Comparative Political Studies

http://cps.sagepub.com/

Is There a Necessary Condition for Democracy? The Role of State

Capacity in Postcommunist Countries
Jessica Fortin
Comparative Political Studies 2012 45: 903 originally published online 5
December 2011
DOI: 10.1177/0010414011428587

The online version of this article can be found at:

http://cps.sagepub.com/content/45/7/903

Published by:

http://www.sagepublications.com

Additional services and information for Comparative Political Studies can be found at:

Email Alerts: http://cps.sagepub.com/cgi/alerts

Subscriptions: http://cps.sagepub.com/subscriptions

Reprints: http://www.sagepub.com/journalsReprints.nav

Permissions: http://www.sagepub.com/journalsPermissions.nav

>> Version of Record - Jun 11, 2012

OnlineFirst Version of Record - Dec 5, 2011

What is This?

Downloaded from cps.sagepub.com at LUND UNIVERSITY LIBRARIES on December 28, 2012

428587 CPS

Comparative Political Studies

Is There a Necessary 45(7) 903–930

© The Author(s) 2012
Reprints and permission:
Condition for sagepub.com/journalsPermissions.nav
DOI: 10.1177/0010414011428587
Democracy? The Role http://cps.sagepub.com

of State Capacity in
Postcommunist Countries

Jessica Fortin1

Abstract
Although postcommunist countries share a common past, the variability of
outcomes in both democracy and economic reform is very large in the re-
gion. Only a few countries have become Western-type democracies in Cen-
tral and Eastern Europe and the Baltic. By contrast, the norm is clearly not
democracy for other Soviet successor states. In this article, the author attri-
butes this variation to differences in the infrastructural capacity of the state.
Using both quantitative and qualitative analyses within 26 postcommunist
countries, the author argues that for democracy to flourish, the state must
first possess the necessary means to maintain law and order and to protect
the rights of citizens, in other words, to ensure the maintenance and delivery
of essential public goods. The results show that the links between a strong
state that has been able to apply a definitive set of rules and democratic
institutions are clear.

Keywords
state capacity, democratization, postcommunism, transition

1
GESIS—Leibniz Institut für Sozialwissenschaften, Mannheim, Germany

Corresponding Author:
Jessica Fortin, GESIS—Leibniz Institut für Sozialwissenschaften, Postfach 12 21 55,
Mannheim D-68072, Germany
Email: [email protected]

Downloaded from cps.sagepub.com at LUND UNIVERSITY LIBRARIES on December 28, 2012

904 Comparative Political Studies 45(7)

For an increasing number of contributors to the discipline, the connection

between a strong state and democratic institutions is obvious (Bunce, 2000;
Fukuyama, 2004; Hashim, 2005; Hendley, 1997; Holmes, 1997; Huber,
Rueschemeyer, & Stephens, 1999; Kaufman, 1999; Linz & Stepan, 1996;
O’Donnell, 1993; Przeworski, 1995; Roberts & Sherlock, 1999; Sharlet,
1998; Suleiman, 1999; Tilly, 2007; Wang, 1999). As Linz and Stepan (1996)
argue, an effective state is essential to support the other building blocks of
democratic consolidation (civil, political, economic, and the rule of law). In
other words, for democracy to flourish, the state must possess the means nec-
essary to maintain the rule of law, to protect the rights of citizens, and to regu-
late economic transactions. However compelling in theory, this proposition
has never been rigorously tested in the context of postcommunist countries.
In this article, I argue that postcommunist patterns of state capacity offer
crucial insights into the types of regimes—democratic or authoritarian—that
emerged after the downfall of communism. For this purpose, I draw on a
mixture of qualitative and quantitative methodologies to evaluate how well
this hypothesis reflects the reality of postcommunist transformations. Using
a five-item index of state capacity as the main independent variable, I
demonstrate the importance of state capacity for regime outcomes in two
steps. First, I establish state capacity as a necessary condition for democracy,
using a qualitative method to test probabilistic hypotheses: the methodology
of necessary conditions as developed by Braumoeller and Goertz (2000). In a
second step, I perform time-series cross-sectional (TSCS) estimations on 26
countries from 1989 to 2006 to explain levels of democracy and confirm the
patterns observed in the first step. Following this demonstration, I conduct
additional tests to assess the direction of the causality between the variables
included in this study.
The combined results of these analyses lead to three interconnected con-
clusions. First, effective state capacity seems to be a necessary—but not
sufficient—condition for democracy. Postcommunist countries with com-
paratively high levels of initial state capacity at the time of transition, under-
stood via proxy of the quality of provision of a basic class of public goods,
were more likely to move toward democracy. Second, when explored in a
multivariate design accounting for the effect of time, the empirical connec-
tion between state capacity and democracy remains robust among postcom-
munist countries. On one hand, the links between a strong state that has been
able to apply a definitive set of rules and democratic institutions are clear.
On the other hand, where state capacity was more limited after indepen-
dence was gained, democracy was a less likely outcome. Finally, however,
the direction of causality among variables cannot be established in a

Downloaded from cps.sagepub.com at LUND UNIVERSITY LIBRARIES on December 28, 2012

Fortin 905

unidirectional way, which indicates that state capacity and levels of democ-
racy likely have mutually constitutive and self-reinforcing effects.

