Trustees of Princeton University
Electoral Systems and the Number of Parties in Postcommunist States Author(s): Robert G. Moser Reviewed work(s): Source: World Politics, Vol. 51, No. 3 (Apr., 1999), pp. 359-384 Published by: Cambridge University Press Stable URL: http://www.jstor.org/stable/25054083 . Accessed: 08/01/2013 08:59 Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at . http://www.jstor.org/page/info/about/policies/terms.jsp
. JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of content in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms of scholarship. For more information about JSTOR, please contact
[email protected].
.
Cambridge University Press and Trustees of Princeton University are collaborating with JSTOR to digitize, preserve and extend access to World Politics.
http://www.jstor.org
This content downloaded on Tue, 8 Jan 2013 08:59:23 AM All use subject to JSTOR Terms and Conditions
SYSTEMS AND THE
ELECTORAL
NUMBER OF PARTIES IN POSTCOMMUNIST
STATES
By ROBERT G.MOSER*
REND Lijphart concluded his book on electoral systems in con solidated
A
democracies
with
note on new democracies:
the following
a first electoral system has to be chosen which will When hopefully guide the new democracy s elections for a long time (or, in the case of a redemocratizing country, a new system thatwill hopefully work better than the old one), it is im portant to examine all of the options aswell as their advantages and disadvan to the extent
tages-Therefore,
democracies twenty-seven to electoral offer engineers old democracies.1
will
that
have
this
some
in the new
study
practical democracies
of
seventy
electoral
it may utility, than in these
have
systems more
in to
twenty-seven
in this statement is an assumption that the effects of elec Implicit toral systems are more or less universal and will therefore hold in new as well as old. Based on a rational choice democracies of understanding the behavior
essence
of voters
the influence of electoral systems in for the shape of party systems powerful explanation democracies. of decades the After testing, empirical of the major hypotheses made famous by Maurice Duverger?
has provided consolidated
and candidates,
a
that single-member elections tend to constrain the plurality a to a much greater ex of significant in parties operating polity tent than multimember (PR) systems?has proportional representation as remained intact.2 democratization has swept Thus, remarkably
namely, number
* An earlier version of this paper was presented at the annual meeting of the American Political Sci ence Association, D. C, August 28-31,1997.1 would like to thank Joel Ostrow, Frank Washington, and four anonymous reviewers for helpful comments on earlier drafts of this paper. Iwould Thames, also like to acknowledge research support provided by the University of Texas at Austin and an IREX Short-term Travel Grant used for fieldwork in Russia. 1 Arend Democracies, Lijphart, Electoral Systems and Party Systems: A Study of Twenty-Seven 1945-1990 2 Maurice
(Oxford: Oxford University Press, 1994), 151-52. in theModern State (New York and Activity Duverger, Political Parties: Their Organization 1963); Douglas W. Rae, The Political Consequences of Electoral Laws, 2d ed., (New Haven: Yale Wiley, H. Riker, "Duverger's Law Revisited," in Bernard Grofman and Press, 1971); William University Arend Lijphart, eds., Electoral Laws and Their Political Consequences (New York: Agathon, 1986); in Bernard Giovanni Sartori, "The Influence of Electoral Systems: Faulty Laws or Faulty Method?"
WorldPolitics 51 (April 1999), 359-84
This content downloaded on Tue, 8 Jan 2013 08:59:23 AM All use subject to JSTOR Terms and Conditions
360
WORLD POLITICS
southern Europe, Latin America, and now the former com through states of east central munist it has been quite nat and Eurasia, Europe
ural for politicians
and political
to new democracies knowledge tablish a new electoral law.3 to be seen, however, of electoral systems, based
It remains effect Western
will
democracies,
to apply this body of
scientists
the momentous
making
to es
decision
whether
the hypotheses the regarding on for the most part the experience of new democracies, in hold actually particu
larly in the very different social and political context of postcommunist states. The
of Russia, Ukraine, and Poland (Senate elec experiences can shown that and elections tions) plurality majoritarian produce s of that the reach very fragmented party systems, suggesting Duverger laws may be limited.4 Such findings have important practical implica have
new
as
to survive. also struggle They new institutionalism, for which important implications a seeks to grant institutions effect that is to some extent inde political of their social and political environments. pendent tions
for these
have
democracies
This
article
examines
the effect
in five (SMD) elections postcommunist in the electoral systems postcommunist some
of those
the standard
states
district single-member to ascertain the effects of
context.
states, most notably Poland pattern of party consolidation
It will
be shown
that
and Hungary, have followed over time in reaction to in
of electoral
have Ukraine, cases can be attributed
Cox,
of PR and
and others, most systems, while notably Russia not. The different in effects of electoral these systems
centives
found
they
theoretical
to different
levels
states. in postcommunist Building I argue that the constraining effects
of party institutionalization on the work of Sartori and of electoral
systems
on
the
number of parties will be mitigated by the institutionalization of the party
system.5
Institutionalization
is defined as
a process bywhich a practice or organization becomes well established and widely known,
if not
universally
accepted.
Actors
develop
expectations,
orientation,
and
and Arend Lijphart, eds., Electoral Laws and their Political Consequences (New York: Agathon, Grofman and Matthew S. Shugart, Seats and Votes: The Effects and Determinants 1986); Rein Taagepera ofElec toral Systems (New Haven: Yale University Press, 1989); Lijphart (fn. 1). 3 for New Democracies," Choices See, in particular, Arend Lijphart, "Constitutional Journal ofDe mocracy 2 (Wmtex 1991). 4 to "From Solidarity 3 (April 1992); Journal ofDemocracy Jasiewicz, Krzysztof Fragmentation," Robert G. Moser, The Impact of Parliamentary Electoral Systems in Russia," Post-Soviet Affairs 13, no. 3 (1997); Marko inMarch-April Elections 1994," Europe Parliamentary Bojcun, "The Ukrainian Asia Studies 47, no. 2 (1995). 5 Votes Count (Cambridge: Cambridge See Sartori (fn. 2); and Gary W. Cox, Making University Press, 1997).
This content downloaded on Tue, 8 Jan 2013 08:59:23 AM All use subject to JSTOR Terms and Conditions
ELECTORAL SYSTEMS/NUMBEROF PARTIES
361
behavior based on the premise that this practice or organization will prevail into the foreseeable future. In politics, institutionalization means that political actors have
Weak
clear
and
stable
about
expectations
institutionalization
of party electoral
and majoritarian plurality ductive effects in postcommunist of strategic voting cases studied here
the behavior
of other
actors.6
can
the failure of explain re to have their systems expected states. Moreover, in the level variance over time among the postcommunist
and learning can be attributed
systems
to a
level of party institu country's states increased postcommunist strategic showing over time are those with most the institutionalized party systems voting no to elec in the those that exhibited region, while signs of adaptation tionalization.
Those
constraints have weaker party systems. toral-system low party institutionalization of some postcommunist that PR systems with may suggests legal thresholds constraint tions.
Finally, states,
given the this study
a greater provide of parties than single-member district elec the status of political them with parties by providing
on the number
PR elevates
a
over nomination monopoly district elections.
procedures
not
found
in single-member
the failure of electoral effects systems to have their predicted Perhaps not in new, unconsolidated democracies should be surprising. Sartori has argued that strong electoral (for systems example, plurality systems) do not have their expected reductive effects on the number of national parties
in countries
with
unstructured
that strategic voting in all approximated political
argued
party certain
requires contexts.8
This
systems.7 Similarly, Cox has conditions that may not be article builds on these hy
support from postcommunist potheses by providing empirical in Latin America, The least institutionalized party systems
states.9 such
as
6
Scott Mainwaring, in the Third Wave of Democratization: "Rethinking Party Systems Theory of the Importance of Party System Institutionalization" (Paper presented at the annual meeting D. C., August 28-31,1997), American Political Science Association, Washington, p. 7.Mainwaring is the key variable distinguishing argues that party institutionalization party systems in established The
Western democracies and party systems in democratizing states, and he explains much of the variance states. to his measures of in democratic performance within the broad class of democratizing According institutionalization?which control over candidate nomi emphasize continuity of party organizations, nations, and the volatility of electoral support of parties?party systems in Eastern Europe and the are among the least institutionalized former Soviet Union of the late democratizers. See also Samuel Political Order in Changing Societies (New Haven: Yale University Press, 1968), 12. Huntington, 7 Sartori, (fn. 2), 62. 8 Cox (fn. 5). 9 is important because scholars have not identified many empirical cases that do not fulfill This Cox's necessary conditions for strategic behavior. Cox's primary example of a country with a weakly institutionalized party system defying the expected effects of a plurality electoral system is Papua New on the Indian a Guinea. See Cox (fn. 5), 85. Sartori concentrated multiparty example which produced and a dominant party system in the electoral realm, but the expected high levels of disproportionality were translated into seats. See Sartori (fn. 2), 55-56. system after votes
This content downloaded on Tue, 8 Jan 2013 08:59:23 AM All use subject to JSTOR Terms and Conditions
WORLD POLITICS
362 have
tended
tionalization,
thus
Brazil,
to
PR electoral
systems that foster party frac to the debate over the con challenge The majoritarian weakly systems.10
adopt
no
providing effects of PR and straining institutionalized party systems of postcommunist cases in which and majoritarian electoral plurality
states present a set of are used (in systems a system), providing unique and the re strategic voting
or as
a electoral part of mixed to test opportunity hypotheses regarding ductive effects of these systems.
