proportional representation tends to maintain the party structure in existence at the time of PR's appearance. Moreover, he notices that many countries using PR also use some method of limiting the multiplicative effect: either a threshold of votes required for reaching the parliament, or a formula privileging the larger parties.
Douglas Rae, using a large sample of European democracies, found a strong link between electoral laws and the political systems in his sample countries. Rae examined the influence of the electoral formula (method of translation of votes into seats in a district), the district magnitude (number of seats in a district), and the ballot structure (vote-splitting) on the proportionality of the electoral outcome ("proximal" effect) and on the degree of multipartyism ("distal" effect). 11 Proportionality is the correspondence between the shares of votes and shares of seats received by the parties participating in the election. Rae concluded that there is a strong relationship between two independent variables (electoral formula and district magnitude) and the disproportionality of the electoral outcome, and another strong relationship between these two independent variables and multipartyism.
William Riker, revisiting Duverger's law, makes explicit the rational choice assumptions behind his original formulations: politicians and voters are rational actors, responding to the incentives and disincentives provided by the electoral system. 12 Riker actually goes a step beyond Duverger, pointing out that PR systems will offer politicians an incentive to form new parties, and no disincentive for splintering. But Riker is far more cautious about assessing the effects of PR on voter behavior, noting that there is relatively little hard evidence that strategic voting exists at a mass level. Moreover, Riker observes that it is not even clear from a theoretical point of view what the rational strategy for a voter in a PR system would be: should such a voter treat the vote as "part of the selection process," or as an "expression of preferences?" Anthony Downs claims that voting is part of the selection process; 13 but Riker responds that because an individual voter under PR is even less able to influence the selection process than under a plurality system, it may be rational to vote as an honest expression of preferences.
By contrast, Sartori is far more concerned with how the voter responds to PR. He considers that the electoral formula is relatively unimportant, and that the magnitude of the district is likely to have a much larger influence over the proportionality (or disproportionality) of the electoral outcome. 14 Sartori also makes the important comment that the most direct effect of the electoral system is generally on the voter at the constituency level, and that the electoral system affects the party system at the aggregate level in a far more indirect fashion. Sartori distinguishes between "strong" and "feeble" electoral systems, depending on whether or not they have a "constraining" or "manipulative" effect on the voter. In Sartori's terms, the plurality system is relatively strong, absolutely pure PR is feeble, and any PR systems with national vote thresholds are somewhere in between the two extremes. Sartori claims that pure PR does not have multiplicative tendencies, and indeed need not have any effect on the party system at all; he explains the general tendency to associate PR with party multiplication as being an "optical illusion," which emerged because PR was introduced at the same time that new groups of voters were being enfranchised. Moreover, when PR does have an effect on the party system, it tends to be restrictive "to the extent that [PR] is nonproportional." 15 Thus, Sartori claims, "the influence of PR merely represents an enfeeblement of the same influence that is exerted by the majority-plurality systems." 16
With respect to the specific cases considered in this essay, the specific predictions that result from Sartori's explanation are relevant to the case of new democracies. Sartori claims that PR systems can "obtain reductive effects... in proportion to their nonproportionality; and particularly whenever they apply to small sized constituencies, establish a threshold of representation, or attribute a premium. Under these conditions PR will eliminate the lesser parties whose electorate is dispersed throughout the constituencies." 17 He concludes with a remark which is especially relevant for the new Central and Eastern European democracies: "the newborn countries that start with PR have set for themselves the least favorable conditions for overcoming party atomization and for attaining structural consolidation." 18
The Institutionalist Approach and the Study of New Democracies: A Hypothesis about the Political Consequences of Electoral Laws
The implications of this theory review for the study of the new Central and Eastern European democracies (and the Polish case in particular) are clear: electoral laws should have a strong influence in shaping the emerging party system. In particular, the district magnitude and the electoral formula should influence both the level of disproportionality of the electoral outcome and the degree of multipartyism. The larger the district magnitude and the weaker/less restrictive (i.e., less distorting) the electoral formula, the more proportional the electoral outcome and the larger the number of political parties will be. Thus, lowering the district magnitude and/or strengthening the formula (or introducing additional features favoring large parties, such as national vote thresholds or an additional pool of seats to be divided among the largest parties) should result in higher disproportionality of the electoral outcome and a fall in the number of parties.
