5Coalition Formation in Eastern Europe

LEGISLATIVE STUDIES QUARTERLY, XXXII, 1, February 2007 5

JAMES N. DRUCKMAN ANDREW ROBERTS

Northwestern University

Communist Successor Parties And Coalition Formation In Eastern Europe

One of the most distinctive features of new democracies is the presence of political parties associated with the old, repressive regime. This article investigates whether or not the Eastern European variant of these parties, which we call communist successor parties (CSPs), has affected coalition politics. It finds that CSPs do have significant effects on the dynamics of coalition formation. CSPs are less likely than other parties to be included in governing coalitions; coalitions that include CSPs are more likely to be oversized (that is, to include superfluous parties); and CSPs that make it into government are penalized, insofar as they receive less than their fair share of governing portfolios. We attribute these results to the salience of the regime divide—the affective dislike of many citizens for the legacies of communism. Our results extend research on coalition behavior to Eastern European contexts and show how affective dislike combined with vote-seeking motivations can affect governing behavior.

How do coalitions form and function in new democracies? Most existing theories suggest that coalition formation in these countries should be driven by the same factors as in established democracies: the size and ideology of parties and the institutional framework in which they operate (see Martin and Stevenson 2001 for a summary of theories).

In this article, we look at one significant fact about new democracies that may produce a different sort of coalition politics. Most new democracies are distinguished by the presence of successors to the former ruling parties. These successors may simply be accepted as normal players in the democratic game, but there are good reasons to believe they will be treated differently by other parties. After all, the successor parties are associated with regimes that, at best, restricted fundamental human rights and, at worst, murdered or imprisoned significant portions of their countries’ populations. It is thus reasonable

6 James N. Druckman and Andrew Roberts

to expect that these parties will be seen as more than simply bearers of legislative strength and ideology.

To address this issue, we examine coalition politics in 15 Eastern European countries. Specifically, we studied the effect on coalition formation of what we call communist successor parties (CSPs), the parties that are the organizational successors to the former ruling communist parties.

We focused on three aspects of coalition politics: Do CSPs enter government to the same extent as other parties? Do the governments that CSPs enter differ from governments composed of other parties? And are CSPs allocated a proportional share of ministries when they enter government? In all three cases, we found evidence that CSPs are treated differently than other parties, even when we controlled for their legislative strength and ideology.

We believe that these anomalies stem from a combination of affective dislike on the part of voters and party activists for CSPs and the vote-seeking motivations of party leaders. Party leaders appear to follow the lead of their electorates and party activists in shunning CSPs or limiting the CSPs’ influence on politics.

Our findings are particularly interesting because there has been relatively little study of coalition formation in new democracies. Consistent with recent institutional theories of coalition behavior (for example, Martin and Stevenson 2001 and Strøm, Budge, and Laver 1994), the results imply that understanding coalitional politics requires contextual knowledge about the substance of national politics. Specifically, our findings suggest that scholars should pay more attention to the affective attitudes of party electorates and elites toward rival parties. In cases where there is strong affective dislike of a particular party, vote-seeking motivations may be as, or perhaps even more, important than office- and policy-seeking motivations of parties.

I. Communist Successor Parties

Few topics in postcommunist politics have been studied as intensively as communist successor parties, the parties that are the organizational successors to the old ruling parties. A review of the literature on this topic reveals at least two books, two edited volumes, and 18 articles devoted exclusively to the comparative analysis of these parties.1 Taking into account articles devoted to individual successor parties or to the place of CSPs in party systems substantially adds to this total. No other type of party—liberal, agrarian, nationalist, or even extremist—has received comparable study.

7Coalition Formation in Eastern Europe

Why have these parties attracted so much attention? In the first place, it came as a surprise to many scholars that these parties survived and, in some cases, transformed themselves. It was assumed that parties so closely associated with a repressive dictatorship would quickly fall by the wayside when faced with genuine competition.2

Moreover, it seemed unlikely that these parties would be able to change their ways, given the heavy baggage that they carried from the past. Scholars expected these parties to remain as authoritarian and ideologically extreme as they had always been. For this reason, the ability of CSPs to win elections, first in Lithuania and Poland and later in other countries, led many observers to worry about a possible reversion to nondemocratic politics. When the election of these parties did not portend a return to communism, scholars wanted to know how some of these parties were able to transform themselves into democratic parties.