dependently
Duverger's
Law and Its Limits
originally formulated three laws of the effects of electoral
Duverger systems:
representation tends to lead to the formation of many inde (2) the two-ballot majority system tends to lead to the for pendent parties,... mation of many parties that are allied with each other,... and (3) the plurality (1) Proportional
rule
final
The
to
tends
a
produce
two-party
law, Duverger
system.11
claimed,
was
to a
"the closest
sociological
law."12
Subsequent
studies
have
better
specified
the causal
nature
of these
correlations both empirically and theoretically. District magnitude number
of representatives
elected
from
was
each district)
found
(the to be
the decisive influence on disproportionality andmultipartism. Low dis trict magnitudes,
particularly
single-member
districts,
have
a
powerful
constraining effect on the number of parties and produce high levels of to seats. district High cause a not of do (but greater proliferation parties of parties) and produce lower levels of disproportional multiplication of effect electoral The systems was found to reside constraining ity.13 disproportionality allow magnitudes
in the translation
from votes
a
at the district level.14 Fi level rather than at the national directly was to not interact the the electoral found with, override, system nally, to structure in society.15 The major modifications Duverger's cleavage
most
laws arise from
these findings.
10 in Comparative Scott Mainwaring, "Brazilian Party Underdevelopment no. 134 (Notre Dame: stitute Working of Notre Dame, Paper, University 11 Maurice Duverger, "Duverger's Law: Forty Years Later," inGrofman 12 Ibid., 69. 13 and Shugart (fn. 2); Lijphart (fn. 1). 14Taagepera Sartori (fn. 2), 54-55. 15 District and Olga Shvetsova, "Ethnic Heterogeneity, Peter Ordeshook ber of Parties," American Journal ofPolitical Science, 38 (1994); Cox (fn. 5),
This content downloaded on Tue, 8 Jan 2013 08:59:23 AM All use subject to JSTOR Terms and Conditions
Perspective, 1990). and Lijphart
Magnitude, 203-21.
" Kellogg
In
(fn. 2), 70.
and the Num
ELECTORAL SYSTEMS/NUMBEROF PARTIES
363
at the district that electoral-system effects are manifested to the correlation level has been used to accommodate be exceptions tween at the For and national level. systems systems two-party plurality fact
The
retained three or four significant par example, Rae argued that Canada a ties electoral because of the existence of geo system plurality despite in Canada concentrated minority graphically parties. Minority parties one of the some dis major parties in supplanted a re tricts while third parties nationally.16 revision Such remaining on the number tained the plurality of parties system's causal influence at the elections still produced because plurality two-party competition district level. With concentrated minority parties, how geographically survived
ever,
because
the same
district,
they
two
resulting
were
parties
in
not
multipartism
in every
always the major players at the national level.
Cox has contributed greatly to this debate by explicating the condi tions under which
strategic
takes place. Using
voting
a rational
choice
model, he argues that strategic voting (and by implication strategic that involve entry and departure by elites) requires certain conditions time horizons, and the availability actors' motivations, of preferences, accurate information. Cox has that Thus, argued single-member plu to reduce the vote for minor if one or rality elections may fail parties more
of the following
conditions
arise:
(1) The presence of voters who are not short-term instrumentally rational; (2) lack of public information about voter preferences and vote intentions (hence about which candidates are likely to be "out of the running"); (3) public belief that a particular candidate will win with certainty; or (4) the presence of many voters who care intensely about their first choice and are nearly indifferent be tween
Even date
their
and
second
if conditions
lower
choices.17
are favorable
races at the district
for the establishment
level, the projection
of two-candi
of this bipartism
to the na
tional level is not assured. Rather it depends on the ability of parties to unite
in single nationwide If this party organizations. two candidates in the elections accomplished produced plurality across at the district to a multitude level may belong of different parties prominent
elites
is not
the country. Cox cites institutional the direct elec forces, most notably as tion of a powerful for the na national executive, primary pushing to tionalization of parties that is essential of Duverger's the realization third
law at the national
tionalization, 16 Rae 17 Cox
however,
level. In the absence the existence
of a general
of nationwide
(fn. 2), 95. (fn. 5), 79.
This content downloaded on Tue, 8 Jan 2013 08:59:23 AM All use subject to JSTOR Terms and Conditions
theory of remains parties
na a
364
WORLD
necessary
POLITICS
condition
only for the constraining to be felt at the national level.18
effect
of plurality
electoral
systems of these preconditions The presence for to the national of local projection bipartism tionable communist
initial during states with electoral
previous may
voters
deny
and the strategic behavior level are particularly ques elections in new democracies, in post especially little or no democratic tradition. The absence of
experience and elites
and accurate
information alone polling to information behave necessary
the
strategically.
Most
importantly, the lack of well-established
the ability of voters serve as the primary mechanism
dermines
ion, while
electoral
and elites
to channel
are a
systems
to behave
secondary
political parties un Parties
strategically.
and aggregate public opin mechanism the influencing
number of viable political parties.19 If significant political parties do not or political nor In weak of public opinion. represent ganizations large segments no voters the absence of identification leaves with party systems, party of characteristics candidates and patron cues, other than the personal exist
cannot
they
age, as to how
aggregate
elites
political
to cast their votes. Due in the most
unstable
organizations enter and leave the scene ally
into nationwide
to the nature of party transitory democracies, parties continu in tandem with the political clout
new
(usually no and provide between electoral continuity periods. voters to cultivate Such little for lasting opportunity provide most For for one party or another, uncommitted. leaving preferences in Russia survey research has put the number of independent example, a party identification at 78 percent, voters without 13 compared with of their
leader) conditions
percent for theUnited States and 8 percent forGreat Britain.20 It is dif to the ficult, then, for us to attribute voting preferences ers in unconsolidated in the same way we democracies to voters in consolidated with democracies preferences concrete party systems. Without strategic preferences, cess based If one essary
of vot
majority attribute
voting institutionalized
voting
as a pro
seems very on a rank unlikely. ordering of preferences the conditions consolidated democracies, only
nec
examines
for strategic
voting
are
usually
approximated.
Exceptions
are few
and easily accommodated through special conditions for geographically parties. This in the larly postcommunist
concentrated
18 Ibid., 182-93. 19 See Sartori (fn. 2), 55-56. 20 Richard Rose, Steven White,
is not world
the case for new democracies, particu where parties are not well-developed
and IanMcAllister,
How
Russia
Votes (New York: Chatham
1997), 135.
This content downloaded on Tue, 8 Jan 2013 08:59:23 AM All use subject to JSTOR Terms and Conditions
House,
ELECTORAL SYSTEMS/NUMBEROF PARTIES and voters
It would
systems.
cannot
and elites
easily respond a mistake to assume
be
to incentives that
365
from
electoral
institutional
effects
found in established democracies will be replicated in the very differ ent social
of new democracies
context
in eastern
Europe
Mdced Electoral
Reconceptualizing
and Eurasia.