In the case of Poland, we expect that the more restrictive 1993 electoral law will result in a more disproportional electoral outcome and a fall in the number of parties. The mechanical effects of the change in the electoral regime are easy to demonstrate. Proving the existence of psychological effects is more of a challenge. If Duverger was right about the psychological effect, then we should be able to observe two phenomena. On the party level, we should see consolidation of the existing parties. We can expect an increase in coalition formation and mergers, and a decrease in party splits. Overall, fewer parties should participate in elections, and this is especially true of the very small parties. Moreover, on the voter level we should observe a shift in voter preferences away from the smaller and towards the larger parties.
These are the predictions derived from Duverger's and Rae's analyses. However, I argue that in order to apply this theory to the case of new democracies, one must modify its assumptions, and, what follows, also the specific predictions. All of the literature discussed above is concerned with the political consequences of electoral regimes in already established party systems. The case of the Central and Eastern European countries, however, is that of totally new democracies, where the period between the last democratic experience and the period of recent democratization was at least 40 years. Thus what we are observing is not the political consequences of electoral laws for existing party systems, but rather, consequences of electoral laws for the formation of party systems.
Once we accept the specificity of new democracies, what does change? I argue that one of the most crucial differences between the political consequences of electoral regimes and their resulting changes in established and in new democracies results from the higher level of uncertainty in the latter cases. The prediction that politicians and voters will respond to the incentives and disincentives of the electoral regime is based on a number of assumptions. One is that both the politicians and voters understand how the electoral system works and what the mechanical effects of its provisions are. Another is that both politicians and voters have access to reliable information regarding the distribution of political preferences within the society. A third is that voters have a clear understanding of the differences among the political parties. Finally, from the politicians' point of view, the presence of other elections (presidential or senate), especially conducted under a different set of rules and at different times than the parliamentary ones, weakens any incentives for party mergers and disincentives for splintering. If the party's political survival is not solely dependent on the results of parliamentary elections, the influence that the electoral regime exerts on the party's strategy will be somewhat weakened.
In new democracies, at least initially, some of these assumptions do not hold. One can plausibly argue that politicians, especially in the larger parties, understand the provisions of the electoral law; after all, any given electoral regime has to be accepted by the parliament, after some deliberation and discussion. However, the assumption that the voters in new democracies understand a priori how the electoral law works and the likely effects of its various elements is much more dubious. I would argue that at best the average voter understands the mechanics of the national vote threshold provision. Moreover, in new democracies there is no prior knowledge about likely voting patterns, on which both the politicians and the voters can base their calculations. In addition, we have no reason to expect the voters to behave strategically (and switch from a smaller party to a larger one which is ideologically close) if it cannot be shown that the voters clearly understand the differences among the various parties and their political platforms.
Under these altered behavioral assumptions I will argue that in new democracies, the link between the electoral regime and the party systems is weakened. Specifically, the reaction of both voters and politicians to a change in provisions of the electoral law that is predicted by the rational choice-based approach will occur only with a lag of one or two rounds of elections, to allow for the learning process to take place.
Discussion of the Variables
In examining these hypotheses, my independent variable will be the strength of the electoral system (where a strong system implies one that is restrictive or constraining), which I will analyze in terms of the electoral formula, the district magnitude, the thresholds of representation, and any other special premiums which tend to privilege larger parties. I follow Lijphart's classification of electoral formulas in the hypothesized declining order of restrictiveness (from strong to weak): plurality and majority formulas, d'Hondt, largest remainders-Droop and modified Sainte-Lague, largest remainders-Hare and pure Sainte-Lague. 19 The district magnitude (M), or number of seats per district, affects proportionality such that the larger the district, the more the final outcome reflects the exact distribution of voters' preferences. National vote thresholds privilege the larger parties, as do national pools of seats reserved for the larger parties; both have the effect of reducing the proportionality of the electoral outcome. 20
I will examine two dependent variables. The first will be the degree of disproportionality of the electoral outcome, as defined above. I will calculate an index of deviation from proportionality, as the sum of the differences between seats and votes received by all the parties, divided by two (D = 1/2[Sigma]|v-s|). 21 I will also use another indicator of disproportionality: D4, which is the difference between the four party concentration index in votes and seats, and which shows the extent to which the electoral system privileges the larger parties. 22 In this way, I will use one index which is sensitive to the number of parties in the party system, and another which does not directly depend on the number of contestants in the electoral competition.
