Historical Legacies and Post-Communist
Regime Change
Grigore Pop-Eleches Princeton University
This article shows that post-communist regime trajectories have been largely circumscribed by historical legacy
differences, but the question about which particular legacy matters most is much harder to answer, since statistical
results are sensitive to model speciﬁcation and to the choice of democracy indicator. While some of these discrepancies
reﬂect the inherent limitations of traditional statistical methods, others reﬂect the different dimensions of democracy
captured by different indicators. Therefore, the article contributes to a more nuanced explanation of post-communist
democratization by showing that different legacies drive different aspects of democratization. Finally, the results
demonstrate that several prominent alternative explanations—initial election outcomes, institutional choices, geographic
diffusion, and external conditionality—played a relatively modest role in explaining democratization
patterns beyond the constraints imposed by historical legacies.
T
he collapse of communism in 1989–91 represented
a unique opportunity for the countries
of the former Soviet bloc to join the global
democratization wave in what was originally envisioned
as a common transition from communist oneparty
regimes to liberal democracies. Instead, the
political developments of the last 17 years have produced
a variety of divergent political trajectories with
endpoints ranging from authoritarian regimes in
Belarus and Central Asia to relatively well-functioning
democracies in East-Central Europe. What accounts
for this dramatic divergence among the erstwhile communist
comrades?
To answer this question, this article analyzes the
stark cross-country historical legacy differences at the
outset of the transition and demonstrates the powerful
and temporally resilient inﬂuence of these overlapping
social, economic, and political legacies on postcommunist
regime trajectories. While the joint effect
of legacies has greatly circumscribed post-communist
democratic prospects, the question of which particular
type of historical inheritance matters most is much
harder to answer with any degree of conﬁdence. In
part, these difﬁculties reﬂect the limitations of the
standard methods through which the legacy-reform
link has been analyzed so far and this article suggests a
few alternative approaches to this question.
However, the article also shows that some of the
statistical discrepancies in the legacy-democracy link
are due to the fact that different regime indicators
capture distinctive aspects of democracy. This point
suggests a more nuanced explanation of how legacies
shape regime trajectories: thus, some legacies (such as
membership in the pre-war Soviet Union and Muslim
religious majorities) primarily affect the postcommunist
institutional conﬁgurations, while other
legacies (e.g., ethnic fragmentation and Western
Christianity) are more useful in explaining the gap
between political institutions and actual rights or
between basic rights and genuine democratic governance.
The ﬁnal section demonstrates that several
prominent alternative explanations—initial election
outcomes, institutional choices, geographic diffusion,
and external conditionality—played a relatively
modest role in explaining democratization patterns
beyond the constraints imposed by historical legacies.
The existing academic literature provides few clear
answers about the role of legacies in post-communist
democratization. Even though Jowitt (1992) predicted
that Leninist legacies would decisively shape postcommunist
trajectories and Janos (1994, 2000) argued
that pre-communist cross-country differences would
continue to be salient despite ﬁve decades of communist
regional equalization attempts, much of the early
The Journal of Politics, Vol. 69, No. 4, November 2007, pp. 908–926
© 2007 Southern Political Science Association ISSN 0022-3816
908
academic literature on the subject downplayed the role
of initial conditions. In the early 1990s, the predominant
theoretical approach to the study of democratization
focused primarily on proximate explanations,
such as elite politics and democratic crafting (Di
Palma 1990; Huntington 1991; Karl and Schmitter
1991; Linz and Valenzuela 1994; O’Donnell, Schmitter,
and Whitehead 1986; Przeworski 1991). A second
group of studies focused on the initial postcommunist
power balance (Fish 1998a, 1998b; McFaul
2002; Roeder 1994) or on the effects of institutional
choices such as presidential versus parliamentary
systems (Ishiyama and Velten 1998). A third group of
explanations analyzed the international dimension of
post-communist regime change in the broader context
of the third wave of democratization (Diamond and
Plattner 1996; Grugel 1999) and highlighted the role
of geographic diffusion (Kopstein and Reilly 2000)
and European integration incentives (Kurtz and
Barnes 2002; Pridham and Ágh 2001; Vachudova
2004; Whitehead 1996).
While these explanations tend to downplay the
role of historical legacies, the past has not been completely
neglected by scholars of post-communism.
Some of the above authors acknowledge the inﬂuence
of certain legacies on regime change (Fish 1998b;
McFaul 2002), but they only consider a small set of
legacy indicators, which are basically treated as control
variables. Other studies focus more directly on the
effects of speciﬁc historical legacies on postcommunist
democratization (Bunce 2005; Kurtz and
Barnes 2002; Roeder 1999), political party development
(Grzymala-Busse 2002; Ishiyama 1997; Kitschelt
et al. 1999; Pop-Eleches 1999), initial election outcomes
(Darden and Grzymala-Busse 2005), and
voting behavior (Wittenberg 2006). However, these
studies emphasize a relatively small number of legacies
at the expense of a more comprehensive discussion of
the implications of the region’s multiple and overlapping
legacy differences.
More recently, a few studies have tried to tackle the
empirical and theoretical complexity of legacy effects.
Kitschelt (2003) provides an important step towards a
more integrated theoretical understanding of the
post-communist legacy-regime link, but then focuses
almost exclusively on bureaucratic legacies in his
empirical analysis. Horowitz (2003) captures the
effects of cultural legacies and agricultural employment
on post-communist regime change but his decision
to combine a number of pre-communist and
communist economic and political legacies into a
“frustrated national ideals” index makes it difﬁcult to
disentangle the contribution of individual legacies.
Crawford and Lijphart (1997) discuss six broad legacy
types affecting post-communist politics, but their
primary emphasis is on how transition countries differ
from other regions rather than on legacy differences
between ex-communist countries. De Melo et al.
(2001) and Katchanovski (2000) survey a large
number of legacy indicators but their primary focus is
on post-communist growth variation, and they
employ factor analysis to extract legacy indices
rather than focusing on individual indicators. Ekiert’s
(2003) analysis of post-communist transformations in
Central and Eastern Europe also emphasizes the
importance of historical legacies, but he focuses primarily
on Communist-era institutional legacies and
argues that their effect declines over time. Not surprisingly,
Kopstein (2003) concludes his review of the literature
on the relationship between democracy and
legacies with a call for a more nuanced treatment of
the mixed communist legacy and for a more careful
analysis of the interplay between domestic and international
legacies.
The present article argues that historical legacies
have to constitute the starting point for any systematic
analysis of democratization in the postcommunist
context. Obviously, this argument does
not imply that regime trajectories were predetermined
by initial structural conditions; indeed, the
analysis shows that no single legacy was either sufﬁcient
or necessary for post-communist democracy.
Nor were the political preferences and choices of
post-communist elites and citizens irrelevant.
However, I claim that the stark cross-national differences
in the region’s overlapping cultural, socioeconomic,
and institutional legacies signiﬁcantly shaped
the preferences of political actors and the constraints
on their choices. As a consequence, the prospects for
democratization and democratic deepening were signiﬁcantly
better in countries with favorable legacies
(such as the relatively developed, ethnically homogenous
countries of East Central Europe with their
longer histories of statehood, democracy, and
bureaucratic competence) than in many of the ﬂedgling
new states emerging from the former Soviet
Union and Yugoslavia.
This article contributes to the post-communist
democratization literature in a number of ways. First,
it covers a broader range of legacy and democracy
indicators than previous statistical analyses on the
subject and illustrates both the importance and the
complexity of the relationship between structural
conditions and democratization. Second, the article
demonstrates the importance of taking historical
legacies seriously, not only because their inﬂuence on
historical legacies and post-communist regime change 909
post-communist democratization has been large and
remarkably resilient, but also because their systematic
incorporation in statistical analyses of democracy
undermines some of the earlier claims about the
importance of more contingent factors, such as
initial elections outcomes, institutional choices, and
geographic diffusion. Third, unlike most previous
work, this analysis addresses the methodologically
crucial—and frequently ignored—issue of the
overlap between cultural, economic, institutional,
and social legacies in the post-communist context
and suggests possible solutions for addressing
these problems. Finally, this paper shows that the
importance of legacies varies as a function of the
particular dimension of democracy captured by different
democracy indicators, and therefore calls for a
more nuanced approach to the statistical study of
democratization in the post-communist context and
beyond.
The article starts with an overview of legacies,
followed by a discussion of democracy scores from
several sources. The second section presents crosscountry
statistical evidence to show that initial conditions
heavily inﬂuence regime trajectories and that
their effect does not diminish (but actually increases)
over the course of the transition. The third section
deals with the methodological difﬁculties inherent in
the study of the post-communist legacy-reform link
and identiﬁes some solutions and a future research
agenda for overcoming these difﬁculties. The ﬁnal
section shows that alternative explanations such as
institutional choices, initial election outcomes, geographic
diffusion, and Western integration provide
only limited explanatory leverage beyond the combined
effect of legacies. The conclusion summarizes
the ﬁndings and discusses their implications for
democracy in the region.
Legacies and Democracy—Concepts
and Measures
For the purpose of this analysis, legacies are deﬁned as
the structural, cultural, and institutional starting
points of ex-communist countries at the outset of the
transition. Even though these transitional starting
points have deep and complicated historical roots in
the region’s pre-communist and communist past, this
article does not attempt to retrace these roots. Instead,
this section sets out to capture a statistical snapshot of
the complex differences in initial conditions in the 28
countries of Eastern Europe and Eurasia, which abandoned
Communism as an ofﬁcial doctrine between
late 1989 and late 1991.1
This overview is not simply a
descriptive exercise but rather a theory-driven attempt
to systematize those aspects of the region’s historical
baggage which can be expected to have affected the
democratic prospects of the former communist countries.
In doing so, I focus on both structural longue
durée factors and more recent institutional features
rooted in the communist period (Ekiert and Hanson
2003).
To help organize the broad array of legacies, the
discussion distinguishes between ﬁve key legacy
dimensions for ex-communist countries. A detailed
description of the variables can be found in Table 1.