Systems
states reflect the influence of postcommunist of some sort states the German of mixed electoral system. Many employ PR contests in both in elected system with representatives party-list or two-round districts and multimember large plurality majoritarian
The
electoral
systems
in districts. No exclusively systems single-member plurality in single-round, all representatives district races, single-member certain obstacles in testing theo and this for those interested produces states. ries of electoral in postcommunist systems Mixed electoral of electoral the systems complicate categorization elections elect
and the analysis of effects. Most scholars have thought of a systems as a PR to mixed electoral modified form of curb the system designed a accurate for is Such party proliferation.21 potential quite conception seats for those mixed that utilize that interlock systems compensatory the two halves
of the electoral
system
does. Another
into one, as the German used in Russia and Croatia,
system, type of mixed seats. Rather, it calculates ever, does not use compensatory the two portions of the system and allows parties separately seats won
other
tier.
in each Shugart
half
of the
of the
system how
the vote to
results
of
keep all of the
system regardless has argued that such a system is actually a modified to Since there is no mechanism the PR tier prioritize assumes that the effects of the plurality por tier, he
system. plurality over the plurality tion of the system will
override
the more
feeble
PR tier.22
levels of mul systems also tend to produce moderate as have intended. In practice, tipartism and disproportionality designers reason such results are laudatory and are a major the system has why Mixed
electoral
been
replicated electoral systems, tween
most
so
widely however,
in east add
and Eurasia. Mixed Europe to the dichotomous debate be
central little
PR and has shown that the systems. Western plurality scholarship on is effect of electoral the number of parties systems powerful
the constraining 21 22Lijphart Matthew
effect
of single-member
plurality
systems.
Taken
at
(fn. 1), 39-46.
Shugart, "Building the Institutional Framework: Electoral Systems, Party Systems, and no. 2.26 and European Presidents," Working Studies, 1994), Paper, (Berkeley: Center for German 10-15.
This content downloaded on Tue, 8 Jan 2013 08:59:23 AM All use subject to JSTOR Terms and Conditions
366
WORLD POLITICS
face value,
the electoral
of east central
systems
Europe
and Eurasia
do
not offer any persuasive cases to test this hypothesis. The high levels of states can be attributed to party fractionalization usually found in these structure and an in the combination of a fractionalized social cleavage state that no postcommunist system. Given strong electoral sufficiently the strongest exclusively system?single-member employs plurality can not test whether a elections?one electoral system would plurality have
been
able
to curb
the party
fractionalization
in many
found
such
states.
Mixed electoral also offer certain opportunities for systems, however, a one can treat the comparison of electoral I that mixed argue systems. electoral system that employs separate votes for each tier of the system as its name a mixture of two separate electoral exactly implies?as a side side. Such arrangements by operating conceptualization provides or PR tiers a to the of effects separate study plurality unique opportunity of a mixed variables nomic
system while holding such as culture, social
constant
other and
cleavages,
intervening possible the level of socioeco
development.
This
conceptualization
of mixed
electoral
systems
follows
the
method of controlled comparison, which studies "cases that differ with regard
to the variables
one wants
to
investigate,
but
similar with
regard
to all other important variables thatmay affect the dependent variables; can then be treated as control vari variables these other important a ables."23 Although tool, cases that are simi powerful methodological are in certain variables lar except very hard to find. This independent studied the interre has been used effectively method by Putnam, who in Italy, institutions and social environment political lationship between over time in electoral examined and by Lijphart, who systems changes a that when countries.24 within individual country argued Lijphart its electoral system "many potentially explanatory important changes can be controlled in the sense that they can be assumed not to variables the same country, the same political differ or to differ only marginally: elec combination the same voters, and so on."25 The by mixed parties, and electoral toral systems of dichotomous (PR systems plurality/ma joritarian)
in a single
social
environment
contributes
a
significant
dimension of comparability not found in Lijphart's study.Despite theoretical
malleability,
electoral
systems
are resilient
23 (fn. 1), 78. 24Lijphart Robert Putnam, Making Democracy Work' Civic Traditions Press, 1993); Lijphart (fn. 1). University 25 Lijphart (fn. 1), 78.
inModern
This content downloaded on Tue, 8 Jan 2013 08:59:23 AM All use subject to JSTOR Terms and Conditions
institutions.
Italy
(Princeton:
their All
Princeton
ELECTORAL SYSTEMS/NUMBEROF PARTIES
the same broad categories changes within the class of PR systems. Mixed elec systems, mostly within to the the diametri compare systems provide unique opportunity
of Lijphart's of electoral toral
cases
367
involved
cally opposed categories of PR and plurality elections upon which the debate
over electoral
systems has been based. to mixed electoral systems
is not without approach precedent. some scholars have test for the PR and strategic voting, compared to one another. mixed tiers of Germany's electoral system plurality and Cox have all used the mixed electoral Fisher, Jesse, Bawn, system of to show the greater level of in West the plu strategic voting Germany This
To
rality tier than in the PR tier. Such analyses have consistently found that inWest received more votes large parties Germany than in the PR tier and vice versa for smaller parties,
in the tier plurality as the vot strategic
ing hypothesis would predict.26 By showing that voters behave differ tiers of a mixed such studies bolster the system, ently in the separate case for tiers of the mixed electoral different systems separately treating as as a vote is cast in each tier. long of different tiers of mixed controlled electoral The sys comparison is not without tems, however, problems. The greatest problem potential cross is cross-contamination the two cases compared. Unlike between across time, the electoral systems analysis Lijphart's test are not in of one another. They this being compared independent form two halves of one electoral system for the same legislative body in same matter the two halves of the the election. No how independently into two sys of mixed electoral the separation system operate, systems two tiers of tems for the purpose remains artificial. The of comparison one another some extent. a mixed to will electoral affect system surely For example, that run in the PR tier of the election have small parties national
analysis
or
costs
of electoral and could be competition as well with districts in the single-member expected in terms of seats. Or, one or two large parties little regard to payoffs the dominate races, having coattail effects may single-member plurality the ef that produce greater vote shares in the PR contest. Consequendy, a PR fective number of parties half of mixed in the system may tend to already
assumed
the entry to run candidates
be lower than if the system had a strictly PR electoral system. 26
Vote Thesis," Comparative Politics 5, no. 2 (1974); Eckhard Jesse, Steven Fisher, "The Wasted of the Federal Elections from 1953 to in the Federal Republic of Germany: An Analysis "Split-voting 1987," Electoral Studies 7, no. 2 (1988); Kathleen Bawn, "The Logic of Institutional Preferences: Ger no. 4 (1993); Cox man Electoral as a Social Choice Outcome," American Journal of Political Science 37, (fh. 5), 82-83; Samuel H. Barnes, Frank Grace, James K. Pollack, and Peter W. Sperlich, "The German 1962). Party System and the 1961 Federal Election," American Political Science Review 56 (December
This content downloaded on Tue, 8 Jan 2013 08:59:23 AM All use subject to JSTOR Terms and Conditions
368
WORLD
POLITICS
to in important keep the danger of cross-contamination it does not ruin this experiment. Incentives for strategic mind, entry in the single-member tier may be and withdrawal district by elites While
weakened
it is
the costs
because
voting, Strategic West Germany.
and payoffs should remain
however, Rational
voters
will
are of competition changed. intact as shown in studies of
still have
incentives
to abandon
small parties in the plurality tier in favor of large parties with a better chance
of winning.
district
elections in favor
ality time
the mechanical effects of single-member Moreover, should remain intact producing greater disproportion PR tier. Over than in the corresponding of large parties should produce strategic behavior in the translation of votes into seats.
such mechanical
ties who
are
effects
punished
The Effects
of pr Systems
Do
electoral systems states and consolidated
affect
in Postcommunist
by par
States
in of parties postcommunist answer this To similarly? question,
the number
democracies
states will be examined. Three in five postcommunist electoral systems a mixed electoral and Lithuania, of these, Russia, Hungary, sys employ tem of various combinations elected of PR and plurality or majoritarian as two tiers of these systems will be seats. The individually analyzed separate scribed Poland These and
de method controlled-comparison with electoral systems, single-tier also be studied.
the systems following countries above.27 Two and Ukraine, cases were
single-member of varying
will
selected district levels
examples
to of PR roughly equal coverage provide cases The also electoral systems. provide of party
institutionalization.