How State Capacity Influences Democratization

A growing body of literature proposes causal mechanisms linking quality of
democracy to state capacity. The most influential of such recent theoretical
developments is what Grzymala-Busse (2007) has termed robust competi-
tion, where high levels of competition among political parties during the
postcommunist period explain why state resources were less exploited by
ruling coalitions in Hungary, Poland, Estonia, Slovenia, and Lithuania, by
contrast with the Czech Republic, Slovakia, Latvia, and Bulgaria. In the
former group of countries, opposition parties have been vocal and critical in
all successive governments as of 1990, even managing to successfully
monitor the level of state exploitation. Grzymala-Busse’s argument goes
together with studies by Weingast and Moran (1983) as well as Keefer and
Stasavage (2000), who argue that the credibility of governments not to
renege on their policy commitments is enhanced by the presence of veto
players. Similar to these works, Hellman, Jones, and Kaufmann (2000) also
contend that levels of state capture can be kept in check by sufficiently devel-
oped civil societies.
Expanding on an earlier literature linking levels of corruption and democ-
racy, Bäck and Hadenius’ (2008) as well as Charron and Lapuente (2010)
propose quantitative demonstrations on a large number of countries using an
improved measurement of state capacity.1 Although the empirical connec-
tions between democracy and state capacity are strong in these works, some
of the theoretical implications of these studies’ findings make it necessary to
question the direction of the causal relationship under study. Both studies
observe a J-shaped or curvilinear relationship between democracy and
administrative performance. As in Grzymala-Busse’s argument, in countries
already above a certain threshold of democracy, further democratization has
positive effects on the state’s administrative capacity. On the flip side, and
more problematic, this reasoning would entail that in highly authoritarian
countries, democratization would serve to reduce state capacity, and in the
case of semiauthoritarian regimes, democratization would then have no effect
on state capacity (Bäck & Hadenius, 2008, pp. 20-21).
To explain these puzzling findings, I propose to look at the reverse causal
mechanism: At low levels of state capacity, democratization is simply less
likely, not so much the opposite. In the face of a curvilinear relationship, it
makes sense to propose a criterion of effective state capacity as a pillar to a

Downloaded from cps.sagepub.com at LUND UNIVERSITY LIBRARIES on December 28, 2012

906 Comparative Political Studies 45(7)

polity’s potential to democratize, implying that some level of capacity is

required for a functioning democracy. If there is no organized and competent
state authority, even highly mobilized citizens cannot possibly influence
policies. In such settings, the state will not be in a position to carry out basic
policies, much less social policies that require even higher state capacity for
effective implementation (Huber et al., 1999). Based on this, I argue that the
likelihood of a state becoming and remaining democratic is considerably
higher when it possesses the means necessary to maintain the distribution of
a basic class of public goods. Although effective state capacity can reason-
ably be considered a necessary condition for democratization, strong states
do not automatically produce democratic regimes, nor do they guarantee their
survival. Far from being a sufficient condition for democracy, a strong or
capable state was also theorized to be indispensable for the maintenance of
stable autocratic rule (Way, 2005); this would explain the curvilinear relation-
ships observed in previous research.
A crucial point of departure in understanding postcommunist trajectories
of reform and democratic performance is the fact that conditions of state
infrastructure were not uniform across the different countries, especially
between Central and Eastern European (CEE) countries and Soviet repub-
lics. In the case of Soviet republics, most of which had never experienced
independent statehood before, the concentration of formal governing insti-
tutions of the Soviet Union in Moscow, in the shape of All-Union Ministries
and Union-Republic Ministries, meant that each one needed to fill a sub-
stantial institutional vacuum in a short period after the dissolution of the
Soviet Union. The disappearance of the Communist Party—the main insti-
tution of oversight—also left a void. In some of the successor republics that
have experienced heavy state engineering, we find intense ethnic strife, con-
flict, territorial divisions, contested territories, and groups such as clans that
can pose a real challenge to the authority of the central government. Georgia
and Tajikistan, for instance, suffered from serious malfunctions at the time of
their emergence, whereas Moldova and Azerbaijan have also struggled to
enforce sovereignty over their secessionist enclaves. Yet these patterns are
not common to all former Soviet republics. Belarus and Ukraine emerged
from the Soviet Union with comparatively more capable state administra-
tions, however burdened with deep-rooted corruption.
Soviet-induced state engineering was less sweeping in Central and Eastern
Europe: Some boundaries were certainly shifted as a result of the conclusion
of World War II, but Moscow did not impose massive changes to facilitate the
establishment of its authority like the carving of states from scratch in Central
Asia.2 Most important, despite the de facto control of the Soviet Union, CEE

Downloaded from cps.sagepub.com at LUND UNIVERSITY LIBRARIES on December 28, 2012