Several recent studies have emphasized the importance
of geography in explaining political development
(Bloom and Sachs 1998; La Porta et al. 1999;
Acemoglu, Johnson, and Robinson 2001). In the East
European context, the most important aspect of a
country’s geographic location is its proximity to
Western Europe, measured by two indicators: the
existence of a border with an EU-member country and
the distance from the closest EU capital. Western proximity
should promote democratization by facilitating
the diffusion of Western values and political institutions
to ex-communist countries (Kopstein and Reilly
2000). Moreover, proximity to Western Europe should
be beneﬁcial to democracy by providing more credible
European integration prospects and hence stronger
democratization incentives to the countries on the
Western fringe of the former Communist bloc.2
The importance of cultural/religious heritage in
explaining political development has received a fair
amount of attention in the East European context,3
as
well as in more recent debates about the compatibility
of Islam and democracy (Fish 2002; Stepan and
Robertson 2003). Two straightforward measures in the
1
This case universe, which is used by most statistical studies of
post-communist reforms, consists of 12 East European countries
(including the ﬁve initial Yugoslav successor states but not Kosovo
and Montenegro) and all 15 former Soviet republics and Mongolia.
I have decided against including other former communist
countries, since China, Cuba, North Korea, and Vietnam are not
technically post-communist yet, whereas other Marxist regimes in
the developing world were of a sufﬁciently different nature so as to
make their inclusion problematic.
2
In a strict sense geographic location is not a historical legacy.
However, it is a structural factor with signiﬁcant implications for
democracy prospects and is therefore treated as a legacy in this
analysis. Moreover, while geographic location is time invariant, the
political implications of location are certainly time dependent.
3
Janos (1989) discusses the different foundations of political
authority (legal-rational in Western Christianity versus traditional
in Eastern Orthodoxy), and Lal (1998) points to the higher degree
of separation between church and state in Western Christianity.
910 grigore pop-eleches
post-communist context are whether the predominant
religion in a given country is Western Christianity
or Islam, thus leaving Eastern Orthodoxy as the
excluded category.4
Additionally, I consider imperial
legacy, since it is likely that centuries of different imperial
inﬂuence affected not only religion but also institutions,
civic values, and national ideas (Bunce 2005)
differently in countries in the Central European sphere
of inﬂuence compared to regions dominated by the
Russian or the Ottoman Empires.
Given the painful process of post-communist economic
adjustment, economic legacies are likely to
impose signiﬁcant constraints not only on economic
reforms but also on democratization. Because the
existing literature offers little help in this respect,5
I use
an indicator of the energy intensity of the communist
economies to capture Soviet-style structural economic
4
The only exception is Mongolia, where Buddhism is the primary
religion.
5
Kitschelt (2001) uses the FSU dummy as a proxy for economic
distortions but such a measure glosses over important differences
between former Soviet republics, as well as between East European
countries.
TABLE 1 Overview of Variables
Variable name Coding/measurement Source(s)
EU border 1 = yes, 0 = no Author
Distance to West Log (Distance to Helsinki, Vienna, Rome or
Berlin in km)
Author
Western Christian Western Christian majority? 1 = yes, 0 = no Author
Muslim Muslim majority? 1 = yes, 0 = no Author
Minority share Log (percentage ethnic minoritiesa
) Author using data from CIA
World Factbook (1992)
Energy intensity (GDP per unit of energy use)-1
/ (%
industry/total GDP)
Author using data from World
Development Indicators (2001)
Natural resources 2 = resource rich, 1 = moderate resources,
0 = resource poor
de Melo et al. (2001)
Non-CMEA Exports Non-CMEA exports/Total exports de Melo et al. (2001)
EBRD Economic Reform
score 1989
1 = none—4 = highest EBRD (2001)
GDP/capita 1989 GDP/capita in 1989 at PPP de Melo et al. (2001)
% Urban 1989 urban population/total (in %) World Development
Indicators (2001)
% High Education 1989 % higher education enrollment in 19–24 age
group
UNICEF (2001)
Communist bureaucratic legacy bureaucratic-authoritarian = 3
national-accommodative = 2 mixed = 1.5
patrimonial = 1 patrimonial/colonial
periphery = 0
Kitschelt (2001)
Interwar statehood 1 = yes, 0 = no Author
Prewar Soviet Republic 1 = yes, 0 = no Author
Prior democracy Average Polity Regime score 1920–39 Author calculations based
on Polity
Imperial legacy 2 = Western, 1 = mixed 0 = Russian/Ottoman Author
FH political and civil rights 0 (least free) to 12(most free)b
Freedom House (2005a)
NIT democratic reform 0 (lowest) to 6 (highest)c
Freedom House (2005b)
WB Voice and Accountability (VOA) -2.5 (lowest) to 2.5 (highest) Kaufman et al. (2005)
Polity IV Regime -10 (autocracy) to 10(democracy) Marshall & Jaggers (2005)
a. Does not include the Roma population both because ofﬁcial censuses are biased and because Roma have not been organized politically
along ethnic lines to the extent of other ethnic minorities.
b. Obtained by adding the scores for political and civil liberties, and then subtracting the sum from 14.
c. The scores were inverted so that higher scores indicate better performance. Also, given the uneven temporal spacing of the surveys, I
created yearly scores by using weighted averages where necessary.
historical legacies and post-communist regime change 911
distortions.6
Since prior work has identiﬁed a “natural
resource curse” effect on democracy (Barro 1999), the
analysis also includes an indicator of natural resource
endowment7
to address the possibility that postcommunist
regime transformations were affected by
the political repercussions of natural resource wealth.
Since closer economic integration with the West prior
to the collapse of communism should make a country
more receptive to Western democratic conditionality, I
include a measure of the exports to non-CMEA countries.
Finally, to capture the possibility that postcommunist
democratization is affected by differences
in pre-1989 economic reforms, I used the EBRD economic
reform index for 1989.
Despite intense communist efforts to erase large
pre-communist differences in social conditions/
modernization, transition countries nevertheless differed
in terms of socioeconomic development levels,
which, according to modernization theorists,8
should
have predicted differential readiness for democracy.
The present analysis focuses on GDP/capita, urbanization
and education levels,9
all of which reﬂect the large
developmental gap between East-Central European
countries on one hand, and parts of the Balkans and
the former Soviet Union on the other. Despite the
obvious theoretical shortcomings of the classical
modernization hypothesis, its predictions receive
individual-level support from public opinion surveys,
which conﬁrm that urban and educated voters are
generally more exposed to Western inﬂuences and,
therefore, more supportive of democracy (Clem and
Craumer 1997).
Despite their common history of de facto oneparty
rule, the transition countries exhibited considerable
institutional legacy differences, traceable not
only to late-communist reforms but also to older historical
legacies. Following Kitschelt (2001), I distinguish
between ﬁve types of bureaucratic legacies,
ranging from bureaucratic-authoritarian (the
Czech Republic) to patrimonial/colonial periphery
(Caucasus and Central Asia). The institutional penetration
of Communism is measured by two straightforward
if somewhat blunt measures—years under
communism and prewar membership in the Soviet
Union—with longer communist spells presumably
complicating democratization. Along similar lines,
one would expect pre-communist democratic experience
in the interwar period10
to facilitate democratization
by allowing for at least some voters with
memories of free elections and by strengthening anticommunist
forces in cases where prewar democratic
parties were revived following the collapse of Communism
(e.g., Romania, Hungary, Czech Republic).
The last two variables in this category address even
more fundamental political challenges: the limited
prior statehood experience of many transition coun-
tries,11
and the challenges of high ethnic fragmentation,
which has signiﬁcantly complicated democratization
efforts in large parts of the Balkans and the former
Soviet Union (Roeder 1999).
Dependent Variables—Democracy Scores
Of the large and growing variety of democracy measures,
this paper focuses on four of the most widely
used indicators for the ex-communist countries,
described in Table 1. Even though much of the democratization
literature (as well as studies using democracy
as an independent variable) tend to use only one
of these measures, thereby implicitly assuming them
to be equivalent measures of the same concept, the
current discussion suggests that their different conceptualization
approaches actually produce different
empirical results and therefore warrant closer scrutiny
than they usually receive. Freedom House (FH) political
and civil rights scores are the most widely used democracy
measure in cross-national research and have the
beneﬁt of the most extensive temporal and geographical
coverage for my sample. Compared to the narrower
rights-based FH democracy measure, Nations in
Transit (NIT) democratization scores attempt to
capture the quality of democracy by evaluating countries
along four dimensions since 1996: the competitiveness
of the political process, the degree of
development of civil society, the existence of an inde-
6
The measure uses the inverse of GDP per unit of commercial
energy use, normalized for the share of industrial output in total
GDP (in order to avoid the risk of scoring a country with a large
but relatively energy-efﬁcient diversiﬁed industry the same as a
country with a smaller but heavily distorted industrial sector.)
7
The measure, based on de Melo et al. (2001), codes countries on
the basis of the abundance of natural resources, rather than on the
share of natural resources in exports. (The latter measure runs the
risk of endogeneity since it is affected by institutional features and
political decisions).
8
See Lipset (1959) but also the ﬁndings in Przeworski et al. (2000).
9
While Communism eroded the large pre-communist literacy differences,
enrollment in higher education in 1989—the measure
used in this paper—still differed signiﬁcantly across countries.
10
The linkage between interwar and post-communist democracy
has been explored before (e.g., Wittenberg 2006) but it has not
been tested in cross-country statistical tests.
11
Indeed, the settlement of statehood and border issues was the
one prerequisite for democratization acknowledged even by
transitologists.
912 grigore pop-eleches
pendent media, and the quality of governance and
public administration. The third measure, Kaufman,
Kraay, and Mastruzzi’s (2005) Voice and Accountability
(VOA) score, aggregates democracy indicators from as
many as 11 different surveys through an unobserved
components methodology. A fourth widely used
democracy indicator is the Polity IV Regime score,
measured as the difference between its democracy and
its autocracy scores, each of which are based on the
weighted aggregation of three components: competitiveness
of political participation, openness and competitiveness
of executive recruitment, and constraints
on the chief executive’s power.
The Persistent Link between
Legacies and Post-Communist
Democracy
Despite their common past as one-party states and
their almost simultaneous abandonment of communist
ideology, the transition countries did not experience
the kind of uniform regime trajectory predicted
by the transitologists’ tabula rasa vision of democratization.
Instead, the high and statistically signiﬁcant
correlations between legacies and democracy scores in
Table 2 conﬁrm that post-communist democratization
was deeply imbedded in historically rooted crosscountry
differences. Moreover, judging by the overtime
change in the correlations between legacies and
democracy, modernization is the only legacy cluster
whose effect weakened over the course of the transition.