Poland
and
Hungary have the highest level of party institutionalization. Parties mo nopolize
the nomination
ginal phenomena. While
process
making
independents
rare and mar
there is significant volatility in popular support
or to election, tenden ideological major parties and Ukraine the time.28 Russia oppo represent site end of the spectrum. Parties do not control the nomination process in single-member the field of candidates dominate and independents
for parties from election cies have survived over
27
data was not available for the single In the case of the 1992 Lithuanian elections, district-level PR tier of the 1992 Lithuanian election was included in the district tier. Therefore, only the
member
study. 28 inHungarian Six major parties have remained dominant politics since the end of communist rule. has argued that the political system the party system in Poland has been more fluid, Tworzecki While is actually more consolidated around a small number of political tendencies. Hubert Tworzecki, Parties of Poland (Boulder, Colo.: Westview and Politics in Post-1989 Press, 1996). The recent consolidation reformist forces into the AWS would tend to support this assessment.
This content downloaded on Tue, 8 Jan 2013 08:59:23 AM All use subject to JSTOR Terms and Conditions
ELECTORAL SYSTEMS/NUMBEROF PARTIES district
elections.
and long-lasting resents a middle cases. Like
369
are fluid, and voters tend not to have a strong a with identification Lithuania rep particular party.29 the east European and former Soviet ground between Parties
to monopo and Hungary, Lithuanian parties tend in lize nominations but are not as stable or firmly grounded society.30 on the electoral 1 Table basic information systems of the provides
five
Poland
countries
cases
Each
individual
election
for which
data were
as an individual
case. sys By separating mixed PR and tiers, fifteen component plurality/majoritarian result: nine PR elections and six elections conducted under a plu
available will tems
analyzed. be treated
into
their
measures common to the electoral system.31 Two majoritarian to examine will be these the effective literature used systems: systems number of parties and the index of The least-squares disproportionality. a measure of parties provides effective number of party system frac rality
or
tionalization seats
The
by counting by their parties weighted so that very small not count as much do parties
shares as
of votes
or
large parties.32
level of disproportionality produced by an electoral system is the
deviation
between
tion
the proportion of seats
and the proportion to cross-national cording
studies
in an elec party receives in the gets legislature. Ac to be tends highest disproportionality of votes
a
it actually
in plurality more parties
small parties and reward large systems, which penalize than multimember district PR systems. Like the effective index of the least-squares measure, number-of-parties disproportional 29 is perhaps best captured by the unusually high level of split-ticket voting in This characteristic Russia. In 1993, 70 percent of voters planned to split their votes in the PR and plurality tiers, voting for different parties or for a party and an independent candidate in the two halves of the election. Only 19 (fn. 20), 139-40. Rose, and McAllister percent planned to vote a straight party ticket. White, 30 Kitschelt provides an index of the chances of program-based party formation for postcommunist to the classification of the cases in of the level of party institutionalization states, which corresponds s scale this study. Using Kitschelt and Poland have the highest scores at 5.5 and 5.0 respec Hungary are lower at 3.5 to 5.0. Russia, Ukraine, and other Soviet republics tively. The Baltic states marginally in Postcommunist have a much lower score of 0.5. Herbert Kitschelt, "Formation of Party Cleavages no. 4 (1995), 457. Moreover, Evans andWhitefield Democracies," argue for a similar Party Politics 1, classification of postcommunist states' potential for the development of stable party systems, with east central Europe (Poland, Hungary, the greatest potential for stable and the Czech Republic) possessing and other Soviet successor states followed by the Baltic states with Russia, Ukraine, party development of stable party systems. Geoffrey Evans and Stephen having much lower chances for the establishment in Eastern Europe," British Journal of Politi the Bases of Party Competition Whitefield, "Identifying cal Science 23, no. 4 (1993). 31 a vote is cast are seats in compensatory Only tiers in which analyzed here. Therefore, Hungary and Poland calculated on the basis of previous votes are not included. 32 The effective number of parties index is calculated by squaring the proportion of the vote or seat shares of each party, adding
these together,
See Lijphart
andTaagepera
1 by this total:
then dividing
Nv=l/X(Vl2)orN=l/Z(Si2) (fn. 1), 67-72;
and Shugart
(fn. 2), 77-81,104-5.
This content downloaded on Tue, 8 Jan 2013 08:59:23 AM All use subject to JSTOR Terms and Conditions
WORLD POLITICS
370
TABLE1
Country Russia
Hungary
in Five Post-Communist
Systems
Electoral Electoral
PRTier
System
Plurality/Majoritarian Tier 225 seats elected by
225 seats elected in one nationwide district with 5% legal threshold
mixed
same
Lithuania
70
mixed
seats
rules
second
single-member top three go round
plus
to
any
in one
71
of first-round
seats
elected
vote
under
nationwide district with
two-round
4%(1992)and5%(1996)
in single-member two go districts; top round second
391 seats elected in 37 multimember districts;
PR
under
candidate with 15% or more
legal threshold
Poland
elected
majoritarian
in
districts;
thresholds
elected
seats
176
two-round
in one nationwide district with
single-member
districts
seats elected
compensatory
in
plurality
152 seats elected in 20 multimember districts with 4% (1990) and 5% (1994) legal threshold; 58
mixed
States
majoritarian
rules
to
none
seats 69 compensatory elected in one nationwide
district; no legal threshold
(1991),5%(1993)
Ukraine
450 seats elected under
none
majoritarian
two-round
majoritarian in single-member districts; top two go to rules
second SOURCE: Cox v 1993 godu
round
(fh. 5), 50-54; Bojcun (fn. 4); "Polozhenie o vyborakh deputatov Gosudarstvennoy dumy in 1993), Rossiiskie vesti, October for elections of deputies of the State Duma
(Provisions
12,1993.
ity weights tions have
the deviations less effect
than
between
seats and votes
so that small devia
large ones.33
33 share differ is calculated by squaring the vote-seat index of disproportionality The least-squared ences and adding them together; this total is divided by 2; and then the square root of this value is taken: LSq For discussion
of these two measures
see
= SqRtofl/2*I(vrs.)2 Lijphart
(fn. 1), 67-72;
and Taagepera
77-81,104-5.
This content downloaded on Tue, 8 Jan 2013 08:59:23 AM All use subject to JSTOR Terms and Conditions
and Shugart
(fn. 2),
ELECTORAL SYSTEMS/NUMBEROF PARTIES 2
Table for Proportional
Effects
Representation
Effective # of Electoral
Legal Case
371
Tier/Election
Effective # of
Least-Squares
Index of Disproportionality
Parliamentary
Threshold
Parties (N )
Parties (N)
5% 5%
7.58 10.68 12.50
6.40
4.94
20.56
Russia 1993 Russia 1995 Poland 1991 Poland 1993 Poland 1997 Lithuania 1992 Lithuania 1996 Hungary 1990 Hungary 1994
5%
4.59
3.32 10.87 3.86 2.95
4%
4.10
2.86
5%
7.94
3.16
17.81 10.63 7.55 16.34
4%
6.71
4.31
9.34
5%
5.49
3.73
8.53
none 5%
9.80
6.11
Cox (fn. 5), 159. Poland: Frances Millard, ed., "Poland," in Bogdan Szajkowski, ofEastern Europe, Russia and the Successor States (Essex, U.K.: Longman Group, 1994), 313-42. Hungary: Benoit Lars Johannsen, and Anette Pedersen, (fn. 40). Lithuania: Ole Norgaard, of Multi-Party "The Baltic Republics Estonia, Latvia and Lithuania: The Development Systems," in SOURCE: Russia:
Political Parties
ed., Political Parties 60; Lithuanian
Szajkowski,
Bogdan
1994),
Longman Group, online at http://rc.lrs.lt/rinkimai/seim96.
Table nist
2 shows
states. The
tern and the Successor States (Essex, U.K.: ofEos Europe, Russia Seimo rinkimo '96," dataset, Seim, "Lietuvos Respublikos
of PR systems striking characteristic
the effects
most
the number
which
postcommu of all the states is the level
party proliferation is expected in PR
of party fractionalization. While elections, states outstrips fective number
in the selected
of significant found in anything
parties
developed for the cases parties
of elective
in
postcommunist The average ef studied here is 7.71,
operating countries.
is higher than all electoral systems in consolidated democracies
except Belgium
and nearly
twice
the average
for consolidated
democra
cies using PR (D'hondt method).34 It is also higher than inmost new democracies with the exception of Ecuador and Brazil, both of which have
ten effective
around
the high
Clearly, in part to fragmented
parties.35 states is due in postcommunist of parties structures. However, the high and fluid cleavage
number
level of disproportionality used
in every
electoral
case but one)
produced by legal thresholds (which were suggests
an absence
of strategic
behavior
in
those cases with the highest level of disproportionality. The average level of disproportionality produced by PR systemswas 11.31 (11.96 ex more than cluding Poland 1991, which had no legal barrier). That is 34 Lijphart (fn. 1), 96,160-61. 35 Cox (fn. 5), 309-11.