Finally, students of postcommunist politics were interested in how citizens and politicians would react to these parties. CSPs were intimately connected with some of the most repressive regimes in human history. Would citizens be able to forgive and forget after the fall of communism? Or would they hold grudges against these parties for the destruction their predecessors had inflicted? How important, in short, were legacies of the past?

A survey of the literature devoted to these parties finds that it is focused almost entirely on party development.3 Virtually all of this work addresses the same two questions: (1) What explains the success of CSPs, either in merely surviving or in winning elections? and (2) What explains a CSP’s ability or inability to transform into a normal democratic party? Most of this literature can thus be classed as addressing the causes of these parties rather than their effects.

Instead of revisiting this research, we ask a different, and in some sense, more-consequential question: do CSPs matter politically? Research has not yet shown that these parties have had a definite effect on politics in the region. Did it matter that CSPs managed to survive or transform themselves in the countries of Eastern Europe? Did they actually affect the substance of politics in the region? In essence, ours is the “so what?” question. We focus specifically on the dynamics of government formation and portfolio allocation. These areas are particularly good for tests of these questions because they have bred a large literature (giving us a baseline of comparison) and they are readily quantifiable.

We believe that these questions are significant not only because their answers may provide a validation of previous work, but also because they have important consequences for governance in the region.

8 James N. Druckman and Andrew Roberts

Government formation and the allocation of portfolios have been shown in numerous studies to shape policy outcomes and general well-being (see, for example, Klingemann, Hofferbert, and Budge 1994 and Lijphart 1999). Thus, if CSPs alter the dynamics of coalition politics, then this influence could ultimately have profound effects on the politics of the region.

II. CSPs and Coalition Theory

Just as studies on Eastern European politics have paid scant attention to how CSPs affect coalition governance, the literature on coalition behavior has given limited consideration to governing dynamics in Eastern Europe, instead focusing largely on Western Europe (see, however, Blondel and Müller-Rommel 2001; Druckman and Roberts 2005, N.d., and Grzymala-Busse 2001). That said, coalition theory, more generally, provides reasons to expect CSPs to matter.

Over the last 20 years, researchers have shown how various external constraints affect coalition behavior (see, for example, De Winter 1995; Druckman and Thies 2002; Huber 1996; Laver and Schofield 1990, 195–215; Martin and Stevenson 2001, 35–38; and Mershon 1996). Most of these constraints come from institutions and include such things as formation rules (for instance, investiture rules), cabinet operation rules (such as votes of no confidence), legislative rules (such as dissolution rules), and party rules.

Most relevant for us, given our interest in CSPs, is the impact of party rules. Strøm, Budge, and Laver (1994) explain that

Party leaders may declare that they will, or will not, form particular coalitions. . . . The systematic exclusion of certain parties from coalition bargaining is the most striking party constraint found with any regularity. . . . The examples of the Gaullists or Communists in the French Fourth Republic, Sinn Féin in Ireland, or the Italian Communists in the 1950s show that certain parties, as a consequence of their strong “antisystem” stance, can effectively be discounted as members of any potential government. (317)

CSPs in Eastern Europe differ from these parties insofar as they have generally not been systematically excluded from government formation (Strøm, Budge, and Laver focus on Western Europe; also see Martin and Stevenson 2001, 37, 46; and Mershon 1994, 57; 2002, 40–41). Yet the basic idea that specific types of parties can alter the dynamics of coalition bargaining extends to the case of CSPs. Indeed, Strøm, Budge, and Laver (1994, 317) even acknowledge that “Besides antisystem parties [with whom other parties refuse to bargain], taboos are often

9Coalition Formation in Eastern Europe

attached to parties associated with the previous political regimes, especially undemocratic ones.” These types of parties impose incentives “that change the calculus of the decision maker” (Laver and Schofield 1990, 202).