For other legacy dimensions—especially cultural
and institutional factors—the cross-country explanatory
power of legacies increased over time across different
democracy measures.
Table 2 reveals important differences in the correlation
between democratization and different legacy
types. Three highly correlated legacy clusters—
geography, religious-cultural, and institutional—
appear to be the strongest correlates of political
trajectories. The two religious variables emerge as
powerful predictors in the hypothesized direction,
suggesting that predominantly Western Christian
countries were more democratic than average,
whereas Muslim countries were more likely to be
authoritarian. Contrary to some earlier claims (Fish
1998a; Kitschelt 2003), ethnic diversity displays a
TABLE 2 Bivariate Pairwise Correlations between Legacies and Reform Scores
Geography Culture/Religion Economy Modernization Institutions/Statehood
EUborder
DistanceW
WChristian
Muslim
Imperiallegacy
Energyintensity
Naturalresources
EBRDReforms1989
Wexports1989
GDP/cap1989
%Urban1989
%HighEduc1989
Commburlegacy
InterwarStatehood
PrewarSovietRepublic
Priordemoc
Minorityshare
FH Democracy 93 .51 -.65 .62 -.64 .46 -.61 -.37 .07 .36 .61 .48 .27 .75 .57 -.58 .46 -.61
FH Democracy 96 .49 -.69 .68 -.66 .54 -.57 -.31 .05 .33 .57 .50 .29 .78 .61 -.66 .59 -.51
FH Democracy 00 .44 -.73 .76 -.73 .65 -.66 -.45 .21 .44 .50 .51 .26 .82 .57 -.81 .70 -.49
FH Democracy 04 .41 -.74 .71 -.64 .66 -.72 -.50 .33 .56 .38 .41 .23 .82 .56 -.86 .68 -.48
NIT Democracy 96 .59 -.78 .80 -.66 .60 -.60 -.37 .17 .33 .58 .63 .33 .84 .57 -.68 .68 -.53
NIT Democracy 00 .51 -.73 .80 -.66 .64 -.67 -.43 .19 .42 .48 .51 .25 .84 .59 -.78 .70 -.52
NIT Democracy 04 .46 -.82 .78 -.60 .68 -.68 -.52 .33 .53 .38 .48 .25 .86 .58 -.85 .73 -.47
WB VOA 96 .65 -.71 .68 -.47 .69 -.53 -.32 .21 .54 .28 .40 .06 .72 .44 -.70 .55 -.68
WB VOA 00 .55 -.86 .86 -.52 .71 -.54 -.47 .27 .37 .41 .62 .33 .84 .47 -.64 .67 -.50
WB VOA 04 .54 -.88 .87 -.60 .74 -.61 -.51 .32 .43 .46 .66 .38 .86 .49 -.74 .73 -.47
Polity Democracy 93 .47 -.47 .44 -.72 .38 -.52 -.39 -.03 .28 .62 .47 .30 .65 .48 -.53 .41 -.39
Polity Democracy 96 .49 -.45 .50 -.65 .45 -.52 -.24 .03 .40 .48 .53 .22 .70 .54 -.65 .52 -.35
Polity Democracy 00 .50 -.55 .49 -.78 .47 -.73 -.46 .32 .59 .43 .53 .23 .70 .47 -.78 .49 -.47
Polity Democracy 03 .45 -.54 .52 -.78 .50 -.72 -.44 .30 .58 .44 .55 .25 .71 .45 -.75 .47 -.49
Note: Bolded coefﬁcients are signiﬁcant at .05 (two-tailed).
historical legacies and post-communist regime change 913
strong negative correlation with democracy, particularly
with respect to FH civil and political rights in the
early transition period. While it is hardly surprising
that state- and nation-building imperatives initially
delayed democratization, the data in Table 2 suggest
that the negative effects of weak statehood and democratic
traditions persist over time, a persistence that
runs counter to the transitologists’ “learning-bydoing”
expectation. The two remaining indicators in
the“institutional legacy”cluster—bureaucratic legacies
and imperial legacies—also emerge as consistently
strong statistical correlates of democracy.
The high predictive power of the strongest economic
legacy predictor—energy intensiveness—
conﬁrms the deleterious effects of uncompetitive,
energy-intensive industries on not only economic
reforms but also democratization. This correlation is
consistent with the strong support for reform opponents
in regions dominated by Stalinist-style heavy
industry in Slovakia, Romania, Russia, and the
Ukraine and reveals a potential causal link between
energy intensiveness and democracy, in the sense that
areas with hard-to-reform industries became crucial
political constituencies for parties opposed to both
economic and political liberalization. The effects of
other economic legacy indicators were less conclusive:
thus, countries with stronger communist-era trade
links to the West and weaker natural resource dependence
made faster progress towards democracy, but
the relationship was statistically weaker than for other
indicators. Finally, the 1989 EBRD economic reform
index emerges as the weakest predictor of political
reforms, suggesting that the pre-transition economic
reform headstart of several countries (most
importantly Hungary and Poland) provides little
analytical leverage for understanding broader postcommunist
reform patterns. The ﬁnal legacy
cluster—modernization—displayed rather consistent
moderate-to-high correlations between reforms and
two of the three modernization indicators: GDP per
capita and urbanization.12
Multiple Regressions and the
Legacy-Regime Link
However, even strong correlations tell us little about
causal links, especially since different historical
legacies are highly correlated. The most widely used
statistical method for analyzing the drivers of postcommunist
democracy has been cross-sectional
regressions (Fish 1998a, 1998b; Horowitz 2003;
Katchanovski 2000; Kopstein and Reilly 2000). Nevertheless,
as Kitschelt (2003) has forcefully argued, the
primary reliance on goodness-of-ﬁt criteria unwittingly
favors proximate causes and may result in
shallow explanations. In this section, I brieﬂy discuss
what multiple cross-sectional regressions can and
cannot tell us about the relationship between legacies
and post-communist reforms.
Table 3 presents cross-sectional OLS regressions,
which illustrate the effect of legacies on different
democracy measures in 1993, 1996, and 2003/2004, to
capture regime patterns at the start, the midpoint, and
the most recent available data point of the transition.
The model speciﬁcation is primarily for illustrative
purposes and makes no claim to include all the potentially
relevant variables (which is precluded anyway
by sample size limitations). Nevertheless, the ﬁve
explanatory variables—Western Christian and Muslim,
Interwar Statehood, Energy Intensity, and Prewar Soviet
Republic—represent three of the legacy clusters discussed
earlier.13
The ﬁrst striking ﬁnding is the extremely high
joint explanatory power of the ﬁve legacy indicators,
which capture more than three-quarters of the crosscountry
variation in seven of the 10 models. Moreover,
comparing Models 3–6 to Models 7–10,
respectively, it appears that regardless of the choice of
democracy measure, legacies were stronger predictors
of regime patterns after 15 years of post-communist
transformations than in the mid-1990s, with Rsquared
statistics of around 90% for three of the four
democracy indexes. In light of these results—all of
them obtained without recourse to what Kitschelt
calls temporally proximate, “shallow” variables—it is
hard to deny that post-communist political reforms
have been largely circumscribed by the long shadow
of the past.
In terms of the effects of individual variables,
however, the statistical results in Table 3 reveal the
signiﬁcant analytical limitations of cross-sectional
OLS regressions. Thus, none of the individual legacy
indicators were consistently statistically signiﬁcant
across the different statistical models, making it difﬁcult
to claim robust links and, therefore, lay the foundation
for establishing credible causal links between
12
The prevalence of higher education turns out to be the weakest
indicator in the group, possibly due to cross-national measurement
differences.
13
The missing clusters are geography and modernization, whose
indicators are not signiﬁcant once we control for other legacies.
914 grigore pop-eleches
individual legacies and democratization on the basis
of cross-sectional regressions.14
Two inherent limitations of cross-sectional
regressions—small sample sizes15
and impracticality
for dealing with over-time regime change patterns16
—
can be addressed at least in part through the use of
panel data.17
Table 4 presents the results of PraisWinsten
regressions with panel-corrected standard
errors based on the longest available time coverage for
each of the four democracy indicators discussed
above. The basic model speciﬁcations include not only
a broader set of legacies than the cross-sectional
models in Table 3 but also a measure of the logged
duration of the transition to capture the temporal
dimension of democratization. Since the initial conditions
at the outset of the transition do not vary over
time, Models 2–4 include interaction terms between
the transition duration and key historical legacies in
order to capture the relative predictive power of legacies
as the transition unfolded.
Even though the overall explanatory power of the
panel regressions is weaker than in the cross-sectional
counterparts (since legacies do not vary temporally),
the results in Models 1 and 6–8 conﬁrm the statistical
and substantive importance of a wide range of historical
legacies for post-communist democratization.
According to Model 1, even controlling for differences
in socioeconomic development and ethnic diversity,
Western Christian countries with a legacy of independent
statehood in the interwar period (such as Poland,
Hungary, or the Baltics) had a predicted democracy
advantage of more than 5 points on the 12-point
Freedom House scale compared to predominantly
Muslim former Soviet republics such as Azerbaijan or
Kazakhstan.
Nor is there any evidence that the effect of legacies
faded away as the transition took its course. Instead,
the interaction terms in Models 2–4 show that the
predictive power of historical legacies increased over
14
None of the other legacies discussed earlier fared any better—in
fact most of them were statistically insigniﬁcant when added to the
current speciﬁcations (which is why they were omitted from the
current models.)
15
One standard solution to this problem—expanding the sample
size—is of limited use for the study of post-communist democratization,
for the reasons discussed in footnote 1.
16
Thus, one could extend the approach in Table 3 and run crosssectional
regressions over a wide range of years but the interpretation
of coefﬁcients across a large number of models is
cumbersome, makes inefﬁcient use of the data, and ignores serial
correlation.
17
This approach has been used at least in part by Kopstein and
Reilly (2000) and especially Kurtz and Barnes (2002).