Fortin 907

countries retained local control over the apparatus of the state. The administra-
tion, the treasury, and the courts, despite the application of socialist law every-
where, remained in the hands of national elites. Given their sovereign status,
most institutions of the state such as the military, the police, and core national
ministries, in particular the institutions of taxation, already existed and did not
need to be created anew in 1989. When CEE countries recovered their auton-
omy from Moscow, they faced fewer acute challenges than in the former
Soviet Union, where leaders were confronted with the additional burden of
having to affirm and enforce the newfound authority of the state on their terri-
tory. Although this represented an advantage over most former Soviet repub-
lics, it must nevertheless be stressed that CEE countries were not immediately
consolidated or stable in the 1990s (Grzymala-Busse & Jones Luong, 2002),
nor were they exempt from issues related to limited state capacity (Ganev,
2007). In fact, all eight EU accession candidates had to undergo a fair amount
of state building, understood as the administrative capacity to implement the
acquis, as a condition for admission (Mungiu-Pippidi, 2008).
My contention here is that some of these conditions of state infrastructure
precede, to a certain degree, the kind of liberalization that followed (or did
not follow) the fall of CEE communist parties and the USSR: There was
much variation in the capacity of successor states and, as a result, in the chal-
lenges new, or incumbent, rulers faced in establishing state authority on their
respective sovereign territories. There were generally fewer constraints on
leaders where the capacity of the state was higher: The continuous delivery of
public goods made it less probable that state authority would come under any
serious strain. Leaders’ capabilities for reform were greatest in postcommu-
nist countries where the successor state had the ability to apply a definitive
set of rules early on and, by extension, to build channels of political support.
These conditions would facilitate a power transition that precluded a single
group of actors wanting to secure access to the state’s power resources to lock
in their victory in an unbalanced constitution.
Where the capabilities of the state were more limited after the collapse of
communism, democracy was a more difficult outcome to achieve. In such
cases, leaders tended to compensate for state weakness by enlarging and arrang-
ing their powers vertically. Leaders in times of crisis often claim the nation
needs firm and decisive leadership that stands above the partisan way and can
pursue the national interest more directly, in addition to providing more imme-
diate benefits for society (Smith, 2005). The most illustrative example was
Boris Yeltsin’s refusal to surrender the emergency decree powers he had been
granted in 1990 and his heavy use of such tools between 1992 and 1994 to
enact privatization that benefited a group of insiders. His successor, Vladimir

Downloaded from cps.sagepub.com at LUND UNIVERSITY LIBRARIES on December 28, 2012

908 Comparative Political Studies 45(7)

Putin, also entered the presidential office with the overt goal of strengthening
or consolidating vertical authority, an undertaking that went hand in hand with
an increase in authoritarian practices (Holmes, 2006, p. 301). Facing weak state
infrastructures in the early 1990s, executives in Georgia, Azerbaijan, and
Central Asia also quickly moved to impose a vertical control structure over
regional and local administrations, hence thwarting the rise of credible checks.
In these settings executives were freer to rely on state resources for patronage
and co-optation to gain allies and buy off political opponents.
The establishment of institutions of oversight can have two potential facil-
itators in postcommunist countries. On the one hand, the early establishment
of such institutions can result from “robust competition,” where political par-
ties establish tools to monitor state institutions (Grzymala-Busse, 2007). On
the other hand, I believe the likelihood that institutions of oversight will
emerge, and how binding these institutions are designed to be, is also linked
to existing levels of state capacity. States inheriting corrupt and ineffective
bureaucracies, large informal economies, and a severely impaired ability to
finance their activities will face more difficulties in establishing functioning
agencies to oversee privatization, securities and exchange commissions,
courts staffed with competent personnel, law enforcement, national account-
ing offices, and other such institutions that are crucial to forestall the abuse of
power. Although Grzymala-Busse (2007) argues convincingly that some for-
mer communist states are weak because they failed to implement these insti-
tutions fast enough, if at all, the flip side of the coin is that most such states
were already weak at the onset of independence and faced additional con-
straints in establishing these institutions. It therefore becomes difficult to dis-
entangle a factor that existed previously and is unequivocally casual, as I
demonstrate in the following analyses.

Cases and Measures

To estimate the two corollary hypotheses according to which state capacity
is a necessary condition for democracy and that the establishment of demo-
cratic regimes is more likely in capable states than weaker ones, I employ a
mixture of qualitative and quantitative techniques using the following 26
cases: Albania, Armenia, Azerbaijan, Belarus, Bulgaria, Croatia, Czech
Republic, Estonia, Georgia, Hungary, Kazakhstan, Kyrgyzstan, Latvia,
Lithuania, the Former Yugoslav Republic of Macedonia (FYROM), Moldova,
Mongolia, Poland, Romania, Russia, Slovak Republic, Slovenia, Tajikistan,
Turkmenistan, Ukraine, and Uzbekistan. The coupling of qualitative and
quantitative tests allows me to tackle the issue of complex causality by har-
nessing the strength of different approaches to causation.