In the case of CSPs, constraints come from the internal party politics and party electorates of the nonsuccessor parties (see Strøm 1990a). That is, party activists and voters of the other parties may oppose coalitions with CSPs for affective reasons unrelated to party strength and ideology. Party leaders will heed these preferences because, in addition to pursuing office and policy, they pursue votes and need to maintain support from activists and voters.

The underlying reason for these attitudes is what a considerable literature has referred to as the “regime divide”—the affective dislike of former opponents of communism for those associated in any way with the communist system (Kitschelt et al. 1999).4 The brutality of the communist regime is said to have led victims of the regime to visceral distaste for representatives of that regime. In societies flat- tened by communism—and thus lacking many of the political cleavages found in established democracies—the division between supporters and opponents of the old regime appears to be one of the strongest political cleavages. Just as the Civil War generated lasting political cleavages in the United States, the trauma of 40 years of communism should leave a lasting imprint on postcommunist politics.

This emotional distaste may be so great that it overwhelms the desire of some CSPs to draw a thick line between themselves and the previous ruling parties and to present themselves as normal players in the democratic game.5 For many citizens, and even politicians, the mere thought of a CSP provokes a negative emotional reaction. We should add that such a reaction is not usually in response to the current ideology of the party—several of these parties have quite moderate ideologies—or to its legislative strength—none of these parties dominates politics in the way that the former rulers did.

We believe there are two ways in which the regime divide might affect coalition politics.6 First, it is possible that party elites affectively dislike CSPs so intensely that the elites sacrifice some of the benefits of office and policy in order to punish CSPs. This dislike may reflect instrumental consequences generated by intraparty politics. Strøm, Budge, and Laver (1994, 318) explain that “senior politicians interact with activists in their own parties. Party leaders may believe that some coalitions would be sufficiently detrimental to their recruitment of candidates or activists to be avoided under all circumstances. Or they may believe that the benefits of a potential coalition would be

10 James N. Druckman and Andrew Roberts

outweighed by the costs their own activists would impose in irate retaliation. . . . [T]hese risks may induce party leaders to refrain from otherwise attractive coalitions [for example, with CSPs].”

Second, and perhaps more important, the regime divide may work through the general electorate. In addition to seeking office and policy outcomes, parties also seek to maximize the votes they receive in present and future elections (Strøm 1990a). If party electorates are sufficiently opposed to coalitions with CSPs for the affective reasons we mentioned, then party elites will likely follow the electorates’ lead. Elites will avoid such coalitions or find ways to distance themselves from CSPs or keep CSPs in check within such coalitions. All of these strategies are motivated by a desire to avoid losing votes in future elections.7

As evidence of this vote-seeking motivation, Grzymala-Busse (2001) has shown that parties that form coalitions with CSPs are punished more strongly at the polls than parties that do not. Anecdotal evidence supports this view as well. Consider the active movement in the Czech Republic, spanning from left to right, called “One Doesn’t Speak with Communists” or the recent mobilizations of the Hungarian Fidesz party and the Polish Law and Justice Party (PiS) against the vestiges of communism embodied in quite liberal CSPs. The fact that parties in many states of the region campaign on the anticommunism and anti-CSP issue even today, 15 years after the transition, indicates that the issue must still touch a nerve with voters.8

III. Hypotheses

Given this logic, we expect that the presence of CSPs and this regime divide will affect at least three areas of coalition formation. The first is participation in government. While CSPs surely desire to participate in government, it is possible that they are rarely invited to join governments. This exclusion could occur because of some combination of affective dislike, intraparty consequences, and/or fear of electoral consequences. Theories of the regime divide would thus predict that CSPs are less likely to serve in government, all other things being equal. The same conclusion could flow from a consideration of the CSPs themselves. If they are so rigid because of their authoritarian past that they are unable to compromise with other parties, then they will be less likely to participate in coalition governments.