TABLE3LegaciesandDemocracy:Cross-SectionalRegressionResults
(1)
Polityregime
1993
(2)
FHdemoc
1993
(3)
Polity
regime1996
(4)
FHdemoc
1996
(5)
WBdemoc
1996
(6)
NITdemoc
1996
(7)
Polity
regime2003
(8)
FHdemoc
2004
(9)
WBdemoc
2004
(10)
NITdemoc
2004
WesternChristian1.309
(2.117)
2.417*
(1.026)
.944
(2.446)
2.635**
(.932)
.733**
(.139)
1.753**
(.391)
-.525
(1.600)
2.259**
(.698)
.864**
(.117)
1.343**
(.328)
Muslim-8.22**
(2.333)
-2.595*
(1.148)
-5.035*
(2.697)
-2.406*
(1.044)
-.289#
(.155)
-.841#
(.438)
-6.598**
(1.764)
-1.639*
(.782)
-.297*
(.131)
-.640#
(.351)
Interwarstatehood4.85*
(2.167)
2.903**
(1.024)
5.187*
(2.505)
3.246**
(.930)
.130
(.138)
1.192**
(.390)
1.861
(1.638)
1.262#
(.697)
.164
(.117)
.699*
(.325)
Energyintensity-.589
(.669)
-.670*
(.313)
-.588
(.773)
-.407
(.284)
-.030
(.042)
-.193
(.119)
-.989*
(.506)
-.475*
(.213)
-.033
(.036)
-.166#
(.095)
Pre-warSovietRep.2.13
(2.89)
1.429
(1.305)
-2.220
(3.347)
.630
(1.185)
-.080
(.176)
.275
(.497)
-4.193*
(2.189)
-3.163**
(.888)
-.214
(.149)
-1.264**
(.398)
Constant4.526#
(2.311)
6.886**
(1.134)
4.590
(2.671)
6.50**
(1.030)
-.48**
(.153)
2.71**
(.432)
1.389**
(1.747)
9.45**
(.772)
-.45**
(.130)
3.645**
(.351)
Obs.27282728282827282827
R2
.64.72.62.77.80.81.82.90.90.90
Standarderrorsinparentheses—#signiﬁcantat10%;*signiﬁcantat5%;**signiﬁcantat1%.
historical legacies and post-communist regime change 915
TABLE4LegaciesandDemocracy:PanelRegressionResults
(1)
FHdemoc
(2)
FHdemoc
(3)
FHdemoc
(4)
FHdemoc
(5)
WBdemoc
(6)
NITdemoc
(7)
Polityregime
(8)
FHdemoc
(9)
NITdemoc
WesternChristian2.06**
(.407)
2.05**
(.416)
2.09**
(.422)
2.09**
(.435)
.726**
(.101)
1.02**
(.343)
-.112
(1.230)
1.885**
(.298)
.964**
(.188)
Muslim-.483
(.584)
-.513
(.583)
-.473
(.605)
-.439
(.622)
-.213*
(.129)
-.812
(.892)
-4.221*
(1.670)
.451
(.495)
.109
(.153)
Interwarstatehood1.25**
(.350)
-1.18#
(.605)
1.26**
(.368)
1.23**
(.382)
.088
(.083)
.330
(.311)
2.520*
(1.045)
.667*
(.260)
.083
(.140)
Energyintensity-.467**
(.121)
-.467**
(.122)
-.467**
(.125)
.320#
(.184)
-.011
(.021)
-.076
(.169)
-1.002**
(.248)
-.199*
(.104)
-.061*
(.027)
Minorityshare-.282*
(.147)
-.313*
(.152)
.456
(.289)
-.302#
(.160)
-.147**
(.039)
-.056
(.087)
-.753
(.534)
-.079
(.117)
-.039
(.036)
%Urban1989.065**
(.018)
.062**
(.019)
.065**
(.019)
.064**
(.020)
-.002
(.004)
.006
(.029)
.098*
(.043)
.043**
(.015)
.009*
(.005)
Pre-warSoviet
Republic
-1.40*
(.596)
-1.37*
(.614)
-1.39*
(.620)
-1.43*
(.644)
-.266*
(.112)
-1.42**
(.524)
-1.563
(1.881)
-.873*
(.443)
-.239*
(.119)
Transitionyear(log).647**
(.143)
.113
(.203)
1.74**
(.337)
1.89**
(.261)
.055
(.068)
-.324
(.245)
.654*
(.333)
.407**
(.124)
-.167#
(.097)
Interwarstate
*Transitionyear
1.33**
(.268)
Minorityshare*
Transitionyear
-.401
(.132)**
Energyintensity
*Transitionyear
-.414**
(.069)
Polityregime.256**
(.024)
FHdemocracy.266**
(.021)
Constant3.77**
(1.03)
5.07**
(1.10)
1.80
(1.20)
1.55
(1.16)
-.072
(.297)
4.31**
(1.468)
-.355
(2.25)
3.47**
(.761)
3.78**
(1.330)
Observations400400400400278272357357269
R2
.60.61.60.60.60.52.29.75.89
Standarderrorsinparentheses—#signiﬁcantat10%;*signiﬁcantat5%;**signiﬁcantat1%.
916 grigore pop-eleches
time: according to Model 2, interwar statehood was a
much stronger predictor of civil and political rights in
2004 (when it accounted for a 2.4-point democracy
gap, signiﬁcant at .001) than during the early transition
years, when its effect was statistically insigniﬁcant
(and even negative). Similarly, the negative interaction
terms in Models 3 and 4 indicate that the adverse
regime effects of ethnic fragmentation and distorted,
energy-intensive economies manifested themselves
much more clearly later in the transition. These ﬁndings
suggest that whereas the chaotic early transition
period had produced a number of deviations from the
“iron law of history,” many of these deviations were
eventually“corrected”over time. This return to history
may help explain the belated democratic progress of
Slovakia, Croatia, and Serbia, countries whose regime
trajectory until the late 1990s was modest compared to
their relatively favorable legacies, as well as some of the
democratic slippage in erstwhile democratic overachievers,
such as Kyrgyzstan and Moldova. The trajectories
of these countries suggest that even
signiﬁcant positive or negative leadership “shocks” do
not usually produce lasting deviations from the broad
parameters of legacy-based regime predictions.18
Nonetheless, a closer comparison of the legacy
coefﬁcients in Models 1 and 5–7 in Table 4 reveals that
even panel data fails to produce a single “foolproof”
legacy, which emerges as statistically signiﬁcant
regardless of the choice of model speciﬁcation and
democracy indicator. For example, Western Christianity
is a statistically and substantively powerful predictor
of democracy according to Freedom House, Nations
in Transit, and the Kaufman et al. measures but is relegated
to insigniﬁcance when using Polity regime
scores. Since similar fates befall indicators for other
potential theories—including modernization, ethnic
diversity, and Islam19
—these ﬁndings should serve as a
call for extreme caution for anyone attempting to
provide monocausal explanations of post-communist
democracy. Even though plausible theoretical arguments
backed by suggestive correlational and case
study evidence can be made on behalf of a variety of
legacies (or—as we will see—nonlegacy accounts),
none of the ﬁndings are sufﬁciently robust to conﬁrm
them conclusively in a cross-country context once
other legacies are taken into consideration.
What are the roots of this remarkable instability of
the legacy-reform link and what are its implications
for our understanding of post-communist democratization?
The next two sections focus on two possible
answers: the intertwined nature of different types of
historical legacies in the post-communist space and
the different dimensions captured by different the
democracy indicators.
Methodological Concerns
and Solutions: Analyzing
Intertwined Legacies
A Romanian proverb tersely states that“money attracts
money,andﬂeasattractﬂeas,”whichaptlydescribesthe
relationship between different types of legacies in
ex-communist countries.The countries of East Central
Europe were not only geographically, historically, and
culturally closer to the West than their Eastern and
Southern brethren, but were also richer, more modern,
less ethnically diverse, with longer histories of democracy
and statehood, and relatively less distorted economies
at the outset of the transition. While high
correlations between indicators within the same
“legacy family” are reassuring about the reliability of
the measures, the high correlation between indicators
of different legacy types—such as between religion and
statehood—create signiﬁcant difﬁculties for interpreting
the role of individual factors.20
From a statistical point of view, the high correlations
of the different legacy indicators (many correlated
at or above .5) can lead to high multicollinearity,
which produces unstable coefﬁcients and inﬂated
standard errors, and, therefore, undermines the utility
of multiple regressions in adjudicating debates
between competing explanations. Under such circumstances,
classical regression analysis is limited to telling
us that the variables matter jointly but the method is
less useful for identifying individual effects.21
This will
18
Thus, the initial underperformance of Slovakia, Croatia, and
Serbia was undoubtedly due to a great extent to the personalities
and leadership styles of Meciar, Tudjman, and Milosevic but
democracy improved almost immediately after these leaders left
power. Lukashenka’s Belarus is still an outlier in this respect but I
would argue that the country’s more favorable legacies nevertheless
make democracy there more likely than in similarly authoritarian
countries, such as Uzbekistan and Turkmenistan.
19
The weaker effect of the“Muslim”variable compared to Table 3 is
due the fact that the panel regressions control for several other
legacy variables (e.g., urbanization and ethnic diversity), which are
strongly correlated with both regime and “Muslim majority” but
which were excluded from the regressions in Table 3 due to concerns
about degrees of freedom.
20
For an overview of these correlations, see Table A in the online
appendix at http://journalofpolitics.org/articles.html.
21
Similarly, the use of F-tests to assess the signiﬁcance of blocks of
variables can help us establish the joint importance of legacies but
it obviously cannot adjudicate between individual legacies.
historical legacies and post-communist regime change 917
be true even if some variables do better than others
in what Kitschelt (2003) terms “the statistical
tournament”—such outcomes are not only sensitive
to model speciﬁcation and the addition/deletion of a
few data points,22
but they may merely reﬂect differences
in measurement error between variables rather
than real differences in substantive effects. In the
current case, regression diagnostics clearly indicate
that a statistical model including all legacy indicators
discussed in the ﬁrst section would suffer from rather
severe multicollinearity problems.23
Meanwhile, the
base models (1 and 5–7) in Table 4 stayed within the
conventional boundaries of multicollinearity24
but did
so at the cost of excluding a number of potentially
relevant explanatory variables, thereby risking biased
estimates due to incorrectly speciﬁed models.