The second dependent variable will be the degree of multipartyism, which I will measure using the Laakso-Taagepera index of the effective number of parties, N = 1/[Sigma]p2, where p is the share of the votes or seats received by a party. 23 Using the effective number of parties, rather than simply the number of parties participating in the elections (elective parties) or those gaining representation in parliament (legislative parties), allows a control for differences in party size. I will measure both the effective number of elective parties (using pv, or the share of votes) and the effective number of legislative parties (using ps, or the share of seats of parliamentary parties). The difference between these two (r), measured as the proportional reduction in the effective number of parties, will be another illustration of the disproportionality of the electoral outcome and of the mechanical effects of the electoral system. I will use the effective number of elective parties as an indicator of the distribution of voters' preferences and of the long-term fragmentation of the party system. Using Nv allows us to test for the presence of the psychological effect on the part of the voters; if voters do not alter their behavior in spite of a change in the electoral system, then the indirect, psychological effect of the electoral system cannot be confirmed. 24
The Electoral System in Poland
The Polish electoral system underwent serious changes in the years immediately following the 1989 revolution. The electoral system for the 1991 Polish elections was an almost pure PR system. 25 Of the 460 seats in the lower house of the parliament, the Sejm, 391 were assigned on the basis of district votes, and 69 seats formed an additional pool of seats reserved for the largest parties which managed to get at least five percent of the national vote or got their representatives elected in at least five districts. There were 37 electoral districts, with their number of seats, or magnitude, ranging from 7 to 17 (determined according to the population of each district), and an average district magnitude of 10.6. The seats on the district level were assigned according to the largest remainders-Hare formula, one of the weakest in terms of restrictiveness. On the national level the premium seats were assigned on the basis of the somewhat more restrictive modified Sainte-Lague formula. There was no threshold of representation required for the parties to win seats in the Sejm.
In May 1993 a new electoral law was adopted which was used in the September 1993 elections. 26 The new law kept 391 seats for district votes and 69 seats as premium for the largest parties. It changed the number of districts from 37 to 52 (3-17 seats per district), which brought the average district magnitude down to 7.5. The new law established a five percent national vote threshold for parties to get into the Sejm (a requirement waived for national minorities), and an eight percent national vote threshold for coalitions. The premium was now awarded only to parties that won at least seven percent of the national vote. All seats, at both the district and the national level, were now to be awarded on the basis of d'Hondt formula. Under both systems, each voter had one vote.
The elections to the Senate in 1991 and 1993 were guided by the same law of May 1991. 27 There were 100 seats in the Senate, elected in 49 electoral districts (generally coinciding with administrative provinces), two in each district (except for Warsaw and Katowice, the most populous provinces). The elections to the Senate were on the basis of plurality formula (the two most successful candidates in each district get elected). Each voter had two votes.
The review of the changes in electoral law in Poland shows that the initial electoral system was very proportional: it had a large average district magnitude, used the largest remainders formula (which favors smaller parties), and had no threshold of representation. The only "impure" element was the pool of national seats awarded to the parties which got more than five percent of the national vote, but because of the clause allowing parties which got elected in five different districts to compete for the national pool seats, this provision hurt only the very small parties. In contrast to 1991, in 1993 the PR system was much more restrictive: the average district magnitude went down from 10.6 to 7.5, the electoral formula was changed to privilege larger parties, and a threshold of five percent for individual parties and eight percent for coalitions was adopted, while the threshold necessary to win the seats from the national pool was raised to seven percent.