It is important to consider the ceteris paribus clause here. Are CSPs less likely to serve in government, all else being equal? The

11Coalition Formation in Eastern Europe

important control variable is party ideology. CSPs may be discrimi- nated against because they are extremists, that is, because they have not broken ties with the past and renounced the politics of the old regime. A number of theories of government formation argue that ideology matters for government participation (for instance, Laver 1998). We wished to determine if the mere fact of being a CSP leads to decreased participation regardless of ideology. Does the CSP variable add any explanatory power above and beyond ideology? Are these parties excluded from power merely for being who they are? We discuss our test of this hypothesis in Section V.

The second possible effect of the regime divide is on cabinet types. If there is intense affective dislike for CSPs, then they might only enter government as part of surplus majority coalitions. This is for two reasons. First, oversized coalitions help other parties to distance themselves from the former communists in the eyes of voters and thus minimize electoral punishment. Second, the inclusion of extra parties helps to moderate the possibly extreme or reactionary policy proposals of the CSPs and the concomitant uncertainty that a CSP may bring to governing (Dodd 1976; Riker 1962). The formation of an oversized coalition with the former communists was widely noted after the Hungarian elections of 1994.9 We aimed to determine if this tendency toward oversized coalitions has more-general applicability. We discuss our test of this hypothesis in Section VI.

The third effect of CSPs is on the allocation of portfolios once a coalition has formed. In Western Europe, coalition parties tend to receive portfolios in nearly one-to-one proportion to their seat share, with a slight small-party bias.10 Dislike for CSPs, however, may lead the successor parties to be compensated less than proportionally, as other parties try to put distance between themselves and the former communists for the benefit of their electorates or party activists. Even when CSPs form governments, they may have to offer their coalition partners excess portfolios in order to induce the other parties to join the government. On the other hand, the possible authoritarian tendencies of these parties may lead them to demand more than their proportionate share. We discuss our test of these hypotheses in Section VII.

One final note is that there are reasons to expect, contrary to our hypotheses, that CSPs are treated the same as other parties. If other parties care mainly about office and policy, as has been emphasized in many theories, then these parties have strong incentives to negotiate with CSPs as they would with other parties. There are likely to be costs to discriminating against CSPs, and parties concerned only about office or policy should try to avoid these costs.

12 James N. Druckman and Andrew Roberts

Our null hypothesis is thus that CSPs are not treated anomalously in government formation. If we cannot reject the null, then we are left to conclude that the party-specific history of the CSPs in Eastern European politics is not a significant factor in explaining government formation in this region. We would view a rejection of the null as evidence consistent with how constraints stemming from party rules (Strøm, Budge, and Laver 1994) in general and Eastern European CSPs in particular shape coalition politics.

IV. Data

We tested these hypotheses with a new dataset that includes government formation decisions in Eastern Europe. The Eastern European data come from nearly all coalitions in the 15 Eastern European countries listed in Table 1, starting with the first democrati- cally elected government in each country and continuing through 2002.11 This sample thus includes all of the parliamentary democracies in Eastern Europe for the lifetime of their existence. We collected the bulk of these data from Keesing’s Contemporary Archives, although in a few instances we drew on country-specific sources, including data directly from the country’s legislative archives.

For each government, we identified all parties in government, every party with at least one seat in parliament (at the point of govern- ment formation), whether or not the party served in the previous government, the seat share of the party, whether or not the party was a communist successor party, the party type of each party, and the partisan affiliation of each minister.

To determine if a party was a CSP, we followed Bozóki and Ishiyama (2002).12 We drew the party type variable from Armingeon and Careja’s (2004) work. This variable labels each party as one of 17 types specified by Lane, McKay, and Newton (1997): communist, postcommunist, left-socialist, socialist, pensioners, greens, agrarian, liberal, religious, conservative, nationalist, ultra-right, ethnic, regional, alliance, protest, and no-label. We labeled a small number of parties not included in their dataset ourselves. We used the party types as a proxy for ideology in the absence of more-direct measures. We should note that CSP is not a party type but a separate dummy variable; for example, some CSPs are classified as “communists,” while others are “socialists.” (We will discuss the details shortly.)