One popular approach to cut through this complexity
has been the use of factor analysis to reduce
the large number of initial conditions to one or two
indices. The approach was pioneered in the postcommunist
context by de Melo et al. (2001). Their
analysis yields two main components, which the
authors interpret as “macroeconomic distortions”
and “over-industrialization.” Such an approach is
attractive considering the limited degrees-of-freedom
in cross-sectional regressions, and the two factors
have in fact been used quite extensively in the literature
(see, e.g., Falcetti, Raiser, and Sanfrey 2002;
Havrylyshyn and van Rooden 1998). However, a
number of drawbacks limit the analytical utility of
this approach. First, the small number of observations
in the sample questions the applicability of
factor analysis, especially with a large number of
variables.25
Second, even if such an approach were
statistically acceptable, the difﬁculty of interpreting
the meaning of these factors critically constrains our
substantive understanding of legacies.
An alternative approach would be to dispense
with regressions altogether and focus instead on case
studies. While such an approach is undoubtedly an
important complement to statistical work, since it
facilitates the detailed tracing of causal mechanisms, it
ultimately does not really sidestep the problematic
nature of intertwined historical legacies. Since countries
rarely differ substantially along only one of the
legacy indicators, it is extremely difﬁcult to identify
cases for which the ceteris paribus assumption underlying
structured case comparisons is actually fulﬁlled.
Nonetheless, such comparisons, as long as they are
carefully selected along a most-similar cases design,
can provide valuable help with untangling the
complex effects of the region’s intertwined historical
legacies.26
In addition to their inherent case-based
insights, structured case comparisons could be used as
the ﬁrst stage in the Bayesian statistical modeling
approach advocated by Western and Jackman as a
potentially promising alternative to multiple regressions
when dealing with small sample sizes and highly
correlated explanatory variables.27
Different Conceptions and Measures
of Democracy
The instability of regression estimates across different
measures of democracy is not a purely postcommunist
phenomenon (Casper and Tuﬁs 2004).
The ﬁndings in Tables 3 and 4 add further support for
Casper and Tuﬁs’ call for a more self-conscious use of
democracy indicators by political scientists. However,
the present discussion goes one step further in that it
links these discrepancies to the different dimensions of
democracy captured by the different indicators used in
this article. Nations in Transit uses the broadest—and
most demanding—conception of democracy, which
requires not only a competitive political process but
also an independent media, a vibrant civil society, and
a well-functioning system of governance and public
administration. By comparison, the other two mea-
22
For example, the negative effect of ethnic fragmentation on FH
democracy in Model 1 of Table 4 disappeared when Latvia and
Estonia were excluded from the analysis. In the same model, the
negative effect for Muslim approached statistical signiﬁcance (.1
one-tailed) if Kyrgyzstan and Azerbaijan were excluded. However,
most other results were less affected by such changes, and no single
country appears to signiﬁcantly affect the overall ﬁndings.
23
Thus, several of the indicators—including Western exports, Communist
bureaucratic legacies, Western Christian, and Prewar Soviet
Republic—had variance inﬂation factors (VIF) above the critical
value of 10, and the condition number for the model was
extremely high at 97.4.
24
All of the independent variables had variance inﬂation factors
(VIF) below 4 and the condition number was around 30.
25
The statistical literature tends to place the lower bounds for
sample size signiﬁcantly above the 28 countries in de Melo et al.’s
(2001) sample. See Cattell (1978) and Tabachnick and Fidell
(2001).
26
Another promising possibility for dealing with overlapping legacies
is to focus on subnational variation but such an approach
suffers from the smaller variation of subnational democracy and
the paucity of subnational democracy indicators.
27
The approach uses similar case comparisons to obtain the starting
values for the relevant coefﬁcients and improve the likelihood
of adjudicating between two or more competing hypotheses measured
by highly correlated indicators (Western and Jackman 1994).
918 grigore pop-eleches
sures28
have a narrower scope but differ in their
emphasis: thus, Freedom House emphasizes political
outcomes (civil and political rights), whereas the
Polity regime score has a strong institutional and procedural
emphasis, thereby reﬂecting a different understanding
of democracy.29
Since different indicators measure distinctive
facets of democracy, their different legacy correlates
can offer a more nuanced, multistage interpretation of
the regime implications of historical legacies. At the
most basic level, democratic countries need institutional
arrangements compatible with genuine democratic
rule (as tracked by Polity). Because the effects of
formal institutions are ﬁltered by informal local
norms and practices, the adoption of “good” institutions
does not necessarily translate uniformly into
civil and political rights (as coded by Freedom House).
Similarly, the quality of democratic governance (captured
by Nations in Transit) hinges on the vitality of
civil society and the competence of the public administration,
and should, therefore, not be expected to
ﬂow automatically from the mere absence of formal
restrictions on civil and political rights. Seen from this
perspective, the achievement of high-quality democracy
requires three distinct steps, each of which may be
affected by different facets of a country’s historical
inheritance.
The legacy correlates of the ﬁrst step—the adoption
of democratic institutions—are reﬂected in
Model 7 of Table 4 and suggest that institutional
choices were more conducive to democracy in urbanized,
non-Muslim countries with longer statehood
traditions and less distorted economies. These correlates
reﬂect the inauspicious institutional choices of
the Central Asian republics, particularly Uzbekistan
and Turkmenistan. Other legacies, such as Western
Christianity, ethnic diversity, and prewar membership
in the Soviet Union, mattered much less at this
stage.
To assess the legacy impact on the second step—
from institutions to civil and political rights—Model 8
in Table 4 includes the Polity regime score as an independent
variable in addition to the legacy indicators
used in Model 7.30
The results suggest that several
legacies turn out to play a signiﬁcant role in understanding
the gap between institutional arrangements
(captured by Polity) and actual freedoms (captured by
Freedom House). The substantively large and highly
signiﬁcant coefﬁcient of Western Christianity conﬁrms
the importance of cultural differences to the
functioning of formal institutions: thus, holding institutions
constant, civil and political rights in Western
Christian countries were signiﬁcantly higher than in
other ex-communist countries. This ﬁnding suggests
that the great democracy boost associated with
Western Christianity occurs at the “implementation”
rather than the institutional choice stage of political
reforms. The gap between formal institutions and
real-life freedoms is also closely related to the ethnic
and state-building challenges that many transition
countries have confronted. Thus, the greater incidence
of rights violations in countries with larger ethnic
minorities (see Model 1) also appears to be mainly a
function of a shortfall between the promises of institutional
design and the actual functioning of these
institutions: thus, the coefﬁcient for Minority share is
moderately large and statistically signiﬁcant in Model
8 but was a weak predictor of Polity regime scores in
Model 7. Meanwhile, based on the comparison of the
same two models, countries with memories of interwar
statehood beneﬁted at both the institutional
choice and the implementation stage, arguably
because the travails of state-building gave strong
leaders more leeway to design favorable (and generally
undemocratic) institutions, while their image as
“fathers” of the country gave them more leeway for
arbitrary measures even within the already loose institutional
constraints on their powers. Finally, countries
with higher urbanization levels at the outset of the
transition also appear to have beneﬁted in terms of
both better formal institutions and superior outcomes
within a given set of institutions, as illustrated by the
signiﬁcant positive effect of %Urban 1989 in Models 7
and 8.
The ﬁnal model in Table 4 address the challenges
of democratic deepening, namely the transition from
basic political and civil rights to a well-functioning,
28
This discussion excludes the fourth measure—Kaufman, Kraay,
and Mastruzzi’s (2005) Voice and Accountability—since its aggregation
of up to 25 different sources makes it signiﬁcantly harder to
interpret substantively.
29
Marshall and Jaggers acknowledge the importance “of civil liberties
to all citizens in their daily lives and in acts of political participation,”
but add that “we do not include coded data on civil
liberties” (2000, 12).
30
This speciﬁcation does not address the endogeneity of these
theoretically prior democracy measures. However, instrumental
variable speciﬁcations (using the lagged average in democracy of a
country’s neighbors as an instrument for theoretically prior
democracy) suggest that the overall ﬁndings of these regressions
with respect to legacy differences are not affected by endogeneity.
However, the instruments are not extremely powerful, leading to
high multicollinearity and inﬂated standard errors, and, therefore,
the results are omitted here.
historical legacies and post-communist regime change 919
democratic society with vibrant intermediary institutions
and a capable and responsive bureaucratic apparatus
(as captured by the Nations in Transit indicator).
The results in Model 9 show that once we control for
basic freedoms, predominantly Western Christian
countries received a large and statistically signiﬁcant
democratic quality boost compared to other transition
countries. The only other statistically signiﬁcant
legacy indicator in Model 9 was Minority share, which
suggests that ethnically diverse countries had a harder
time translating basic freedoms into well-functioning
democratic governance. While the mechanisms
behind these links require further research, these ﬁndings
suggest that the higher quality of democracy in
East-Central Europe may have been due to the fact
that their Western Christian religious/cultural heritage
and their greater ethnic homogeneity contributed to
more effective civil societies.
Finally, the comparison between the coefﬁcients
for the logged transition year in Models 8 and 9 adds
an interesting nuance to the temporal dynamics of
post-communist democratization. The fairly large
positive effect of time in Model 8 suggests a gradual
democratic learning process that allows for greater
political and civil rights to be achieved within a given
set of institutional arrangements. Further research will
be necessary to determine whether this boost is primarily
due to the increasing ability of pro-democracy
political forces to take advantage of formal rules to
pursue their goals, or whether the progress reﬂects the
increasing willingness of the authoritarian camp to
abide by formal democratic constraints. On a less optimistic
note, the negative effects of transition year in
Model 9 suggest that progress in democratic quality
has lagged behind the signiﬁcant increase in basic
rights.
These ﬁndings suggest that the search for a set of
silver bullet explanations of post-communist democracy
is likely to remain elusive for two reasons. First,
the high empirical correlation between analytically
distinct historical legacies is likely to produce unstable
statistical estimates, which make it difﬁcult to draw
deﬁnitive conclusions about the importance of individual
legacies without engaging in multimethod
work. Second, given the conceptualization differences
of different democracy indicators, it should come as
no great surprise (despite the relative neglect of this
insight in much of the democratization literature) that
different facets of democracy should correlate more
strongly with certain legacies than with others. While
such differences may preclude general statements
about the drivers of Democracy, this discussion has
shown that the variation in correlational patterns for
analytically distinct democracy indicators can actually
be used to achieve a more nuanced understanding of
the political dynamics of the democratization.