Downloaded from cps.sagepub.com at LUND UNIVERSITY LIBRARIES on December 28, 2012

Fortin 909

State Capacity
I propose to capture the concept of state capacity via proxy of the quality of
the provision of public goods given their crucial role in establishing state
authority on territories (Spruyt, 1994). This conceptualization is anchored in
a minimal definition of the concept of state capacity, solely in terms of infra-
structural power. Here, infrastructural power is defined as “the institutional
capacity of a central state, despotic or not, to penetrate its territories and
logistically implement decisions” (Mann, 1993, p. 59). This understanding of
state capacity is operationalized by examining the quality of provision of a
class of collective goods even a minimally redistributive state should theo-
retically provide, as proposed by recent work (Fortin, 2010). Rather than
focusing on the measurement of absolute resources, this approach assumes
that it is insufficient that a state should possess important human and natural
resources for the production of wealth: It must also be able to effectively
make use of those resources. As well, this intentionally leaves aside the des-
potic or coercive power of states since such aspects of power do not neces-
sarily go hand in hand with infrastructural power.
The index used in this article is based on the aggregation of five indica-
tors, rather than the single or two-indicator measures proposed in previous
studies. The indicators retained are progress in infrastructure reform
European Bank for Reconstruction and Development(EBRD), levels of
corruption (Heritage Foundation), the quality of property rights protection
(Heritage Foundation), and the ratio of Contract Intensive Money (CIM),
calculated from International Monetary Fund financial statistics. Last but
not least, the delivery of such public goods depends on the crucial factor of
funding. Thus, a state’s ability to collect taxes will be paramount to the
delivery of public goods and consequently a core measure of state capacity.
I use the ratio of tax revenue to GDP to illustrate state taxing capacity. To
make the final index, all indicators were standardized and combined to be
used as a single measure. Factor and correlation analyses revealed that the
association among the items is strong, that they measure a similar latent
construct, and therefore that the combination of these items in a composite
index is the best option in the present case.3

Regime Outcome
Most observers agree that no single index of democracy offers an adequate
solution to all three challenges of conceptualization, measurement, and
aggregation, accompanying the transformation of an abstract concept into a

Downloaded from cps.sagepub.com at LUND UNIVERSITY LIBRARIES on December 28, 2012

910 Comparative Political Studies 45(7)

measurable indicator (Munck & Verkuilen, 2002, p. 28). However what ulti-
mately constrains researchers in choosing a quantitative measurement of
democracy is data availability. Since only two studies offer coverage from
1989 continuously until 2006, the dependent variable used in the present
study is an additive combination of Freedom House (2008) scores and the
Polity IV index (Marshall & Jaggers, 2009) in a single composite measure.4
By grouping the two indices, I hope to maximize both the validity and the
reliability of the composite index since Freedom House uses a maximalist
definition of democracy and Polity a more restrictive one. In particular,
Freedom House (2008) considers characteristics such as socioeconomic
rights, freedom from gross socioeconomic inequalities, and freedom from
war. Polity’s coding of democracy is based on an assessment of the com-
petitiveness of political participation, the regulation of political participation,
the competitiveness and openness of executive recruitment, and the con-
straints placed on the chief executive. Therefore, combining both indices will
serve to alleviate problems of systematic and random measurement errors
undoubtedly present when the two measures are considered separately.

State Capacity as a Necessary

Condition for Democracy
The first step in testing the hypothesis of state capacity as a necessary condi-
tion for the establishment of democracy consists in visually inspecting the
graphical representation of both variables of interest. Figure 1 presents aver-
age values of state capacity during the period 1989–2006 with combined
Freedom House scores in 2006. The two vertical lines represent Freedom
House’s classification cutoffs (respectively, free, partially free, and not free),
whereas the horizontal line simply is the mean of state capacity over time.
The distribution of cases in Figure 1 offers clear evidence (Pearson’s r =
–.80) in favor of the state capacity hypothesis as most cases, except FYROM,
fall neatly within three categories. As expected, the relationship between the
two variables of interest is not straightforward but slightly curvilinear as
illustrated by the locally weighted scatterplot smoothing (lowess) smooth
regression line. The most important finding is that nearly all regimes classi-
fied as “free” by Freedom House are located in the upper-left cell of the
graph. Conversely, all authoritarian and semiauthoritarian regimes are posi-
tioned in the lower half of the state capacity quadrant.
Figure 1 only provides a partial confirmation of Lucan Way’s (2005) claim
that “key elements of a strong state are critical for maintaining nondemo-
cratic rule” (p. 235). Belarus and Uzbekistan, the two most repressive

Downloaded from cps.sagepub.com at LUND UNIVERSITY LIBRARIES on December 28, 2012

Fortin 911

Figure 1. Average state capacity 1989–2006 and democracy in 2006

regimes, indeed display higher state capacity scores than all other autocra-
cies. Yet it is crucial to notice that both exhibit much lower scores than
democracies and that a highly repressive regime, Turkmenistan, displays a
low overall state capacity score. Although democracies are neatly clustered in
the upper-left corner of the graph, we notice that variability in state capacity
scores increases as we move toward the more authoritarian end of the spec-
trum. The direction of the lowess regression line changes in the bottom right
cells: State capacity becomes weakly associated with more repressive regimes
but with too many outliers to offer unambiguous evidence.
Technically, FYROM’s positioning in the top-middle cell could require
discarding the necessary causation hypothesis altogether since in many nec-
essary causation tests, we should reject the theory on the discovery of a single
counterexample. However, such a strict approach to testing for necessary