The Dependent Variables
In this section I present the election results from the 1991 and 1993 elections in Poland, and discuss the measures of disproportionality of the electoral outcome and party system fragmentation. Even a quick glance at the evolution of the party system in Poland shows a persistently high level of fragmentation of the party system. Most major and middle-sized parties have experienced splits at some point in time. There have been few important mergers, with the exception of the UW (Union of Freedom, formed after the merger of UD, the Democratic Union and KLD, the Liberal-Democratic Congress) and the UP (Union of Labor, founded by the Solidarity of Labor, or SP, and a few small left-wing groups). There have been some important coalitions: 1) SLD (the Democratic Left Alliance, predominantly the former communists and the post-communist trade union OPZZ), which, however, was formed before the more restrictive 1993 law, so it cannot be seen as a response to the electoral law provisions; 2) Ojczyzna (Catholic-right coalition), which was formed before the 1993 elections in response to the 1993 law, but failed to gain repre-sentation in the parliament; 3) ROP (Movement for the Reconstruction of Poland); and 4) AWS ("Solidarity's" Electoral Action). The latter two are important coalitions on the right that formed in the anticipation of the 1997 election. But on the whole, especially before the 1993 elections, splinters and new parties were more common than coalitions and mergers.
Election Results
The 1991 elections in Poland produced an extremely fragmented Sejm and Senate. 28 Overall, 111 parties and groups participated in the elections, and 29 got representation in the Sejm, whereas 13 got representation in the Senate. There were also 18 independents in the Senate.
|
Parties in the 1991 Polish Sejm
elections
|
Percent of votes
|
Percent of seats
|
|
UD - Democratic Union |
12.3 |
13.5 |
|
SLD - Democratic Left Alliance |
12.0 |
13.0 |
|
ZChN - Christian National Union |
8.7 |
10.7 |
|
PSL - Peasant Party |
8.7 |
10.4 |
|
KPN - Confederation for Independent Poland |
7.5 |
10.0 |
|
PC - Center Alliance |
8.7 |
9.6 |
|
KLD - Liberal Democratic Congress |
7.5 |
8.0 |
|
PL - Agrarian Alliance |
5.5 |
6.1 |
|
Solidarity |
5.1 |
5.9 |
|
PPPP - Beer Lovers |
3.3 |
3.5 |
|
German minority |
1.2 |
1.5 |
|
ChD - Christian Democrats |
2.4 |
1.1 |
|
SP - Solidarity of Labor |
2.1 |
0.9 |
|
PChD - Party of Christian Democrats |
1.1 |
0.9 |
|
UPR - Union of Real Politics |
2.3 |
0.6 |
|
other (14 parties/committees) |
10.1 |
4.7 |
Source: Monitor Polski, 41/1991.
Only 1.5 percent of the vote was "wasted" in the sense that it did not win any representation (the sum of votes for the parties represented in the Sejm was 98.5 percent). The data above shows the extreme fragmentation of the party system, where the largest party barely won more than 12 percent of the vote and 13 percent of the seats. The Senate elections similarly produced a fragmented picture: 13 parties won representation. The effective number of legislative parties in the Senate was 12.9.
The 1993 elections produced very different results. "Only" 26 parties and coalitions participated in the elections. With the new electoral law, only seven parties made it to the parliament (this includes the German minority, which was exempt from the five percent national vote threshold).
|
Parties in the 1993 Polish Sejm
elections
|
Percent of votes
|
Percent of seats
|
|
SLD |
20.4 |
37.2 |
|
PSL |
15.4 |
28.7 |
|
UD |
10.6 |
16.1 |
|
UP (former SP) |
7.3 |
7.6 |
|
KPN |
5.8 |
4.8 |
|
BBWR (Non-party Bloc in Support of the Reforms) |
5.4 |
3.5 |
|
German minority |
0.7 |
0.9 |
|
Ojczyzna (Catholic Coalition "Fatherland") |
6.4 |
0 |
|
Solidarity |
4.9 |
0 |
|
PC |
4.4 |
0 |
|
KLD |
4.0 |
0 |
|
UPR |
3.2 |
0 |
|
Samoobrona |
2.8 |
0 |
|
Party X |
2.7 |
0 |
|
KdR (Coalition for the Republic) |
2.7 |
0 |
|
PL |
2.4 |
0 |
|
other |
1.1 |
0 |
Source: Monitor Polski, 50/1993.