Alternative Explanations
Having established the strong and lasting effect of historical
legacies on post-communist political reforms, I
now brieﬂy address the implications of these ﬁndings
for other explanations of democratization. I ﬁnd that
legacies remain powerful predictors of regime type
even when we control for several prominent alternative
explanations such as initial election outcomes,
institutional choices, geographic diffusion, and
external conditionality. Meanwhile, the explanatory
power of these alternatives theories is substantially
diminished—and in some cases disappears altogether
—once legacy differences are taken into account.
One alternative explanation focuses on the regime
effects of institutional choices in the period immediately
after the collapse of Communism. One of
the central post-communist institutional design
dilemmas—the choice between presidential and parliamentary
systems—can serve as a lens for understanding
the relationship between historical legacies,
institutional choice and post-communist democracy.
While several authors have explored the determinants
of institutional choice (Bunce 1997; Frye 1997; Luong
2000), few have employed large-N methods to assess
the relationship between presidential powers and
post-communist democracy.31
However, the potential
pitfalls of presidentialism for democracy have been
extensively discussed in the broader democratization
literature (Linz and Valenzuela 1994; Mainwaring
1993; Stepan and Skach 1993).
Another prominent theoretical approach for
understanding post-communist regime variation
focuses on the initial power balance between the communists
and their democratic challengers. Thus, Fish
(1998a, 1998b) argues that the extent of the political
displacement of the former communists in the ﬁrst
post-communist elections sets in motion a pathdependent
economic and political reform process,
which eventually yields regime outcomes closely
related to the outcome of the initial elections. Similarly,
McFaul (2002) argues that the power balance
between democrats and authoritarians at the outset of
31
Two exceptions are Ishiyama and Velten (1998) and Horowitz
(2003), who conclude that presidential powers had a negligible
effect on democratization.
920 grigore pop-eleches
the transition determines future democracy prospects,
as the stronger party imposes its will on the political
losers.
A third alternative explanation focuses on Western
conditionality. The history of the last decade abounds
with instances of democratic reforms initiated in
response to external incentives ranging from economic
incentives to diplomatic pressures, and even military
force. Due to space constraints, the current analysis
focuses on the most prominent and most unevenly
distributed potential beneﬁt of policy compliance with
Western conditionality: EU accession. Kurtz and
Barnes (2002) have argued that the prospects of EU
accession were an important predictor of postcommunistregimetypeacrosstheirsampleof
28countries,
even after controlling for several legacies.
Moreover, analyses of the democratization trajectories
of individual East European countries have identiﬁed
numerous instances in which EU conditionality has
signiﬁcantly shaped speciﬁc political decisions on
issues pertaining to civil rights and better governance.32
Finally, Kopstein and Reilly (2000) have focused
on the role of geographic diffusion and have argued
that in addition to domestic factors, post-communist
regime trajectories were affected by cross-national
inﬂuences. The authors use the distance from the West
and the lagged degree of democracy in neighboring
countries as a proxy of geographic diffusion and found
it to be a powerful predictor of democratization even
when controlling for initial election outcomes and
bureaucratic rectitude.
While all of the alternative explanations discussed
above are theoretically plausible and are in fact highly
correlated with the different democracy indicators, the
crucial question is how these explanations fare when
pitted against the temporally and theoretically prior
historical legacies discussed in the ﬁrst part of the
article. Table 5 answers this question by testing these
alternative theories against the “standard battery” of
legacy indicators used in the earlier sections.33
Before discussing the predictive power of alternative
theoretical explanations, it is worth noting the
remarkably stable coefﬁcients of the different legacy
indicators in Table 5, most of which remained signiﬁcant
predictors of democracy irrespective of the other
variables introduced in Models 2–8. Similarly, except
for a noticeable (but moderately sized) boost in Model
2, the overall explanatory power of Models 3–8
increased only minimally compared to the baseline
legacy speciﬁcation.34
Turning to the individual theoretical alternatives,
Model 2 suggests that higher presidential powers had
deleterious consequences for civil and political rights
even once we account for the important legacy differences
between ex-communist regimes: thus, a one
standard deviation increase in presidential powers35
was equivalent to a 1-point decline on the 12-point FH
scale.36
Models 3 and 4 tested the two versions of the
initial elections hypothesis, using Fish’s (1998a) sixpoint
initial election score and McFaul’s (2002) three
categories of initial power balance, respectively. While
the initial election explanation has come under attack
for focusing on proximate causes (Kitschelt 2003), this
criticism can be taken a step farther considering that,
once we control for legacies in Models 3 and 4, the two
power-balance indicators are substantively and statistically
insigniﬁcant, while coefﬁcients of the legacy
indicators are virtually unchanged compared to the
base model. These ﬁndings suggest that rather than
being critical junctures in the democratization
process, initial elections may be signals about the
nature of historical legacies, which drive the long-term
prospects of democracy in the region.
Using Kurtz and Barnes’ (2002) potential EU candidate
indicator, Model 5 in Table 7 conﬁrms the
importance of EU accession incentives, which were
associated with a 1.3 point democracy increase even
after we control for legacies. However, Model 6 raises
some questions about the robustness of the ﬁnding:
thus, by simply recording Croatia from a noncandidate
to a potential candidate at the outset of the
transition,37
the size of the coefﬁcient is reduced by
32
See e.g., Vachudova’s (2004) discussion of EU integration in six
East European countries.
33
Model 1 presents the basic legacy speciﬁcation for reference purposes.
The tests in this section used Freedom House democracy
scores, but the results were remarkably similar for other measures
of democracy.
34
The R-squared jump in Model 7 is simply an artifact of the
smaller sample size in that model.
35
The measure used here is Frye’s update of the Tucker-Hellman
index of presidential powers, which had a 1–21 range for the
countries in the present sample. Similar results were obtained
using a dichotomous super-presidentialism measure.
36
These statistical results are conﬁrmed by the surprisingly democratic
trajectories of Moldova and Mongolia, two countries whose
rejection of presidentialism appears to have helped them overcome
otherwise unpromising legacies.
37
This recoding is not merely idle speculation: as Kurtz and Barnes
(2002, fn4) acknowledge, “Croatia was also in position to join
these negotiations in 1991 but the war in the former Yugoslavia
intervened.” However, one can argue that the Yugoslav war would
not have changed Croatia’s medium-term integration prospects if
it had not also created a semiauthoritarian, and nationalist regime
under Tudjman (in which case coding Croatia as a non-candidate
amounts to posthoc reasoning.)
historical legacies and post-communist regime change 921
TABLE5LegaciesandAlternativeExplanations:PanelRegressionResults
(1)
FHdemocracy
(2)
FHdemocracy
(3)
FHdemocracy
(4)
FHdemocracy
(5)
FHdemocracy
(6)
FHdemocracy
(7)
FHdemocracy
(8)
FHdemocracy
WesternChristian2.060**
(.407)
2.134**
(.402)
1.969**
(.518)
1.596**
(.517)
1.366**
(.453)
1.632**
(.484)
1.569**
(.599)
1.969**
(.421)
Muslim-.483
(.584)
-.796
(.576)
-.410
(.674)
-.116
(.678)
-.468
(.591)
-.458
(.574)
.020
(.743)
-.284
(.605)
Interwarstatehood1.253**
(.350)
1.288**
(.335)
1.234**
(.363)
1.272**
(.343)
.851*
(.369)
1.123**
(.352)
1.653**
(.474)
1.440**
(.375)
Energyintensity-.467**
(.121)
-.252*
(.123)
-.456**
(.125)
-.481**
(.120)
-.484**
(.123)
-.467**
(.120)
-.574**
(.173)
-.465**
(.121)
Pre-warSoviet
Republic
-1.400*
(.596)
.573
(.751)
-1.355*
(.599)
-1.446*
(.588)
-1.241*
(.624)
-1.319*
(.617)
-1.253#
(.737)
-1.062#
(.625)
%Urban1989.065**
(.018)
.043*
(.019)
.065**
(.019)
.068**
(.018)
.059**
(.019)
.064**
(.019)
.111**
(.021)
.059**
(.018)
Minorityshare-.282#
(.147)
-.492**
(.143)
-.287#
(.152)
-.250#
(.144)
-.196
(.156)
-.260#
(.148)
-.450*
(.212)
-.197
(.151)
Transitionyear(log).647**
(.143)
.783**
(.140)
.648**
(.144)
.644**
(.142)
.645**
(.143)
.647**
(.143)
.748*
(.352)
.553**
(.152)
Presidentialpowers-.219**
(.042)
Initialelection
outcome
.069
(.182)
Initialpowerbalance.423
(.301)
PotentialEU
candidate(1)
1.325*
(.569)
PotentialEU
candidate(2)
.627
(.625)
DistancetoWest.228
(.289)
Openness.012
(.093)
Avg.neighborFH
democracy
.136*
(.066)
Constant3.773**
(1.030)
5.984**
(1.169)
3.544**
(1.289)
3.154*
(1.134)
3.779**
(1.038)
3.691**
(1.011)
-.138
(2.920)
2.974**
(1.102)
Observations400400400400400400167400
R2
.60.64.60.61.60.61.76.60
Standarderrorsinparentheses—#signiﬁcantat10%;*signiﬁcantat5%;**signiﬁcantat1%.
922 grigore pop-eleches
half compared to Model 5 and no longer even
approaches statistical signiﬁcance. Of course, it is conceivable
that a more ﬁne-tuned measure of the overtime
variation in EU accession incentives would
produce stronger results but analyzing such a measure
would have to deal with signiﬁcant endogeneity problems
(as discussed below).
Finally, Kopstein and Reilly’s (2000) geographic
diffusion explanation also receives mixed support
once pitted against historical legacies. On one hand
Model 7 shows that the effect of geographic distance
from the West38
and the country’s international open-
ness39
are statistically and substantively negligible once
other legacies are taken into account. This ﬁnding suggests
that the international openness index, which
Kopstein and Reilly interpret as a mechanism for the
cross-border diffusion of the democratic norms, does
not appear to have an independent effect beyond the
role of domestic legacies, thereby raising the possibility
that the high correlation between openness and
democracy is merely a reﬂection of the index’s strong
modernization dimension.40
On the other hand,
Model 8, which uses the average prior year democracy
of a given country’s neighbors as a geographic diffusion
proxy, indicates that the regime type in neighboring
countries matters for democracy even once we
account for the powerful (and geographically correlated)
domestic structural conditions: thus, Model 8
predicts a 1.3-point civil and political rights difference
between the country with the least and the most
democratic neighbors in 2004, which represents a reasonably
large substantive effect, but one that is nonetheless
smaller than that of key domestic legacies. This
ﬁnding suggests one possible mechanism for overcoming
unfavorable domestic legacies; however, even
the beneﬁts of diffusion are ultimately unevenly distributed,
as countries fortunate enough to have democratic
neighbors are much more likely to adopt outside
democratic practices than countries stuck in the
“wrong” neighborhood.