For most of the article, our unit of analysis is the party, since we are interested in explaining the inclusion of a party in government and the proportion of portfolios each party receives. We also use the

13Coalition Formation in Eastern Europe TA

B L

E 1

D es

cr ip

tio n

of E

as te

rn E

ur op

ea n

D at

a

N um

be r

of N

um be

r of

N um

be r

of Y

ea rs

o f

N um

be r

of N

um be

r of

C ou

nt ry

G ov

er nm

en ts

Pa rt

ie s

in L

eg .a

Pa rt

ie s

in G

ov .b

Fi rs

t a nd

L as

t G ov

. C

SP s

in L

eg .

C SP

s in

G ov

.

A lb

an ia

8 64

33 19

92 –2

00 2

13 %

( 8/

64 )

18 %

( 6/

33 )

B ul

ga ri

a 4

22 12

19 90

–2 00

1 18

% (

4/ 22

) 33

% (

4/ 12

) C

ro at

ia 2

30 11

20 00

–2 00

1 7%

( 2/

30 )

18 %

( 2/

11 )

C ze

ch os

lo va

ki a

2 17

5 19

90 –1

99 2

12 %

( 2/

17 )

0% (

0/ 5)

C ze

ch R

ep ub

lic 3

19 9

19 93

–2 00

2 11

% (

2/ 19

) 0%

( 0/

9) E

st on

ia 6

46 15

19 92

–2 00

2 0%

( 0/

46 )

0% (

0/ 15

) H

un ga

ry 5

39 14

19 90

–2 00

2 13

% (

5/ 39

) 14

% (

2/ 14

) L

at vi

a 9

70 35

19 93

–2 00

0 9%

( 6/

70 )

0% (

0/ 35

) L

it hu

an ia

5 28

12 19

96 –2

00 1

11 %

( 3/

28 )

0% (

0/ 12

) M

ac ed

on ia

4 30

13 19

94 –2

00 0

13 %

( 4/

30 )

15 %

( 2/

13 )

Po la

nd 9

94 32

19 89

–2 00

1 10

% (

9/ 94

) 16

% (

5/ 32

) R

om an

ia 9

63 31

19 91

–2 00

0 14

% (

9/ 63

) 19

% (

6/ 31

) Se

rb ia

1 8

2 19

98 13

% (

1/ 8)

50 %

( 1/

2) Sl

ov ak

ia 4

25 14

19 93

–1 99

8 8%

( 2/

25 )

14 %

( 2/

14 )

Sl ov

en ia

5 38

18 19

93 –2

00 0

13 %

( 5/

38 )

11 %

( 2/

18 )

A ll

C ou

nt ri

es 76

59 3

25 6

19 89

–2 00

2 11

% (

62 /5

93 )

13 %

( 32

/2 56

)

a I nc

lu de

s an

y pa

rt y

w ith

a t l

ea st

o ne

le gi

sl at

ur e

se at

, o ve

r t he

c as

es o

f a ll

go ve

rn m

en ts

fo rm

ed . T

he re

w er

e 76

to ta

l g ov

er nm

en ts

th at

in cl

ud ed

a to

ta l o

f 2 56

pa rt

ie s

ou t o

f a

to ta

l o f

59 3

pa rt

ie s

in th

e le

gi sl

at ur

es .

b S ev

en p

ar tie

s en

te re

d a

go ve

rn m

en t d

es pi

te h

av in

g no

le gi

sl at

iv e

se at

s: f

ou r

pa rt

ie s

in a

s in

gl e

A lb

an ia

n go

ve rn

m en

t, on

e pa

rt y

in tw

o B

ul ga

ri an

g ov

er n-

m en

ts , a

nd o

ne p

ar ty

i n

on e

H un

ga ri

an g

ov er

nm en

t.

14 James N. Druckman and Andrew Roberts

coalition as our unit when we consider the influence of CSPs on cabinet types. In total, there are 76 coalition governments, 593 parties in the legislature, and 256 parties in government. In Table 1, we report, for each Eastern European country, the number of governments, the number of parties in the legislature and government, the years of the first and most recent governments, and the frequency of CSPs in the legislature and in government.