Thus in the 1993 elections 34.6 percent of voters "wasted" their votes: 34.6 percent of the popular vote did not win any representation in the parliament. Therefore the parties which did gain representation in the Sejm split the empty seats among themselves, with the largest party gaining only 20 percent of the votes, but 37 percent of the seats. This is a stark contrast with the highly proportional outcome of the 1991 elections. The elections to the Senate also produced very different results: the SLD controlled 37 seats, PSL 36, Solidarity 9, UD 4, UP and BBWR 2, and ZChN, PC, KLD and PL 1, and the independents 10. 29 The number of effective legislative parties in the Senate decreased quite dramatically, to 3.7.
Measures of Disproportionality and Multipartyism
The measures of disproportionality and multipartyism in the 1991 and 1993 elections in Poland are presented in the table below:
|
|
1991
|
1993
|
|
Nv
|
13.9 |
9.8 |
|
Ns
|
12.9 |
3.7 |
|
D
|
10.25 |
37 |
|
r
|
21 |
60 |
|
D4
|
5.9 |
35.9 |
|
80 percent ES
|
11 |
9 |
|
80 percent LS
|
8 |
3 |
Source: author's calculations.
Where:
Theories of the Political Consequences of Electoral Law and the Polish Case
The Polish case clearly illustrates the mechanical effects of changes in the electoral law. The 1993 electoral regime was much stronger than the 1991 regime as a result of the change in the electoral formula, which lowered the average district magnitude and established a national vote threshold privileging larger parties. As a result, the number of legislative parties decreased from 12.9 in 1991 to 3.7 in 1993. More telling is the reduction ratio, which measures the proportional difference between the number of elective and legislative parties: whereas in 1991 this ratio was 21 percent, in 1993 it was 60 percent. The
5.9 The psychological effects of the change in the electoral law are much more difficult to prove. Following Duverger and Riker, we would expect party leaders to form coalitions or merge parties whenever they could expect difficulties with passing the five percent threshold. However, this did not happen. Among the more politically relevant parties only one such coalition (Ojczyzna) was formed -- KdR was also a coalition, but a minute one. The psychological effect did take place, but to a limited extent: the number of parties participating in the elections decreased (from 111 to 26), but this change reflected the elimination of very minute parties and organizations which would not influence the formation of cabinets. Much more important is the index measuring the effective number of parties participating in elections: here we clearly see that the number of parties that "count" on the political scene did not decrease as drastically (from 13.9 to 9.8). Moreover, the psychological effect on the part of the voters was basically non-existent. The 34+ percent of votes which were "wasted" on parties that did not get represen-tation in the parliament are one indication of this. The extremely high disproportionality index in the 1993 elections suggests that, on the whole, voters and party leaders failed to adapt to the new electoral rules. The new system did not lead to strategic voting on the part of the voters, who in spite of its high restrictiveness still spread their vote among 10 effective parties. A counterfactual simulation of the 1993 elections using the 1991 electoral law, outlined below, shows that there was little change in terms of voters' behavior. The table below presents calculations of the results of the Polish 1993 elections, using the 1991 electoral law (and in the last column, using the 1991 electoral law, but 1993 districts). These results demonstrate the strong mechanical effect of the 1993 electoral law. The difference between the real and the counterfactual results (assuming that voters vote according to their preferences and not strategically) is quite dramatic. What is also interesting is the comparison between the counterfactual results from 1993 and the actual results from 1991. There has clearly been a shift in voters' preferences in favor of left-wing parties (seen in the Senate elections results); but on the whole, in 1993 the voters (and party leaders, who in general refrained from forming coalitions) behaved as if the rules of the game had not been changed for the Sejm elections. This contention is supported by the similar levels of disproportionality between the 1991 elections and the counterfactual 1993 elections. In the counterfactual elections, the disproportionality index D was 8.5, and in the 1991 elections it was 10.25. Similarly, the number of effective legislative parties in the 1993 counterfactual elections (Ns = 8.8) was closer to the 1991 elections (Ns = 10.9), than to the actual 1993 elections (Ns = 3.7).