Downloaded from cps.sagepub.com at LUND UNIVERSITY LIBRARIES on December 28, 2012

912 Comparative Political Studies 45(7)

conditions is likely counterproductive in the present situation, not only

because of the arbitrariness of the position of category lines but also because
of possible measurement errors and the fact that these figures correspond to
averages over time. The arbitrariness of the horizontal line has two potential
consequences for necessary causation testing. Given their median position in
Figure 1, Romania, FYROM, Moldova, Ukraine, and Belarus could be
grouped either with the more capable states or with the less capable states, a
problem that often arises when continuous concepts are dichotomized and
assigned a discrete boundary (Braumoeller & Goertz, 2000). The same argu-
ment also undeniably applies to the classification of democracies and nonde-
mocracies; it is not clear where the border of demarcation between
democracies and nondemocracies should be located or whether the establish-
ment of a truly discrete empirical cutoff is even possible.
Table 1 presents the results from Figure 1 in a four-cell table to test for
necessary or sufficient causes. Observation of Table 1 reveals, unsurprisingly,
that the data do not satisfy the sufficient causation requisites. In particular, for
state capacity to be a sufficient cause of democracy, we should find some
cases located in cell A whereas no cases should be present in cell C (cells B
and D are not considered in this design). Because authoritarian regimes can
also be a plausible outcome in strong states, state capacity is not a sufficient
cause for democracy. Using the positive outcome design, where the analyst
focuses only on cases where the outcome variable is present (Braumoeller &
Goertz, 2000; Dion, 1998; Ragin, 1998, 2000) to test for necessary condi-
tions, we observe that democracy has occurred only in countries displaying a
certain level of state capacity. The presence of variation in both independent
and dependent variables further indicates that state capacity is a nontrivial
necessary condition to democracy.
According to rules of the All Cases Design (Seewright, 2002) for estab-
lishing necessary causation, all valid cases should be distributed in cells A, C,
and D, whereas no cases should be found in cell B: The pattern outlined in
Table 1 satisfies these requirements. Using simple models of Bayesian infer-
ence where we set prior probabilities that the necessary condition is true, the
results from these data are consistent with the necessary causation pattern. In
a model where we have no reason to believe that either hypothesis is more
likely to be confirmed, the confidence level that the present data are consis-
tent with the necessary causation pattern is 97%.5 In other words, we can be
confident with a level of 97% that the data are generated in a pattern that is
consistent with necessary causation even in the face of conservative priors.6
Further confirmation of these findings will be established in the following
section using TSCS estimations.

Downloaded from cps.sagepub.com at LUND UNIVERSITY LIBRARIES on December 28, 2012

Fortin 913

Table 1. Necessary and Sufficient Cause Design

Democracy Cell A (12) Cell B (0)

Bulgaria
Czech Republic
Croatia
Estonia
Hungary
Latvia
Lithuania
Mongolia
Poland
Romania
Slovakia
Slovenia
Nondemocracy Cell C (1) Cell D (13)
FYROM Albania
Armenia
Azerbaijan
Belarus
Georgia
Kazakhstan
Kyrgyzstan
Moldova
Russia
Tajikistan
Turkmenistan
Ukraine
Uzbekistan
Stronger states Weaker states

Time-Series Cross-Sectional Analyses

Pooled time-series and cross-sectional (TSCS) estimations are sensible
when cross-national and temporally comparable quantitative indicators are
only available for a short period and for a small group of countries (fixed,
not sampled). Although it offers clear sample size advantages, the pooling
of time series from a number of cross-sectional units can increase estimation
difficulties. While the data are not temporally dominant, the nature of the

Downloaded from cps.sagepub.com at LUND UNIVERSITY LIBRARIES on December 28, 2012

914 Comparative Political Studies 45(7)

units and the arguably endogenous variables all point toward probable het-
eroscedasticity. When time-series data are pooled from geographic entities,
cross-sectionally correlated errors can become more problematic (Kmenta,
1971). More precisely, the variance of error processes likely differs from
unit to unit. To reduce the risks associated with overconfidence in the per-
formance of estimators, the following analyses are performed using TSCS
models of linear regression with panel-corrected standard errors (Beck &
Katz, 1995).
In the following analyses, I estimate three models using varying specifi-
cations to test the robustness of the state capacity hypothesis. First, I esti-
mate a base model of regime outcome containing regressors of state
capacity, levels of development captured by GDP per capita, and the role of
the European Union, crudely measured as the distance of each country’s
capital to Brussels. To curb the problem of spurious correlations arising
when certain values of the dependent variable vary independently, but in
the same consistent direction over time, I include a trend variable. This
trend variable, the number of years elapsed since independence from com-
munism, will also serve to capture effects of maturation and regime devel-
opment that can account for some extent of variance in regime outcomes.
Following this initial setup, I then add a lagged value of the dependent
variable (LDV) in a second model to control for autocorrelation. Third and
last, I include a series of country dummy variables for each unit of the
pooled model. TSCS equations suppose that countries are homogeneous,
that is, that they differ only in levels of explanatory variables they possess,
an unreasonable assumption in the case at hand. The least square dummy
variable estimators (LSDV) will account for the possibility of intercept dif-
ferences across units but also for variance from potentially influential vari-
ables that were left out of the model (Hicks, 1994; Judge, Griffiths, Carter
Hill, Lutkepohl, & Lee, 1985).7 This last model specification should pro-
vide the strongest test of the robustness of the coefficient of state capacity
in the face of an overspecified model.