So far, the statistical evidence has provided
mixed support for nonlegacy explanations of postcommunist
regime change, whereas the explanatory
power of legacies was only minimally affected by the
inclusion of alternative explanations. Furthermore,
except for geographic diffusion, these explanations
have to cope with a serious endogeneity problem due
to the fact that presidentialism, initial power balance,
and potential EU membership were all highly correlated
with a variety of historical legacies. These high
correlations conﬁrm the anecdotal evidence about the
high prevalence of superpresidential regimes in the
former Soviet Union, as well as about the much better
European integration prospects of the more developed
Western Christian countries.41
Similarly, the correlations
suggest that initial elections did not occur in a
vacuum, but were considerably inﬂuenced by systematic
cross-country legacy differences, which explains
the much greater resilience of unreformed or minimally
reformed Communists in the former Soviet
Union and parts of the Balkans, whereas in EastCentral
Europe the “change of guard” had happened
in the ﬁrst election.42
The endogeneity suggested by these correlations
makes it very difﬁcult to determine what—if any—
independent democratization effects can be attributed
to institutional choice, initial elections, or European
integration incentives. Since historical legacies seem to
drive both these intermediary factors and regime outcomes,
it cannot be ruled out that the correlation
between these factors and democracy could be entirely
spurious (merely reﬂecting their common correlation
with legacies.) Under such circumstances it is not sufﬁcient
to simply include legacy controls in the regression
models that analyze the effect of such endogenous
variables. Instead, one would have to resort to instrumental
variable regressions, which would require
ﬁnding adequate instruments, i.e., variables which are
strong predictors of institutional choice, initial elections,
or European integration but which only affect
democracy through this channel. However, such
instruments are extremely difﬁcult to ﬁnd in the
current context and are therefore beyond the scope of
the present analysis. Absent such instruments, it is
38
These results are not affected by the fact that I used the log
instead of the absolute distance (as Kopstein and Reilly do) from a
country’s capital to Vienna, Berlin, or Helsinki (whichever is
closest).
39
I would like to thank the authors for sharing their data. The
four-year lag in the openness measure is in accordance with the
authors’ expectation that learning due to openness would take
three to four years to show its full effects (Kopstein and Reilly 2000,
fn 30). While the current analysis does not replicate the spatial lag
statistical tests of the original paper, I believe that the average
neighbors’ score captures their logic.
40
For example, the openness index is correlated at .73 with GDP/
capita.
41
For example Prewar Soviet Union membership was correlated at
.89 with presidential powers, while Potential EU membership was
correlated at .76 with Western Christianity. For a more detailed
overview see Table C in the online appendix.
42
Both McFaul (2002, 238) and Fish (1998a) acknowledge this
potential endogeneity but neither author pursues the issue in more
detail, and Fish points out that“the outcome of the initial elections
certainly cannot readily be traced to a particular structural, cultural,
or institutional factor” (78).
historical legacies and post-communist regime change 923
impossible to determine the precise contribution of
such explanations to our understanding of crosscountry
democracy differences but it is highly likely
that the relatively modest effects revealed in Table 5
constitute an upper bound of this inﬂuence. Either
way, this analysis has demonstrated that historical
legacies need to be taken seriously not only because of
their own intrinsic importance in post-communist
democratization but also because our understanding
of alternative explanations has to be embedded in the
complicated reality of the region’s intertwined historical
legacies.
Conclusion
This article has provided a systematic analysis of the
effect of historical legacies on post-communist
democratization. Contrary to early optimistic expectations,
democracy has not become “the only game in
town” among the ex-communist countries, and the
patterns underlying this wide variety of outcomes
were shaped to a remarkable degree by the past.
History matters not only because some countries had
a democratic headstart, but because countries with
different legacies experienced divergent trajectories
over the course of the post-communist transition. In
other words, historical legacies seem to matter more
rather than less as the post-communist transformation
takes its course.
The analysis does not provide a deﬁnitive answer
to the question of which structural conditions matter
most for the establishment of democracy in the
region. Indeed, the analysis of different democracy
indicators suggests that the search for such “universal”
explanations is likely to be misguided. Instead, the
article has shown that different facets and dimensions
of democracy are affected to different degrees by the
various legacies. Thus, the institutional dimension of
democracy, reﬂected by Polity regime scores, suffered
primarily in Muslim countries with no prior statehood
experience, energy intensive economies and low urbanization.
Meanwhile, the political and civil rights measured
by Freedom House were also affected by Western
Christianity, prewar Soviet Union membership, and
ethnic fragmentation, which account for much of the
gap between formal political institutions and the
actual freedoms granted to citizens. The challenge of
overcoming the second important gap—between
formal rights and democratic quality (as measured
by Nations in Transit)—was easier for ethnically
homogenous and predominantly Western Christian
countries.
The high correlation between different historical
legacies suggests that post-communist regime outcomes
were overdetermined, in the sense that the
region’s democracy gradient coincides with overlapping
legacy differences in geographical and cultural
proximity to the West, state- and nation-building challenges,
and socioeconomic development/distortions.
The analytical difﬁculties arising from this legacy
overlap should caution us against excessively conﬁdent
causal claims about the importance of individual
variables. These difﬁculties emphasize the need for
methodological pluralism through the combination of
theoretical/conceptual work, statistical analysis, and
in-depth case studies/small N comparisons. In particular,
much important work remains to be done on
establishing the mechanisms for the legacy-reform
link. Nonetheless, this article has established the
extremely powerful joint inﬂuence of historical legacies
and post-communist political transformations
and has shown that prominent theoretical
alternatives—such as institutional choice, initial election
outcomes, and European integration—have
played a much more modest role and need to be analyzed
in the context of these legacy differences.
The specter of determinism raised by these ﬁndings
should be carefully weighed in both theoretical
and practical terms. On one hand, the last 17 years
have not produced a Fukuyama-style “end of history,”
but rather a return to historically rooted differences.
But while historical legacies have largely circumscribed
the degree of democratic progress in excommunist
countries, this does not mean that outcomes
were predetermined or that there was no room
for agency or constructive institutional choices. Some
countries—such as Moldova, Mongolia, and for a
while the Kyrgyz Republic—were more democratic
than their legacies would have predicted, whereas
others—such as Belarus, Croatia, and Yugoslavia performed
below legacy-based expectations. Moreover, a
variety of subtypes have developed among both
democratic and authoritarian regimes in the region,43
suggesting that countries do not simply follow a deterministically
preordained path towards a uniform endpoint.
Finally, the recent events in Georgia, Ukraine,
and Kyrgyzstan serve as reminders that the postcommunist
transition is not yet ﬁnished, which means
that the correlational patterns between legacies and
reforms could change signiﬁcantly in the not too
distant future.
What the close link between historical legacies and
reforms does suggest, however, is that countries trying
43
See, for example, Roeder (1994) and Comisso (1997).
924 grigore pop-eleches
to escape their past face an uphill battle in trying to
develop well functioning democratic institutions. The
optimistic “possibilism” of the early transition should
be replaced by a more historically grounded realism
about the prospects of political liberalism in the
former Leninist countries. An analytical emphasis on
legacies should not be construed as a fatalist acceptance
of the status quo or as an alternative to searching
for institutional or international solutions to the
region’s political challenges. However, unless these
solutions are grounded in a thorough understanding
of the relationship between legacies, institutions, and
reform outcomes, such efforts are likely to perpetuate
the uneven success of democracy in the region by
focusing on the wrong questions.
Manuscript submitted 7 June 2006
Manuscript accepted for publication 18 January 2007
References
Acemoglu, Daron, Simon Johnson, and James A. Robinson. 2001.
“The Colonial Origins of Comparative Development: An
Empirical Investigation.” American Economic Review 91 (5):
1369–33.
Bloom, David, and Jeffrey Sachs. 1998. “Geography, Demography,
and Economic Growth in Africa.”Brookings Papers on Economic
Activity 2: 207–73.
Barro, Robert. 1999.“The Determinants of Democracy.” Journal of
Political Economy 107 (6): 158–83.
Bunce, Valerie. 1997. “Presidents and the Transition in Eastern
Europe.” In Presidential Institutions and Democratic Politics, ed.
Kurt von Mettenheim. Baltimore: Johns Hopkins University
Press, 161–76.
Bunce, Valerie. 2005. “The National Idea: Imperial Legacies and
Post-Communist Pathways in Eastern Europe.” East European
Politics and Societies 19 (3): 406–42.
Casper, Gretchen, and Claudiu Tuﬁs. 2004. “Correlation versus
Interchangeability: The Limited Robustness of Empirical Findings
on Democracy using Highly Correlated Datasets.” Political
Analysis 11 (2): 1–11.
Cattell, Raymond. 1978. The Scientiﬁc Use of Factor Analysis in
Behavioral and Life Sciences. New York: Plenum Press.
Clem, Ralph, and Peter Craumer. 1997. “Urban-Rural Voting Differences
in Russian Elections, 1995–1996: A Rayon-Level
Analysis.” Post-Soviet Geography and Economics 38 (7): 379–95.
Crawford, Beverly, and Arend Lijphart, eds. 1997. Liberalization
and Leninist Legacies: Comparative Perspectives on Democratic
Transitions. Berkeley, CA: International and Area Studies.
Comisso, Ellen. 1997.“Is the Glass Half Full or Half Empty? Reﬂections
on Five Years of Competitive Politics in Eastern Europe.”
Communist and Post-Communist Studies 30 (1): 1–21.
Darden, Keith, and Anna Grzymala-Busse. 2005. “The Great
Divide: Pre-communist Schooling and Post-communist
Trajectories.” Presented at the 2005 annual meeting of the
American Political Science Association, Washington, DC.
de Melo, Martha, Cevdet Denizer, Alan Gelb, and Stoyan Tenev.