Parties in the 1993 Polish Sejm
elections
Percent of votes
Percent of seats
Percent of seats acc. to `91 law
(`91 districts)
Percent of seats acc. to `91 law (`93 districts)
SLD 20.4 37.2 21.1 21.5 PSL 15.4 28.7 16.1 16.5 UD 10.6 16.1 11.5 12.8 UP (former SP) 7.3 7.6 8.5 9.6 KPN 5.8 4.8 7.2 7.0 BBWR (Non-Party Bloc in Support of the Reforms) 5.4 3.5 5.6 4.6 German minority 0.7 0.9 0.9 0.9 Ojczyzna (Catholic Coalition "Fatherland") 6.4 0 8.3 8.7 Solidarity 4.9 0 5.9 5.2 PC 4.4 0 4.6 3.3 KLD 4.0 0 3.0 3.0 UPR 3.2 0 1.9 2.2 Samobrona 2.8 0 1.1 0.9 Party X 2.7 0 1.3 0.6 KdR (Coalition for the Republic) 2.7 0 1.3 1.5 PL 2.4 0 1.7 1.7 other
Source: Monitor Polski, 50/1993, and Dziennik Ustaw, 59/1991; author's calculations
The Role of Uncertainty Revisited
At the beginning of this essay, I argued that the standard assumptions of the theory of the political consequences of electoral laws are not directly applicable to the cases of new democracies. I proposed instead that uncertainty on the part of both the voters and the politicians will cause a lag in the response to changes in electoral law provisions.
The review of the Polish case demonstrates the strong mechanical effects of a change in the electoral law. Based on the 1991 and 1993 election results, and the counter-factual 1993 elections according to the 1991 law, I also argue that the psychological effects proposed by Duverger, Rae and Riker were negligible in the 1993 Polish elections. We see more evidence of the psychological effect among the politicians before the 1997 elections, with the formation of the two important party coalitions on the right (ROP and AWS). I contend that this supports my hypothesis that higher levels of uncertainty in new democ-racies weaken the effects of electoral regimes on voters and especially on politicians. Riker's alternative assumption about voting according to preferences holds in the case of the first elections conducted under a new set of rules. Thus voters, not knowing the exact effects of the electoral law provisions, not knowing what to expect from individual parties, and not knowing the exact levels of support for the various parties, will vote according to their ideological preferences. Party leaders, on the other hand, will be willing to risk running alone, because by doing so they ensure party name recognition. Moreover, in the Polish case, the presence of Senate elections conducted under a different electoral regime (a variation of the plurality system) counters the incentives inherent in the more restrictive 1993 electoral law. If a party can get into the Senate, it may be more willing to risk not getting elected to the parliament for the price of not joining an electoral coalition.
The high level of multipartyism in the 1993 Polish elections can thus be explained by the "adaptation with a lag" approach. In 1993, voters and parties were uncertain of the long-term distribution of preferences of the electorate (and voters were most likely unable to assess the restrictiveness of the new electoral system). Thus only one medium-sized coalition formed in anticipation of the obstacle of the threshold requirement. The KLD's behavior before and after the 1993 elections is suggestive here. Before the elections, KLD decided to run alone, even though it could attempt to form a coalition with the UD or run together with its parliamentary club, the Polish Economic Program. Once the KLD failed to get into the parliament, its leadership decided on a full merger with the UD.
The experience of other Central and Eastern European countries gives credence to the "adaptation with a lag" explanation. Slovakia changed its electoral law before the 1992 elections, raising the national vote threshold from three to five percent. The Slovak 1992 elections resulted in a much higher disproportionality level than elections conducted in 1994, by which time the parties and voters had the time to adapt to the new rules (the percentage of "wasted" votes fell from 24 percent to 12). The 1994 elections in Slovakia saw more coalition formation than the 1992 ones. Similarly in Bulgaria the first elections conducted under a new set of rules (d'Hondt formula and a four percent national vote threshold) in 1991 resulted in high levels of disproportionality (D = 24.9), whereas the 1994 elections demonstrated voter adaptation to the new rules (with D falling to 15.7).
The true test will come in the form of the 1997 elections in Poland. If the explanation presented in this paper is correct, in the 1997 elections we will see a decrease in the level of disproportionality and a fall in the number of effective elective parties. These changes will indicate that the politicians and the voters have adapted to the new rules.
Some Thoughts on Electoral Reform in New Democracies
Earlier, I discussed the link between the electoral regime, party system formation, and democratic consolidation. I cited studies indicating that high party system fragmentation leads to cabinet instability and can thus undermine democratic consolidation. I also cited studies linking legitimation and popular support for democracy with low party system fragmentation and polarization.