Data Imputation
Despite being high in validity and reliability, the index of state capacity I
employ as an independent variable in the following models suffers from an
important weakness: Missing data in the early years of transition. On the
potential 432 cases the present analyses could benefit from, a mere 263
remain when full models are estimated, even if only 5% of cases are missing.
The majority of missing cases is distributed in the extremities of the time

Downloaded from cps.sagepub.com at LUND UNIVERSITY LIBRARIES on December 28, 2012

Fortin 915

variable and in the least advanced countries, and they are therefore not miss-
ing at random. Here, listwise deletion is a substandard alternative for three
reasons. First, listwise deletion would be especially damaging given the
distribution of the missing cases in the first year of independence since this
would remove much crucial information about the early 1990s from the final
analyses and, consequently, render it impossible to perform appropriate tests
for the hypothesis. Second, the concentration of missing cases in Central
Asia and the Caucasus would undoubtedly introduce systematic bias in esti-
mating the effects of low state capacity on outcome variables. Last, because
TSCS requires a minimum of continuous panel information to produce reli-
able estimates, listwise deletion would reduce country observations and
introduce gaps between years in a way that is insufficient to produce reliable
estimates.8
To mitigate the problem of missing data, I chose to impute a portion of the
cases (for tax revenue and contract intensive money) using (Honaker, King,
& Blackwell, 2009; King, Honaker & Scheve, 2001), the best option avail-
able to generate plausible and unbiased results (Allison, 2000).9 In the case of
corruption levels, and the quality of private property protection, I chose a
more conservative data interpolation technique where starting values are
similar to the first year of actual data coverage by Heritage Foundation (in
most cases around 1995). The hypothesis underlying the interpolation is that
in no case were the scores for these indicators higher in early years of transi-
tion: They were likely lower, or the same as a few years after, at best. The
final index of state capacity therefore contains both actual measurements and
estimated measurements of individual indicators and should therefore be
handled and interpreted with a measure of prudence in the very early years.
With this, I believe the advantages of using estimates produced from this
approximate information clearly outweigh the option of dropping nonrandom
cases.

Findings
Estimation results of the baseline state capacity model are presented in the
first column of Table 2, whereas the two alternative models, one including an
LDV variable as well as a third containing LSDV estimators, are displayed
in the last two columns. For all models presented, the upper section of the
table represents estimates of the parameters of the causal factors included in
the regression. Where applicable, the middle section of the table represents
nation-specific intercepts, net of the effects of the other independent vari-
ables. Since I have suppressed the constant in the model concerned (Model

Downloaded from cps.sagepub.com at LUND UNIVERSITY LIBRARIES on December 28, 2012

916 Comparative Political Studies 45(7)

Table 2. Time-Series Cross-Sectional Analyses of the Effect of State Capacity on

Regime Outcomes

Model 1: Baseline Model 2: LDV Model 3: LSDV

Independent
variable OLS est. PCSE OLS est. PCSE OLS est. PCSE
Regime 0.789*** (0.043) 0.441*** (0.044)
outcomet-1
Time elapsed 0.030*** (0.009) –0.009*** (0.004) 0.002 (0.004)
GDP per 0.179*** (0.046) 0.001 (0.016) 0.148*** (0.036)
capita
Distance to −0.315*** (0.046) –0.049*** (0.016)
EU
State capacity 0.257*** (0.088) 0.206*** (0.036) 0.068* (0.039)
Albania −0.388 (0.384)
Armenia −0.520 (0.361)
Azerbaijan −1.134*** (0.371)
Belarus −1.034*** (0.207)
Bulgaria 0.712*** (0.394)
Croatia −0.635 (0.224)
Czech −0.074 (0.454)
Republic
Estonia −0.295 (0.438)
Georgia −0.376 (0.358)
Hungary −0.105 (0.441)
Kazakhstan −1.140*** (0.396)
Kyrgyz −0.740** (0.325)
Republic
Latvia −0.166 (0.432)
Lithuania −0.037 (0.437)
FYROM −0.345 (0.402)
Moldova −0.280 (0.320)
Mongolia −0.027 (0.321)
Poland 0.130 (0.438)
Romania −0.236 (0.402)
Slovak −0.144 (0.438)
Republic
Slovenia −0.112 (0.486)
Tajikistan −1.026*** (0.309)
Turkmenistan −1.485*** (0.377)

(continued)

Downloaded from cps.sagepub.com at LUND UNIVERSITY LIBRARIES on December 28, 2012

Fortin 917

Table 2. (continued)