This content downloaded on Tue, 8 Jan 2013 08:59:23 AM All use subject to JSTOR Terms and Conditions
372 twice dated
WORLD the average democracies
POLITICS
level of disproportionality than and is also higher
in consoli for PR systems the average of majoritarian
systems.36 to
According proportionality,
and Shugart's Law of Conservation of Dis Taagepera the number of parties and hence the level of dispropor
tionality produced by a systemwill be underestimated by the actual vote shares make
to real voter because voters and elites (as opposed preferences) decisions the vote even strategic favoring larger parties before
takes place.37 Thus,
parties
postcommunist
ticipated the disproportionality
and voters
should
have
an
that would be produced by the legal
threshold
toward larger parties capable of imposed and have gravitated the threshold. Then there would have been fewer electoral overcoming on sure losers, and votes would have been wasted parties, fewer dispro would have been lower. This process should increase over portionality time as voters and elites learn the rules and adapt to the incentives of the system. Among varied ways.
our cases Poland
this phenomenon seemed to occur in significantly to the the strongest evidence of adaptation provides
incentives of a legal threshold and learning over time. Poland had the party proliferation was legal threshold. There
highest
there was no among our cases in 1991 when limited consolidation and very high dis very
proportionality in 1993 when a 5 percent legal threshold was first in an absence of troduced, strategic behavior. But in 1997 elites suggesting to learn from the effect of the 5 percent seemed devastating legal on small in broad electoral and consolidated threshold parties, they blocs
the effective
cutting
number
of electoral
parties
in half. As
a con
sequence, the level of disproportionality dropped significantly because there were
A smaller small parties left without representation. over in the effective number of parties and disproportionality decrease time in Hungary also shows support for the learning of strategic be havior both
fewer
over countries
time.
Russia
the number
and Lithuania of parties
show an opposite in the PR tier increased
trend.
In
substan
tially from the first to the second election raising disproportionality exponentially. accounts What
in the learning of strategic behav for this difference our ior among cases? Party fragmentation does not seem to be the cul on the one hand, had the most fractionalized party system prit. Poland,
in its first election, yet it displayed the greatest amount of adaptation to 36 (fn. 1), 96. 37Lijphart and Shugart Taagepera
(fn. 2), 123; Lijphart
(fn. 1), 97; Cox
(fn. 5), 173-78.
This content downloaded on Tue, 8 Jan 2013 08:59:23 AM All use subject to JSTOR Terms and Conditions
ELECTORAL SYSTEMS/NUMBEROF PARTIES over
constraints
electoral-system
time. Hungary
373 a
also has
relatively
high degree of party fractionalization but a low degree of dispropor tionality. On the other hand, Lithuania had the lowest degree of party fragmentation in 1992 but almost doubled the number of significant electoral
parties
at the next
election
threshold from 4 to 5 percent.
an increase
despite
in the
legal
I argue that the difference in learning over time is caused by the de most of institutionalization. has the stable gree party party Hungary same six in this It is dominated system sample. by the major parties
that have consistently crossed the legal threshold and left little room for
tumul party system has been much more marginal parties. Poland's but Tworzecki that underneath has its fluid tuous; surface, Pol argued ish society is divided and is evolving along several dominant cleavages a
toward
is different
structure similar toWestern cleavage in countries of the former Soviet
tion of the Baltic
these
countries
or even statehood governance independent is Politics and century. polarized, parties with political social constituencies have been slow to emerge. In Russia
with
experience in the twentieth identifiable the most
republics, democratic
situation Europe.38 The Union. With the excep recent have no significant
the extreme ends of the political occupy parties a small of dedicated followers. This leaves spectrum minority a broad and center that encompasses voters the of amorphous majority a multitude a lesser ex and is represented of minor fluid, by parties. To saw the former communist which tent, the same is true for Lithuania, a return to in 1992 with of seats, only to be re power party majority developed and have
of the right at the next elec over time. Under of the center increasing such voters to defect from the conditions incentives ignore marginal parties no clear because for a major party representing ei they have preference placed by the Lithuanian tion, with minor parties
Conservatives
ther end of the political spectrum. Moreover, which
parties
in most This
are viable
or "out of the running" are the "undecideds."39
opinion polls is not to say that voters
and elites
it is difficult to decide since
in Soviet
the largest group
successor
never respond to the incentives of legal thresholds. The
con by social we of political parties. Thus, to in response strategic behavior
in these states, however, strategic behavior that retard the institutionalization ditions may
expect
Lithuania
to
experience
states will
likelihood of
is undermined
38Tworzecki(fn.28),194. 39 This is particularly true in Russia where surveys regularly report that 40 percent of respondents an swer "don't know" to questions about whom they will vote for in the next election, more than twice the and McAllister (fn. 20), 141. percentage supporting the most popular political party. Rose, White,
This content downloaded on Tue, 8 Jan 2013 08:59:23 AM All use subject to JSTOR Terms and Conditions
374
WORLD POLITICS sooner systems there is a better
electoral because tion
than Russia
or other
environment
for future party
former
Soviet
republics institutionaliza
there.
of Plurality and Majoritarian Postcommunist States
Effects
Systems
in
even more
ex conundrum perplexing by postcommunist presented has been in and elec party proliferation perience plurality majoritarian have been deemed the most reductive electoral tions, which powerfully
An
in the
systems
In two
literature.
not
and Ukraine,
cases, Russia
only
have single-member district elections failed to produce local bipartism at the district chanical due
to
level, but
they also have the number
effect?reducing
failed
is seen
disproportionality?that
to
produce
a dramatic
me
of parties entering the legislature in comparative experience. The
failure of strategic behavior in single-member district elections typically leads to an overwhelming
advantage
for the largest party
in the system
and thus to a dominant party system as found in India until the 1990s. Russia
and Ukraine, therefore, in the legislature arising
ation
are from
cases of party prolifer district elections. single-member
truly unusual
Table 3 shows the effects of single-member district plurality andma joritarian elections. In three cases (Lithuania 1996, Hungary 1990 and 1994) there was a significant mechanical effect in the SMD tier. In all cases
these half
the effective
the effective
number
of parliamentary parties of elective parties. Moreover,
number
was the
less than level
of
disproportionality in SMD elections in Lithuania and Hungary was extremely high with values equal to or higher than themost dispropor tional
PR election
isons
further
district
in the study, Russia 1995. Within-country compar the increased reductive force of single-member support
elections
number
in these
of parliamentary
cases. parties
In Lithuania
and Hungary the effective in the single-member district elections
was lower and the level of disproportionality higher than in the corre PR tier, just
sponding
Hungarian number tiers
expect
from
the
experience is particularly striking. Although
of electoral
in the 1990
mentary
as one would
parties
parties and 1994
produced
literature.
The
the effective
was
PR and majoritarian quite similar in the the effective number of parlia elections, in the single-member districts was less than
half of that in the PR tier. Indeed, the impressive victories of theHungar ianDemocratic Forum (MDG) in 1990 and theHungarian Socialist Party (MSZP)in 1994 were driven in large part by seatswon in single-member districts. The MDF won 67 percent of the 176 single-member district
This content downloaded on Tue, 8 Jan 2013 08:59:23 AM All use subject to JSTOR Terms and Conditions
ELECTORAL SYSTEMS/NUMBEROF PARTIES
375
TABLE3 for Plurality
Effects
and Majoritarian Effective
Effective #of
*of Electoral Case
Electoral
Russia 1993 Russia 1995 Lithuania 1996 Hungary 1990 Hungary 1994 Ukraine 1994
Parties (N ) 6.13 5.00
maj.
3.23 6.10 7.75
maj.
7.14
2.03
maj.
6.17
1.35
maj.