V. Government Participation

Are CSPs discriminated against in government formation? Table 1 gives an indication of the prevalence of CSPs in Eastern Europe. Of the 593 parties that received legislative sets in our dataset, 62 (11%) were CSPs. As far as government participation goes, CSPs constituted 32 of 256 parties (13%) in government. On the surface there does not appear to be discrimination, but these results do not control for party size and ideology.

Table 2 is a cross-tabulation of CSP and party type. As mentioned, CSPs come from various different types (that is, party type reflects ideology, whereas CSP status reflects historical origin). The table shows that CSPs fall into five different party types: communist, postcommunist, left-socialist, socialist, and ethnic. Note that CSPs make up only 31% (62/202) of these five party types (the other 69% of these party types are not CSPs). This distinction will later allow us to control for ideology. There are other parties that share the same ideological space but do not carry the historical baggage of the CSPs.

To determine the causes of government participation, we would, ideally, examine the factors that lead to the creation of different coalitions. This analysis would require specifying all of the possible coalitions that might form and then measuring all factors that might affect formation, including ideology and institutional variables (Martin and Stevenson 2001).

We did not have access to all of these variables, and thus we conducted a more-exploratory analysis to determine if there is any basis for believing that CSPs are treated differently than other parties. Our dependent variable in these regressions is whether or not a party participated in government. Because our dependent variable is dichoto- mous, we used logit regressions in all of the analyses that follow.

We identified a number of independent variables that might affect government participation. First, conventional theories of size predict that larger parties are more likely to participate in government (Martin and Stevenson 2001). To account for this possibility, one of our

15Coalition Formation in Eastern Europe

TABLE 2 CSP Legislative Parties

Party Typea Not CSP CSP Total

Communist 1 (number) 8 9 11% (row %) 89% 100%

1% (column %) 13% 5%

Postcommunist 2 18 20 10% 90% 100%

1% 29% 10%

Left-socialist 14 1 15 93% 7% 100% 10% 2% 7%

Socialist 72 33 105 69% 31% 100% 51% 53% 52%

Ethnic 51 2 53 96% 4% 100% 36% 3% 26%

Total 140 62 202 69% 31% 100%

100% 100% 100%

aTable includes only the parties belonging to the 5 listed party types. Not included are the following other party types, comprising 391 parties: Pensioners, Green, Agrarian, Liberal, Religious, Conservative, Nationalist, Ultraright, Regional, Alliance, Protest, Non-Label/others, for a total of 17 types (and 593 legislative parties). Source: Armingeon and Careja 2004.

independent variables is the seat share of each party in parliament.13 A second variable that might affect participation is incumbency (Martin and Stevenson 2001; Strøm, Budge, and Laver 1994). Parties that have served in the previous government are more likely to serve in the current government; they are part of the status quo.

Finally, we included a measure of party ideology. Most theories of government formation postulate that parties are at least partially motivated by policy. Parties thus judge the suitability of coalition partners by their policy stances. As a result, a party’s ideology will help to determine whether or not it enters government. Our measure of ideology is less precise than those of other studies, which use expert

16 James N. Druckman and Andrew Roberts

judgments or analyses of party manifestos to pinpoint specific ideological locations; such measures are not yet publicly available for Eastern Europe. We therefore sought to capture ideology with Armingeon and Careja’s (2004) coding of party types for all political parties in Eastern Europe. As previously noted, this measure classifies all parties in Eastern Europe into one of 17 party types, and each CSP is classified into one of these party types. Because CSP itself is not a party type, when we added party type controls we were able to control for the ideology of CSPs (see Table 2).

We ran all analysis with and without the incumbency variable, because including incumbency reduces the size of the data (from 593 to 475 parties), since the initial governments are excluded. Fortunately, the results are virtually identical, regardless of whether or not we include incumbency; thus, in Table 3, we only present the results that include incumbency (since it is itself significant). We therefore report results with 475 cases rather than 593. We report one-tailed tests because we clearly predicted a negative relationship between CSP, our key variable, and government participation.

Model 1 tests whether or not seat share and incumbency states predict participation i