Model 1: Baseline Model 2: LDV Model 3: LSDV

Independent
variable OLS est. PCSE OLS est. PCSE OLS est. PCSE
Ukraine −0.349 (0.362)
Uzbekistan −1.455*** (0.355)
Constant 0.183** (0.096) 0.198*** (0.040)
N 432 406 406
Number of 26 26 26
countries
Obs. per 16.62 15.62 15.62
group avg.
Rho .75 .27 .22
Adj. R2 .28 .88 .95

OLS = ordinary least squares; PCSE = panel-corrected standard error; LDV = lagged depen-
dent variable; LSDV = least square dummy variable. Table entries are unstandardized regression
coefficients from a time-series cross-sectional analysis using PCSEs (in parentheses) performed
on five imputed data sets using Kenneth Scheve’s “miest” program in Stata 10.0. Rho and R2
values represent the average of five models ran separately.
*p < .10. **p < .05. ***p < .01.

3), these represent the actual intercepts of the groups and can be directly
interpreted as such. These additions allow for slope heterogeneity of state
capacity effects across nations. Last, the lower section of the table displays
the goodness-of-fit and residual statistics.
The baseline model (Model 1) in the first column of Table 2 provides
unambiguous support for the principal proposition regarding the importance
of state capacity for democracy: The parameter estimate for state capacity is
positive and significant. Also as expected, the coefficient acting as a proxy
for EU influence is negative and significant. As well, the parameter estimate
for the time elapsed is also positive and significant, provisionally suggesting
that countries are becoming more democratic over time. Last, the coefficient
depicting level of development (GDP/capita) is also significant and positive.
The overall fit of the model is moderate, with an (average) adjusted R2 of .28,
leaving ample space for model improvement.
A LDV was added to the baseline model in Model 2 to alleviate the high
level of serial correlation of the errors visible in Model 1 (Beck & Katz,
1996).10 Because of the real causal impact that past values of democracy

Downloaded from cps.sagepub.com at LUND UNIVERSITY LIBRARIES on December 28, 2012

918 Comparative Political Studies 45(7)

have on subsequent scores, the inclusion of such an item is theoretically

justified. In addition, because a LDV also serves as a proxy by picking up
some of the unmeasured variables, its addition is also empirically sound
(Burkhart & Lewis-Beck, 1994). As it was to be expected with a model
presenting some amount of serial autocorrelation, the LDV became the
dominant explanatory factor in Model 2, absorbing much of the strength of
the other variables and dramatically improving the adjusted R2 (Achen,
2000). However, even with the inclusion of this autoregressive term, the
parameter estimate for state capacity remains statistically significant,
despite underestimated effects attributable to the LDV. Yet the inclusion of
the LDV has considerable impacts on the other independent variables: The
parameter estimate for GDP per capita loses statistical significance, whereas
the sign of the parameter estimate for the variable depicting trend is
reversed.
Results from Models 1 and 2 confirm previous research concerning the
role of the EU being strongly associated with positive democratic progress
(Cameron, 2007; Dimitrovna & Pridham, 2004; Grzymala-Busse & Jones
Luong, 2002; Kopstein & Reilly, 2003; Risse, Green Cowles, & Caporaso,
2001; Vachudova, 2005, 2010).11 Although I recognize these important con-
tributions as well as the value of adding this information to avoid omitted
variable bias, “proximity to the EU” measured in distance has limited validity
and raises issues of multicollinearity. The strong association between each
country dummy and distance to Brussels suggests that this item encapsulates
many other unobserved phenomena that are specific to each country and not
necessarily related to the EU. Given the crudeness of the measurement, not-
withstanding the laudable theoretical underpinnings of existing research, I
would advise readers to employ a conservative approach to interpreting these
parameter estimates as purely depicting the effects of the EU’s leverage on
regime outcomes.
Since Models 1 and 2 do not necessarily contain all the variables influenc-
ing regime outcomes, the addition of country-specific variables in Model 3
serves to alleviate remaining concerns of model heterogeneity and doubles as
a two-way fixed effects model.12 Overall, Model 3 explains close to 95% of
the linear variation in regime outcomes in postcommunist countries, which
indicates that models that integrate state capacity, levels of development, and
country-specific variables manage an almost perfect fit to the data and would
unlikely benefit from additional variables.13 Given that parameter estimates
from TSCS models encompass the combined average partial effect of the
cross-section dimension and time, their substantive meaning is less readily
interpretable than in traditional cross-section models (Firebaugh, 1980;

Downloaded from cps.sagepub.com at LUND UNIVERSITY LIBRARIES on December 28, 2012