2.46
4.15
plurality plurality
Least-Squares
Parliamentary
Parties (N )
System
Tier/Election
Index of Disproportionality
Effective # of Candidates
4.27 11.09 20.37 31.88 40.89
7.06 7.36 5.59
3.06
n/a
5.97 5.64 5.44
Information 8, 1993; Service, Report on Eurasia, December Foreign Broadcasting na v Commission of the Russian Federation, "RezuTtaty golosovaniya vyborakh on elections Dumu po odnomandatnym of (Results Gosudarstvennuyu voting izbiratel'nym okrugam" to the State Duma in single-member Election report, Central voting districts) (Unpublished o 1 okruzhnykh komissii Commission, Moscow, 1994); "Dannye protokolov No. izbiratel'nykh SOURCE: Russia:
Central
Electoral
rezultatakh vyborov deputatov Gosudarstvennoy Federatsii Dumy Federal'nogo Sobraniya Rossiiskoy no. 1 of distict electoral of (Data vtorogo sozyva po odnomandatnym izbiratel'nym okrugam protocol on the results of elections of commissions deputies of the second State Duma of the Federal Assembly of the Russian "Dannye protocol Hungary:
Federation electoral district), Rossiiskaya gazeia, January 17,1996; by single-mandate no. 2 ob izbiratel'nomu po federal'nomu protokolov itogakh golosovaniya okrugu (Data of no. 2 on results of vote for federal electoral okrug)," Rossiiskaya gazetay January 24, 1996. Benoit and Anette Pedersen, "The Baltic (fn. 40). Lithuania: Ole Norgard, Lars Johannsen, Latvia
Estonia,
Republics
ed., Political
Szajkowski, Longman
1994),
Group,
and Lithuania:
of Multi-party in Bogdan The Development Systems," of Eastern Europe, Russia and the Successor States. (Essex, U.K.: Seimo rinkimo '96," at 60; Lithuanian Seim, "Lietuvos Respublikos Ukraine parliamentary dataset this online (fn. 4), 239; Bojcun parties:
Parties
http://rc.lrs.lt/rinkimai/seim96. is an estimate of partisan affiliations in April Project, districts
for 338 deputies who were elected after the first run-off election for Election parties: International Foundation Systems Ukrainian at this is a dataset of first-round election results for all 450 http://ifes.kiev.ipra.ua;
1994. Ukrainian
elective
1996, in Ukraine.
least-squares
to the incongruence Due of the data for electoral index of disproportionality for Ukraine was not computed.
and parliamentary
a
parties
seats on 24 percent of the first-round vote in 1990 while the MSZPwon 86 percent of those seats on 31 percent of the first-round vote.40 a different situation Russia and Ukraine present altogether. The
ef
fective number of parties reaching parliament (4.15 for Ukraine 1994 and over 5.00 for Russia 1993 and 1995) show none of the dramatic mechanical probably
effect
found
underestimates
in the other the amount
cases. Moreover, this measure of party fractionalization pro
40 from data in Kenneth Benoit, "Votes and Seats: The Hungarian Electoral Computed in G?bor T?ka, ed., The 1990 Election to theHungarian the 1994 Parliamentary Elections," and Data. Dataset, (Berlin: Edition Sigma, forthcoming). Assembly: Analyses, Documents fas. harvard.edu/stafi7ken_benoit. http://data.
This content downloaded on Tue, 8 Jan 2013 08:59:23 AM All use subject to JSTOR Terms and Conditions
Law
and
National online
at
376
WORLD POLITICS
duced
by Ukraine's
system
majoritarian
in which
independents
made
up the largest "bloc"of representatives entering parliament (40.2 percent) a cohesive and can hardly be considered group. They were, however, a measure treated as in the of the effective num single entity computing
ber of parliamentary parties.41The high effective number of candidates per district in Russia shows that Russia's plurality system failed to have its most
basic
an effect What
effect?local
Sartori
claimed
two-candidate even held the cases
distinguishes
contests
at the district
level?
in unstructured
party systems.42 and Russia from the others
in Ukraine
in single-member is the of district candidates proliferation independent a small In Ukraine, of the candidates elections. (11 per only minority were vot nominated rather than cent) by parties by groups of officially ers or worker collectives.43 This percentage, overestimates the however, a number of independents because many candidates with partisan affil iation
chose
to be nominated rather
membership sure of partisanship vote went Russian
than party
methods. by nonpartisan as the more nomination
Using accurate
party mea
61 percent of the first-round candidates, among to the This candidates.44 independent corresponds roughly in which 48 candidates percent experience nonpartisan gained to
of the vote in 1993 and 36 percent in 1995. In Lithuania independents won
only
3 percent
of the vote.
In Hungary
the vote
for
independent
candidates dropped from 7 percent in 1990 to only 2 percent in 1994. in Ukraine and Russia not only made up a large propor Independents tion of candidates for office, for the they also accounted competing in the winners. winners of the of percent Fifty-two largest proportion
Russian elections in 1993 were independents and 34 percent in 1995, while 40 percent of Ukraine's winning candidates did not belong to a party. In Lithuania's 1996 andHungary's 1990 elections, independents made
up only
6 percent
and 3 percent
of the winners,
respectively.
41 on estimates of The effective number of parliamentary parties is based partisan affiliation in the were of Ukrainian for 338 450 who deputies successfully elected after the first run-off elec parliament tion inApril 1994. Bojcun (fn. 5), 239. Only 338 of the 450 district elections were declared valid after the first run-off because the other districts failed to fulfill the required criteria of both 50 percent par candidate. The rest of the seats were filled in special ticipation and 50 percent support for the winning the fluid and unstable nature of partisan affiliation in elections held until 1996. Given make-up In the Russian this figure should be considered only an estimate of party fractionalization. Ukraine on in parliamentary case, the effective number of parliamentary factions, parties is based membership in the legislature. which renders a more accurate reflection of party fractionalization 42 Sartori (fn. 2), 62. 43 Bojcun (fn. 4), 233. 44 for Election Systems (iFES), which This figure is based on data from the International Foundation voter group, workers and the for a candidate listed both the mode of nomination collective) (party, party affiliation of the candidate. The latter was used to estimate the percent of vote going to inde ipri. kiev. ua. pendents. See IFES, online, http://ifes.
This content downloaded on Tue, 8 Jan 2013 08:59:23 AM All use subject to JSTOR Terms and Conditions
ELECTORAL SYSTEMS/NUMBEROF PARTIES There
were
There
no
is some
tion procedures
dates without
377
1994 election.45 winners in Hungary's nonpartisan over nomina are control that evidence gaining parties at least in Russia where the number of winning candi
a partisan affiliation declined in 1995. Both Russia and
to have a continue of independent however, Ukraine, large contingent a weakness in candidates winning elections, which signals continuing countries in their party systems not found in the other postcommunist this study.46 is just another candidates The proliferation of independent sign of as in Russia and Ukraine the lack of party institutionalization opposed not to the other cases in the and Russian parties have study.47 Ukrainian in in districts these the nomination controlled process single-member countries
and have
not
enjoyed
a consistent
level of significant
support
nationwide (with the possible exception of the Communist Party of the one or in 1995). Consequently, without beginning or to benefit from two strategic voting disproportionality, large parties the number of candi elections fail to constrain district single-member their own districts dates per district. Rather, produce single-member candidates and allow individ of partisan and nonpartisan proliferation name resources to find financial ual candidates and with recognition
Russian
Federation
success
in parliament, of party affiliation. Once regardless as a not act does unified group but candidates independent ther. Some join parties they eschewed during the campaign;
this mass
of
splinters fur others form
new
to electoral associations factions unrelated fielding parliamentary in the members of still others become atomized candidates campaign; a affiliation with any consistent larger group. parliament without 45 on the vote for and their success in gaining seats are based on the following independents Figures sources: IFES (fn. 44); Bojcun Information Service, Report on (fn. 4). Russia: Foreign Broadcasting of the Russian Federation, Eurasia, December 8, 1993; Central Electoral Commission "Rezul'taty na v Dumu po odnomandatnym izbiratel'nym okrugam," Gosudarstvennuyu golosovaniya vyborakh re in single-member (Results of voting on elections to the State Duma voting districts) (Unpublished 1 No. izbi Central Election 1994); Commission, Moscow, port, okruzhnykh "Dannye protokolov o rezultatakh Sobraniya Dumy Federal'nogo vyborov deputatov Gosudarstvennoy ratel'nykh komissii Federatsii vtorogo sozyva po odnomandatnym izbiratel'nym okrugam" (Data of protocol Rossiiskoy on the results of elections of no. 1 of district electoral commissions deputies of the second State Duma of the Federal Assembly electoral district), Rossiiskaya of the Russian Federation by single-mandate izbi po federal'nomu gazeta, January 17, 1996; "Dannye protokolov No. 2 ob itogakh golosovaniya ratel'nomu okrugu" (Data of protocol no. 2 on results of vote for federal electoral okrug), Rossiiskaya Benoit Seim, "Lietuvos Respub (fn. 40). Lithuania: Lithuanian gazeta, January 24, 1996. Hungary: 1rs. It/rinkimai/seim96. likos Seimo rinkimo '96," dataset, online at http://rc. 46 elections under a new mixed electoral system similar InMarch 1998 Ukraine held parliamentary to Russia's, which was not included in this study for lack of data. The results show a continued preva district elections in Ukraine. Of the 225 deputies candidates in single-member lence of independent elected in plurality elections, 114 (51 percent) were independents. 47 uses the candidates as one characteristic of his of strong independent Mainwaring proliferation index of party institutionalization. 6). (fn. Mainwaring
This content downloaded on Tue, 8 Jan 2013 08:59:23 AM All use subject to JSTOR Terms and Conditions
378
WORLD POLITICS
The fact that proliferation of nonpartisan candidates did not occur cases in all the that party institutional postcommunist again suggests are ization varies in the region. While and fluid relatively weak parties a in all the emerging clear distinction democracies, postcommunist needs to be made between those states with parties that dominate the and control the choices voters face and those that process electoral role in the system plays an important develop on center attention of parties in this context. PR systems party-list
nomination
do not. The ment
as the for nomination, while dis parties only vehicles single-member to enter the contest. At the trict elections candidates allow nonpartisan same time, the effects of electoral on the number of are systems parties
mitigated by the institutionalization of the party system itself. Finally, itmust be reiterated that while plurality and majoritarian effect in some postcom systems had a significant mechanical to have no munist districts states, single-member appeared surprisingly in any of the cases under ex effect on the number of electoral parties electoral
In no case does
amination.