2001. “Circumstance and Choice. The Role of Initial Conditions
and Policies in Transition Economies.” World Bank Economic
Review 15 (1): 1–31.
Di Palma, Giuseppe. 1990. To Craft Democracies: An Essay on
Democratic Transitions. Berkeley: University of California Press.
Diamond, Larry Jay, and Marc F. Plattner. 1996. The Global Resurgence
of Democracy. 2nd ed. Baltimore: Johns Hopkins University
Press.
EBRD. 2001. Transition Report 2001: Energy in Transition. London:
European Bank for Reconstruction and Development.
Ekiert, Grzegorz. 2003.“Patterns of Post-Communist Transformation.”
In Capitalism and Democracy in Central and Eastern
Europe, ed. Grzegorz Ekiert and Stephen Hanson. New York:
Cambridge University Press, 89–119.
Ekiert, Grzegorz, and Stephen Hanson. 2003. “Time, Space, and
Institutional Change in Central and Eastern Europe.” In Capitalism
and Democracy in Central and Eastern Europe, ed.
Grzegorz Ekiert and Stephen Hanson. New York: Cambridge
University Press, 15–48.
Falcetti, Elisabetta, Martin Raiser, and Peter Sanfey. 2002.“Defying
the Odds: Initial Conditions, Reforms, and Growth in the First
Decade of Transition.” Journal of Comparative Economics 30
(2): 229–50.
Fish, M. Steven. 1998a. “Democratization’s Requisites: The Postcommunist
Experience.” Post-Soviet Affairs (July–September):
212.
Fish, M. Steven. 1998b.“The Determinants of Economic Reform in
the Post-Communist World.” East European Politics and Societies
12 (1): 31.
Fish, M. Steven. 2002.“Islam and Authoritarianism.” World Politics
55 (1): 4–37.
Freedom House. 2005a. Freedom in the World 2005: The Annual
Survey of Political Rights and Civil Liberties. Lanham, MD:
Rowman & Littleﬁeld Publishers.
Freedom House. 2005b. Nations in Transit 2005: Democratization
from Central Europe to Eurasia. Lanham, MD: Rowman &
Littleﬁeld Publishers.
Frye, Timothy. 1997. “A Politics of Institutional Choice—PostCommunist
Presidencies.” Comparative Political Studies 30 (5):
523–52.
Grugel, Jean. 1999. “Democracy without Borders: Transnationalization
and Conditionality in New Democracies.” Routledge/
ECPR Studies in European Political Science 10. New York:
Routledge.
Grzymala-Busse, Anna. 2002. Redeeming the Communist Past: The
Regeneration of Communist Parties in East Central Europe.
New York: Cambridge University Press.
Havrylyshyn, Oleh, and Ron van Rooden. 1998. “Recovery and
Growth in Transition Economies 1990–97: A Stylized Regression
Analysis.” IMF Working Paper No. 98/141.
Horowitz, Shale. 2003. “Sources of Post-Communist Democratization:
Economic Structure, Political Culture, War, and Political
Institutions.” Nationalities Papers 31 (2): 119–37.
Huntington, Samuel P. 1991. The Third Wave: Democratization in
the Late Twentieth Century. Norman: University of Oklahoma
Press.
Ishiyama, John. 1997. “The Sickle or the Rose? Previous Regime
Types and the Evolution of the Ex-Communist Parties in PostCommunist
Politics.” Comparative Political Studies 30 (3): 299–
330.
Ishiyama, John, and Mathew Velten. 1998.“Presidential Power and
Democratic Development in Post-Communist Politics.” Communist
and Post-Communist Studies 31 (3): 217–33.
Janos, Andrew C. 1989. “The Politics of Backwardness in Continental
Europe, 1780–1945.” World Politics 41 (3): 325–58.
Janos, Andrew C. 1994. “Continuity and Change in EasternEurope—Strategies
of Post-communist Politics.”East European
Politics and Societies 8 (1): 1–31.
historical legacies and post-communist regime change 925
Janos, Andrew C. 2000. East Central Europe in the Modern World:
The Politics of the Borderlands from Pre- to Post-communism.
Stanford, CA: Stanford University Press.
Jowitt, Ken. 1992. New World Disorder: The Leninist Extinction.
Berkeley: University of California Press.
Karl, Terry L., and Philippe C. Schmitter. 1991. “Modes of Transition
in Latin America, Southern and Eastern Europe.” International
Social Science Journal (May): 269–84.
Katchanovski, Ivan. 2000. “Divergence in Growth in PostCommunist
Countries.” Journal of Public Policy 20 (1): 55–81.
Kaufman, Daniel, Aart Kraay, and Massimo Mastruzzi. 2005.
“Governance Matters: Governance Indicators for 1996–2004.”
Washington: World Bank Policy Research Department
Working Paper.
Kitschelt, Herbert, Zdenka Mansfeldova, Radoslaw Markowski,
and Gábor Tóka. 1999. Post-Communist Party Systems: Competition,
Representation, and Inter-Party Cooperation. Cambridge:
Cambridge University Press.
Kitschelt, Herbert. 2001. “Post-Communist Economic Reforms:
Causal Mechanisms and Concomitant Properties.”Presented at
the 2001 annual meeting of the American Political Science
Association.
Kitschelt, Herbert. 2003. “Accounting for Outcomes of Postcommunist
Regime Diversity: What Counts as a Good Cause?”
In Capitalism and Democracy in Central and Eastern Europe, ed.
Grzegorz Ekiert and Stephen Hanson. New York: Cambridge
University Press, 49–88.
Kopstein, Jeffrey 2003. “Post-communist Democracy—Legacies
and Outcomes.” Comparative Politics 35 (2): 231–50.
Kopstein, Jeffrey, and David Reilly. 2000. “Geographic Diffusion
and the Transformation of the Post-communist World.” World
Politics 53 (1): 1–37.
Kurtz, Marcus J., and Andrew Barnes. 2002. “The Political Foundations
of Post-Communist Regimes—Marketization, Agrarian
Legacies, or International Inﬂuences.” Comparative Political
Studies 35 (5): 524–53.
La Porta, Rafael, Florencio Lopez-de-Silanes, Andrei Shleifer, and
Robert Vishny. 1999. “The Quality of Government.” Journal of
Law, Economics, and Organization 15 (4): 222–79.
Lal, Deepak. 1998. Unintended Consequences: The Impact of Factor
Endowments, Culture, and Politics on Long-Run Economic Performance.
Cambridge: MIT Press.
Linz, Juan J., and Arturo Valenzuela. 1994. The Failure of Presidential
Democracy. Baltimore: Johns Hopkins University Press.
Lipset, Seymour Martin. 1959.“Some Social Requisites of Democracy:
Economic Development and Political Legitimacy.”American
Political Science Review 53 (1): 69–105.
Luong, Pauline Jones. 2000. “After the Break-Up: Institutional
Design in Transitional States.” Comparative Political Studies 33
(5): 563–92.
Mainwaring, Scott. 1993. “Presidentialism, Multipartism, and
Democracy—the Difﬁcult Combination.” Comparative Political
Studies 26 (2): 198–228.
Marshall, Monty G., and Keith Jaggers. 2000. POLITY IV PROJECT
Political Regime Characteristics and Transitions, 1800–1999
Dataset Users’ Manual. College Park, MD: CIDCM.
Marshall, Monty G., and Keith Jaggers. 2005. POLITY IV PROJECT
Political Regime Characteristics and Transitions, 1800–2003.
College Park, MD: Integrated Network for Societal Conﬂict
Research, University of Maryland.
McFaul, Michael. 2002. “The Fourth Wave of Democracy and
Dictatorship—Noncooperative Transitions in the Postcommunist
World.” World Politics 54 (2): 212–44.
O’Donnell, Guillermo A., Philippe C. Schmitter, and Laurence
Whitehead. 1986. Transitions from Authoritarian Rule.
Baltimore: Johns Hopkins University Press.
Pop-Eleches, Grigore. 1999. “Separated at Birth or Separated by
Birth? The Communist Successor Parties in Romania and
Hungary.” East European Politics and Societies 13 (1): 117–47.
Przeworski, Adam. 1991. Democracy and the Market. Cambridge:
Cambridge University Press.
Przeworski, Adam, Michael E. Alvarez, Jose Antonio Cheibub, and
Fernando Limongi. 2000. Democracy and Development: Political
Institutions and Well-Being in the World, 1950–1990. New
York: Cambridge University Press.
Pridham, Geoffrey, and Attila Ágh. 2001. Prospects for Democratic
Consolidation in East-Central Europe. Manchester and New
York: Manchester University Press.
Roeder, Philip G. 1994. “Varieties of Post-Soviet Authoritarian
Regimes.” Post-Soviet Affairs 10 (1): 61–101.
Roeder, Philip G. 1999. “Peoples and States after 1989: The Political
Costs of Incomplete National Revolutions.”Slavic Review 58
(4): 854–82.
Stepan, Alfred, and Cindy Skach. 1993. “Constitutional Frameworks
and Democratic Consolidation—Parliamentarianism
versus Presidentialism.” World Politics 46 (1): 1–22.
Stepan, Alfred, and Graeme B. Robertson. 2003. “An ‘Arab’ More
Than ‘Muslim’ Electoral Gap.” Journal of Democracy 14 (3):
30–44.
Tabachnick, Barbara, and Linda Fidell. 2001. Using Multivariate
Statistics. 4th ed. Needham Heights, MA: Allyn & Bacon.
Vachudova, Milada Anna. 2004. Europe Undivided: Democracy,
Leverage, and Integration After Communism. Oxford: Oxford
University Press.
Western, Bruce, and Simon Jackman. 1994.“Bayesian Inference for
Comparative Research.” American Political Science Review 88
(2): 412–23.
Whitehead, Laurence. 1996. “Democracy and Decolonization:
East-Central Europe.” In The International Dimensions of
Democratization: Europe and the Americas, ed. Laurence Whitehead.
New York: Oxford University Press, 356–89.
Wittenberg, Jason. 2006. Crucibles of Political Loyalty: Church Institutions
and Electoral Continuity in Hungary. Cambridge:
Cambridge University Press.
Grigore Pop-Eleches is assistant professor of politics
and international affairs, Princeton University,
Princeton, NJ 08544.
926 grigore pop-eleches