Such studies suggest that attempts at political engineering through electoral reform which lead to limitations of multipartyism should have a positive influence on democratic consolidation. The findings of this paper, however, suggest that the effects of electoral reforms in new democracies are more complex. The atmosphere of uncertainty regarding the distribution of political preferences of the voters, as well as low party recognition among the electorate, cause a lag in adaptation to the changed rules both among the voters and among the party leaders. This lag leads to extremely high levels of disproportionality in the election results. Whereas in the established democracies disproportionality hovers between one and nine percent, 30 in the Polish 1993 elections it was close to 37 percent (and around 24 percent in 1991 in Bulgaria and 1992 in Slovakia).
This "adaptation with a lag" in new democracies could thus in itself be a threat to democratic consolidation. The 34 percent of Polish voters who "wasted" their votes are without representation in the parliament. More importantly, these voters include most of the Polish right. The lag in adaptation also allows for fortification of temporary advantages. In a system which officially follows the principle of proportional representation, this lag offers a temporary advantage to the largest parties, which is similar to the systems based on plurality vote. During a period of democratic consolidation as well as difficult and important economic reforms, the lag in adaptation to the changes in electoral regime gives a small number of parties a very important role in shaping the future political institutions as well as a new democracy's social and economic policy.
Footnotes
Note 1: I would like to thank Chris Gowlland, Vello Pettai, Dr. Steven Solnick, and Sherrill Stroschein for their comments and suggestions. Back.
Note 2: Arend Lijphart, Democracies: Patterns of Majoritarian and Consensus Government in Twenty-One Countries (New Haven, CT: Yale University Press, 1984), 124-26. A similar point is made by Douglas Rae, The Political Consequences of Electoral Laws (New Haven, CT: Yale University Press, 1971), 98-99. Back.
Note 3: Robert Harmel and John D. Robertson, "Government Stability and Regime Support: A Cross-National Analysis," Journal of Politics 48 (1986), 1029-40. Back.
Note 4: Frederick D. Weil, "The Sources and Structure of Legitimation in Western Democracies: A Consolidated Model Tested with Time-Series Data in Six Countries since World War II," American Sociological Review 54 (1989), 682-706. Back.
Note 5: Rein Taagepera and Matthew Soberg Shugart, Seats and Votes: The Effects and Determinants of Electoral Systems (New Haven, CT: Yale University Press, 1989), 2. Back.
Note 6: Maurice Duverger, Political Parties: Their Organization and Activity in the Modern State (New York: John Wiley and Sons, 1954), 217. Back.
Note 8: William H. Riker, "Duverger's Law Revisited," in Bernard Grofman and Lijphart, eds., Electoral Laws and Their Political Consequences (New York: Agathon Press, 1986), 29. Back.
Note 11: It is important to note that these twin effects are not the same as Duverger's "mechanical" and "psychological" effects explained above. Duverger suggested that the two effects combine and jointly produce a certain degree of disproportionality and a given party system. Back.
Note 13: Anthony Downs, An Economic Theory of Democracy (New York: Harper and Row, 1957). Back.
Note 14: Giovanni Sartori, "The Influence of Electoral Systems: Faulty Laws or Faulty Method?," in Grofman and Lijphart, eds., 53. See also Sartori, Comparative Constitutional Engineering (London: Macmillan, 1994). Back.
Note 15: Sartori, "The Influence of Electoral Systems," 58. Back.
Note 16: Sartori then links the effects of the electoral laws to those of the party system, extending the work of Duverger and Rae. After differentiating strong and feeble electoral systems, he disting-uishes strong (structured) and feeble (unstructured) party systems. In strong party systems, the basis of party identification is ideology; in feeble party systems, the basis of party identification is personalistic allegiance. The effect of the electoral system varies depending on the party system: in feeble party systems, feeble electoral systems have no effect, whereas strong electoral systems have a restraining effect on the number of parties at the constituency level. In strong party systems, feeble electoral laws are counteracted by the party system, whereas strong electoral laws have a re- ductive effect on the number of parties (corresponding to Duverger's law). To the extent that Sar-tori links the party and electoral systems, his analysis differs from the mainstream institutionalist approach. However, in stressing the reductive role of less proportional PR, Sartori places himself in the institutionalist camp. Given that all party systems in Eastern Europe at the onset of the democratization process would fall in the unstructured category, the addition of the party system as an independent variable would only complicate matters without providing much new insight. Back.