the effective
number
of elective
parties
in
SMD elections approach the number expected in the literature (M + 1 where M signifies the number of candidates winning election). Of are more
elections
course, majoritarian
in this sense for they
ambiguous
provide more incentives for party proliferation in the first round. For majoritarian
mains M vance
cases,
to the
second
the expectation
however,
+ 1with M
of reductive
influence
re
signifying the number of candidates that can ad For Lithuania
M
+ 1 would
equal three is between effective candidates the two top per district since the run-off an vote getters, not come close to Lithuania does expectation approxi allows the top three finishers mating. Hungary plus any candidate with round.48
15 percent of the vote into the second round. Thus, only in Hungary can the to a of attributed number effective candidates be high permis sive electoral of compensatory
since Hungary tier has a third national system, especially seats that uses votes not used to win seats from both
the single-member district tier and the territorial PR tier.This situation provides candidate
chief
incentive
no chance
with
can be used The
for party proliferation because a vote for a to the next round is not wasted; it of getting seats in the national tier. for compensatory
even more
consequence
of the failure
of voters
and parties
to act
strategically is a high level of disproportionality in those countries with relatively developed dent candidates. Up 48 Cox
of the proliferation parties that constrain indepen to now, most elites (with the party postcommunist
(fn. 5), 123.
This content downloaded on Tue, 8 Jan 2013 08:59:23 AM All use subject to JSTOR Terms and Conditions
ELECTORAL SYSTEMS/NUMBEROF PARTIES
379
exception of Poland) have failed to anticipate being punished by the mechanical effect of effective thresholds of single-member districts. to fail to achieve representa thus have entered competition They only voters have tended not to tion. Likewise, postcommunist anticipate or do not care. on that their vote will be wasted marginal parties cases
all the postcommunist
Thus,
employing
dis
single-member
tricts examined here provide some surprising and intriguing findings. In Ukraine
and Russia,
single-member
districts
produce
party prolifer
ation defying both the expectation of strategic voting and mechanical effects due to disproportionality. In Lithuania andHungary, large par ties benefit
from
greatly
the mechanical
effect
of single-member
dis
tricts due to high disproportionality, but voters and elites did not seem or able to react to these
willing
incentives
and behave
strategically.
The big question for both PR and single-member district electoral systems iswhether the lack of strategic voting found in this study is a or elections. Will there permanent transitory feature of postcommunist over time as be greater party consolidation hold and takes democracy to elites and voters learn the rules and better adapt the incentives pro
vided by the electoral system? Perhaps time and repeated electoral cy cles are all that found
for electoral
so to the types of outcomes regularly produce democracies. Cox has argued that time is required to take effect in new democracies:
is needed
in established systems
The typical scenario in emerging democracies, whereby a great number of par ties spring up in the first elections, and there is a relatively slowwinnowing out pro cess, makes sense. A large number spring up in the first election because it is not clearwho will be viable andwho not. As information is revealed about voter pref erences,
the more
serious
groups
will
continue
to enter,
even
against
poor
short
term odds, in the hopes of convincing less committed competitors to drop out.49 While
there have
not yet been
enough
elections
to make
any defini
tive judgments about learning, these initial findings suggest some in cases and between In differences electoral among systems. teresting both PR and single-member with district the countries the elections, more showed institutionalized the strongest party systems signs of over time. In PR elections, saw their ef Poland and learning Hungary over fective number of electoral and disproportionality parties drop rose the number of parties and disproportionality time, while sharply in to the second. In and Lithuania Russia from the first election single member district elections, showed signs of party consol only Hungary
idation over time, and Russia (the only other country with full data for 49
Ibid., 159.
This content downloaded on Tue, 8 Jan 2013 08:59:23 AM All use subject to JSTOR Terms and Conditions
WORLD POLITICS
380 more
saw fractionalization increase over time. In election) PR and the district elections, under single-member comparing learning in PR elections was, contrary evidence of learning of strategic behavior to evidence greater than in SMD elections. This suggests expectations, one
than
states PR elections in postcommunist in controlling effective the number
that more
the dramatic
Given
with legal thresholds may be of parties than SMD elections. and effects produced by effective
mechanical
legal thresholds in both PR and single-member
in
district elections
states, one could expect the winnowing-out process de postcommunist as as are some to scribed by Cox give long developed enough parties in of party preferences. semblance Voters, then, can behave strategically are weakest can not even of and those Where parties light preferences. the nomination
control
process,
however,
such as in Russia
and Ukraine, district elec
in single-member less likely, particularly learning becomes are allowed to candidates tions where proliferate. independent
System Effects
Electoral The
and elites.
New Democracies
systems in postcom suggest that electoral a to of level led strategic behavior by significant cases the number of effective In most electoral parties
findings presented states have not munist voters
in Other
here
to ex the carrying of the electoral system leading capacity a trend to post Is this of levels tremely high peculiar disproportionality. in all new in initial elections communist states, or can it be found
exceeded
democracies?
of electoral
the proliferation
Perhaps
parties
in postcom
munist states is a normal condition of initial elections thatwill subside as increased out
information
should tion
similar
show
in the number most
and experience If so, comparisons
ones.
nonviable
party proliferation of electoral parties
reward viable other
with
in initial over
parties and weed new democracies
elections
and contrac
time.
the electoral sys comparison use tems of the states and new democracies, which plu postcommunist show whether the most powerful would rality systems. The comparison The
would
effective
be between
electoral system has faced party proliferation in initial elections in other new democracies elites
learn
mocratizers
to
only
to
consolidation produce the system. Unfortunately,
navigate have adopted
plurality
over
time as voters few
systems.50 Therefore,
50 Eleven
third-wave the most
and de in
elections in the 1980s adopted plurality systems initial democratic countries experiencing islands. Such cases do not provide the most suitable very small states, mostiy Caribbean states. Nevertheless, these new democracies did not experience comparison with large postcommunist states in the study. The exhibited by the postcommunist the same high level of party proliferation
but all were
This content downloaded on Tue, 8 Jan 2013 08:59:23 AM All use subject to JSTOR Terms and Conditions
ELECTORAL SYSTEMS/NUMBEROF PARTIES structive
is a historical
comparison democracies
new
older, Two postwar
that have
one with now
of
elections
initial
consolidated.