Note 19: Under the plurality formula, the person with the most votes in a district gets elected. Under the d'Hondt formula, the numbers of votes received by individual parties in a district are divided by a series of divisors (1, 2, 3, 4 etc.), and seats are allocated to the parties with the highest averages. Under the Sainte-Lague formula, the divisor series used is: 1, 3, 5 etc.; and under the modified Sainte-Lague it is: 1.4, 3, 5 etc. The d'Hondt formula privileges larger parties, the Sainte-Lague method favors small parties, and the modified Sainte-Lague favors medium-sized parties. Under the Hare quota system, the number of votes in a district is divided by the number of seats (M), and parties which get the quota of votes gain seats while the remaining seats are distributed to parties with largest remainders. The Droop quota works in a similar way, except that the number of votes is divided by (M+1), which hurts the smaller parties and results in a somewhat less proportional outcome. For a discussion of the various electoral formulas, see Taagepera and Shugart, 29-35, and Lijphart, "The Political Consequences of Electoral Laws, 1945-85," American Political Science Review 84, no. 2 (1990), 484. Back.
Note 20: Here I am outlining the general characteristics of the electoral system as the independent variable, according to the hypothesized effects of the various provisions of the electoral law on the proportionality of the outcome. Back.
Note 21: This is the Loosemore and Hanby index, which measures the total deviation from proportionality in a given election and is outlined in John Loosemore and Victor J. Hanby, "The Theoretical Limits of Maximum Distortion: Some Analytical Expressions for Electoral Systems," British Journal of Political Science, no. 1 (1971), 467-77. An alternative index was proposed by Rae (84) and was equal to D = 1/N[Sigma]|v-s|, where N = the number of parties participating in the election. As noted by Lijphart, Rae's index is too sensitive to small parties participating in the elections, and as a result overstates the proportionality of the electoral system. It is also more difficult to calculate, because one needs to include differences between shares of votes and seats of even the smallest parties (Lijphart, Democracies, 162-63). The Loosemore and Hanby index, on the other hand, is too sensitive to the number of parties, and as a result, it understates the proportionality of the electoral system (especially in highly fragmented party systems). Lijphart proposes instead an index which averages only the vote-seat share differentials of the two largest parties in an election. Back.
Note 22: Given the generally high degree of party system fragmentation in the Eastern European cases, this is a better measure than Lijphart's index, which is based on only the two largest parties. Back.
Note 23: Laakso and Taagepera, 3-27. Back.
Note 24: From the point of view of the normative implications of this paper, Nv is important in that it indicates the degree of fragmentation of the political scene from the point of view of the voter. Ns corresponds more closely to the expected cabinet durability. The ratio of reduction in the number of parties (Ns - Nv/Nv) indicates the extent of potential voter alienation: if the difference is large, a large segment of the electorate is not represented in the parliament. Back.
Note 25: For details, see Dziennik Ustaw Rzeczypospolitej Polskiej, 59/1991. For a general overview of Eastern European electoral laws see Christian Lucky, "Comparative Chart of Electoral Regimes," East European Constitutional Review 3, no. 2 (1994), 65-77. Back.
Note 26: Dziennik Ustaw Rzeczypospolitej Polskiej, 45/1993. Back.
Note 27: See Dziennik Ustaw Rzeczypospolitej Polskiej, 58/1991. Back.
Note 28: For the official results, see Monitor Polski: Dziennik Urzedowy Rzeczypospolitej Polskiej, 41/1991. For the 1993 elections, see Monitor Polski, 50/1993. An English-language summary of the results is published in the "Bulletin of Electoral Statistics and Public Opinion Research Data," Eastern European Politics and Societies 8, no.2 (1994), 369-379. Back.
Note 29: Monitor Polski, 50/1993. Back.
Note 30: Shugart, "Electoral Reform in Systems of Proportional Representation," European Journal of Political Research 21 (1992), 207-24. Back.
Email this citation