become
relatively Federal Republic,
the German
democracies,
the
381
its
with
mixed electoral system employing both plurality elections and PR, and India, with its pure plurality system, will be examined to shed some light on the degree of party proliferation in initial elections under a plu on the system rality system and sult of learning.
s reductive
over
properties
time as a re
support for the history provides a common in new idea that party proliferation may be phenomenon do but that reductive effects of the electoral take democracies system Postwar
Germany's
over
hold
time.
electoral
early
Scholars
found
have
of strategic voting totals for parties between
under
evidence
in vote
the PR rules in the difference plurality two tier. The district tier and the single-member par largest German more votes in the the CDU/CSU and the have SPD, ties, single gained
member districts than in the PR racewhile lost votes. This
situation
suggests
smaller parties like the FDP
that voters
from
defected
small par
ties less likely to win in single-member districts and gravitated toward in the discrepancy for representation. While a tiers is not that significant, averaging gain for the three parties most 2 percent affected
parties with better chances the vote between the two or
loss
of
less
than
evidence of strate (CDU/CSU, SPD, and FDP), it does provide persuasive the vote does not really in rules even when gic voting under plurality of seats among parties.51 the final distribution fluence This
did not occur
phenomenon
in the first election
under
the mixed
system in 1953. In that election the SMD tier produced over three effec tive electoral
parties
a
(3.38),
higher
effective
of electoral
number
par
ties than in the corresponding PR tier (3.31). By the next election the in the single-member parties produced to 2.75, tier pro district and the single-member duced fewer effective electoral parties than the PR tier (2.78), as the lit erature would This trend has grown over the years reducing expect.52 in the single-member dis the chances for victory by marginal parties effective
number
districts
had
of electoral
fallen
tricts to virtually
nothing
and solidifying
a
two-party
system
in the SMD
tier. Since the third election held under the mixed system in 1961, no party
other
than
the CDU/CSU
average effective number of electoral parties in Russia's lower than the average produced Cox (fn. 5), 309-11. 51Jesse(fn.26),112. 52 of effective Calculations
number
or SPD has won
a
single-member
district
is 2.26, which ismuch for this group of new democracies cases see two plurality elections. For data regarding these
of parties were
based on data in Barnes
This content downloaded on Tue, 8 Jan 2013 08:59:23 AM All use subject to JSTOR Terms and Conditions
et al. (fn. 26), 906.
382
WORLD POLITICS
seat. Thus,
Germany
a case of
provides
increased
consolidation,
pre
sumably due to learning, which began to take place in the second elec tion. German
voters
and elites did not need much
time
to
adjust
to the
incentives
of the electoral system. an India provides of the opposite example phenomenon?a ent absence of a system. Party strategic voting under plurality
persist prolifer
ation at the district level has remained high in India despite the fact that the plurality system consistently produced very high dispropor tionality,
severely
penalizing
marginal
parties.
As
a consequence,
until
the late 1980s, India experienced a dominant party system inwhich the Congress Party maintained votes while the opposition no evidence of consolidation first
two
elections,
a of seats based on a minority of majority was is there Unlike Germany fragmented. over time as a result of In India's learning.
the effective
number
of electoral
parties
stood
at
4.21 despite the fact that the high disproportionality of the plurality system
narrowed
that number
down
to less than
two effective
parlia expect voters and elites to against small parties, the high level of adjust not subside over several decades of did fractionalization party repeated In six elections from 1962 to 1984, India actually democratic elections. had more effective electoral parties, an average of 4.31, even though the
mentary
one would
(1.79). Although parties to the severe disincentives
electoral
system's mechanical
effect
repeatedly
reduced
the number
of
effective parties entering parliament to 221P By the end of the 1980s the mechanical effect of India's plurality electoral system even failed as the Congress majorities Congress party lost power and the manufactured In 1996 thirty decades for gave way to coalition governments. enjoyed were in parliament with 5.88 effective parliamentary parties represented a similar outcome 1998 parliamentary elections The produced parties. similar to the frac strikingly district elections.54 tionalization single-member two in initial elections These show that party proliferation examples a a common occurrence. while be However, may Germany provides with
5.29
effective
parliamentary produced by Russia's
parties,
case of learning that produced strategic voting beginning in the second 53 (fn. 1), 161. 54Lijphart elections were based on seat distribu Calculations for the 1996 and 1998 Indian parliamentary indiavotes. com. One crucial difference be tions provided by the India Votes '98website, http://www. tween the Indian and Russian cases is that the Indian case tended to produce two- or three-candidate races at the district level. The multiparty system at the national level was produced by the fact that the two in each district did not belong to the same two major parties from district to dis major candidates saw an average of seven significant candidates compete in trict. This was not the case in Russia, which in Indian elections see districts. For district-level analysis of the number of candidates single-member and the Number of Parties in India and and Ken W. Kollman, "Party Aggregation Pradeep Chhibber the United States," American Political Science Review 92 (1998), 332.
This content downloaded on Tue, 8 Jan 2013 08:59:23 AM All use subject to JSTOR Terms and Conditions
383
ELECTORAL SYSTEMS/NUMBEROF PARTIES
peated
that the passage of time and re to promote sufficient strategic behavior.
the Indian
election,
elections
shows
experience alone are not
As in the postcommunist cases, Iwould argue that the key determining factor
is the
institutionalization
tivelywell-established strategic
voting
while
India failed to do so.
of the party
system,
in which
a rela
party system inGermany provided the basis for a more
weakly
institutionalized
party
in
system
Conclusions The findings presented here suggest that electoral systems affect the states and in more established of parties in postcommunist de in very different ways. These findings provide greater empir ical support for Cox s theoretical work on the conditions for necessary a to take Without institutionalized strategic voting place. relatively party system, voters and elites may not have the ordered party prefer ences to behave or may not have to information strategically enough differentiate viable from nonviable contenders. Under such conditions one would not expect the reductive of elec impact powerful plurality tions to take effect. Cox anecdotal evidence from only provided Papua number
mocracies
to illustrate how such conditions in the realworld may produce party proliferation under plurality rules.55The findings pre
New Guinea
states may suggest that many postcommunist under such circumstances. transition ing democratic
sented
here
be undergo
This study has reemphasized the importance of party institutional ization
as an
variable intervening systems and the number
electoral
the relationship influencing of parties. The constraining
between effects
of
thresholds seemed to vary with the institutionaliza legal and effective states. Countries tion of the party system in postcommunist like Poland and Hungary with more developed showed party systems greater ten to electoral system incentives dencies for strategic adaptation and learn
ing over time while
countries like Russia and Ukraine with weakly
no or systems showed signs of strategic behavior an eventual not guarantee learning. The passage of time will necessarily to electoral in found consolidated systems along patterns adaptation institutionalized
party
never a institution relatively produces as case alized party system the of India powerfully demonstrates. Under conditions of extreme party underdevelopment, the electoral use that the of system party labels?proportional promotes representa
democracies
55 Cox
if the social context
(fn. 5), 85.
This content downloaded on Tue, 8 Jan 2013 08:59:23 AM All use subject to JSTOR Terms and Conditions
WORLD POLITICS
384 be more
tion?may
effective
in constraining
the number
of parties
than
the plurality system, provided a legal threshold is used. Indeed, inRus sia a PR system with effect than
a 5-percent its plurality time PR elections with legal
legal
chanical over
in a
threshold
had
Poland
counterpart. can reduce thresholds
state as elites
and voters
a has
stronger shown
me that
the number
of
learn the incentives
parties postcommunist of the system. Our cases have shown no such dramatic reduction over time in of electoral number district single-member parties
in the elec
to curb party fractionaliza systems may find better results with
tions. Therefore, electoral engineers tion in states with developing party
hoping
PR and a legal threshold than with single-member district elections. states
Postcommunist
provide that defies
a
to electoral studies with challenge even the electoral systems. strongest such as those found in Russia systems,
party fractionalization The most party underdeveloped such fragmentation and Ukraine, produce fect of single-member district elections nomena
suggest
that
in integrating
that even does
not
the mechanical
take hold.
postcommunist
states
ef
Such phe into the field
of electoral studies greater attention will need to be paid to the impli cations puzzle which
of party systems. The main under is the process and conditions this analysis states. in democratizing become institutionalized
of low levels of institutionalization arising party
from systems
This content downloaded on Tue, 8 Jan 2013 08:59:23 AM All use subject to JSTOR Terms and Conditions