Fortin 919

Kittel, 1999). Although it is impossible to determine the exact magnitude of

the relative contribution of each dimension to the parameter estimates, the
robustness of the state capacity variable in the models holding country effects
constant can be considered as evidence of developments over time that are
common to all cases (Kittel & Winner, 2005).
A surprising finding concerns the coefficients for the trend variable which
are unstable between the three models. The positive trend noted in Model 1
reverses on the addition of the LDV (Model 2) and is too insubstantial to
remain statistically significant when country dummy variables are added
(Model 3). It appears then, all things being equal, that although postcommu-
nist regimes have tended to become more democratic over time when we
look at overall democracy scores, this improvement is not the result of time
itself but rather changes in the other factors included in the fully specified
model. The weakness of the variable representing time in the face of a fully
specified model suggests that democratic development was neither unidirec-
tional nor automatic in the region.
In an article reviewing the evidence in favor of modernization theory
(Lipset, 1959), Burkhart and Lewis-Beck (1994) demonstrated that economic
development alone accounted for more variance in democracy than any of the
other variables they included in their model, a finding that echoed many ear-
lier studies on the topic (Bollen, 1979; Bollen & Jackman, 1985; Brunk,
Caldeira, & Lewis-Beck, 1987; Jackman, 1973).14 The present study nearly
replicates these findings. The variable depicting levels of development is sta-
tistically significant in the baseline and fully specified models, even in the
face of very high multicollinearity, although it temporarily loses statistical
significance in the presence of a LDV in the model (Model 2).15 Despite the
strong association between state capacity and GDP per capita,16 and the prob-
lematic multicollinearity brought by the LSDV, both variables retain statisti-
cal significance in the flooded model (Model 3), allowing us to disentangle
some independent effects of state capacity and levels of development on
regime outcomes. Under such conditions, the fact that the coefficient depict-
ing levels of state capacity retains some explanatory power, even in an over-
specified model, demonstrates beyond much doubt that it is one of the key
variables explaining regime outcomes.

A Further Exploration of Causality

On the whole, and over time, the index of state capacity is a strong predictor
of levels of democracy. However, this close connection raises a last, but
nontrivial, methodological concern over the direction of causality. The issue

Downloaded from cps.sagepub.com at LUND UNIVERSITY LIBRARIES on December 28, 2012

920 Comparative Political Studies 45(7)

Table 3. Results From Granger Causality Testing

Dependent: Democracy Dependent: State capacity

Variable Coefficient t Variable Coefficient t

Democracyt-1 0.88*** 45.2 Democracyt-1 0.05*** 3.74
State capacityt-1 0.15*** 6.23 State capacityt-1 0.91*** 48.23
Democracyt-2 0.84*** 32.36 Democracyt-2 0.08*** 4.13
State capacityt-2 0.20*** 6.41 State capacityt-2 0.87*** 36.71

Cells contain results from four ordinary least squares regressions performed on two different
dependent variables. Regressions for 1-year lags were performed separately from 2-year lags.
Estimates were obtained from one of five imputed data sets. Nevertheless, the same estima-
tions performed in the other four data sets allow very close replication of the above table.
***p < .01.

is particularly thorny here because of the impossibility of establishing an

experimental design to isolate time order of exposure to the independent
variable, thus unequivocally demonstrating directional causality. One alter-
native way of establishing the direction of causation is to perform Granger
causality tests (Granger, 1969). Granger causality evaluates only whether a
phenomenon happens before another and helps predict it, but it does not
represent the concept causality in any other deeper theoretical sense
(Granger, 1980). This test is conducted by regressing the dependent variable
(yt) on lagged values of yt-k and an independent variable (xt-k). The null
hypothesis is that xt-k does not Granger cause (g-cause) yt: This null hypoth-
esis can be rejected when one or more of the lagged values of x are signifi-
cant. Table 3 presents the results from the estimations performed according
to the above specifications.
In the present case, results indicate that state capacity (xt-k) g-causes
democracy (yt) but that democracy (yt-k) also g-causes state capacity (xt);17 we
are facing a feedback stochastic system where causality is not unidirectional
and values from both variables help predict future values in the other.18
Substantially, the causal arrow between state capacity and democracy is
likely to run in both directions. Feedback effects are undoubtedly present. As
demonstrated by Grzymala-Busse (2007), the causal arrow also runs from
democracy or party competition to state capacity. In her analysis of postcom-
munist state reconstruction, Grzymala-Busse showed that opportunistic
behavior by ruling parties was curbed in the presence of robust political
oppositions, a point similar to Hellman et al.’s (2000) finding that levels of
state capture can be kept in check by sufficiently developed civil societies.

Downloaded from cps.sagepub.com at LUND UNIVERSITY LIBRARIES on December 28, 2012

Fortin 921

Thus, the building of formal state institutions of monitoring and oversight

was a more likely outcome in the most competitive systems. Conversely,
unchecked governing coalitions have tended to engage in more resource
exploitative, clientelistic, and rent-seeking behavior. Consequently, ruling
parties in uncompetitive settings fashioned state institutions that facilitated
such predatory behavior by thwarting the development of formal institutions
of monitoring and oversight.

Discussion and Conclusions

The view I proposed in this article is essentially state centric, not, however,
because I believe other factors are less relevant to explain postcommunist
democratization. The role of civil society, formal political institutions, val-
ues held by key elites and elite constellations, feelings of national unity and
ethnic strife, levels of development, contagion effects, political parties, and
other institutions have purposely been relegated to the background to isolate
the significance of a single factor that had not yet been submitted to com-
parative empirical verification. Remembering that Huntington (1991) listed
no fewer than 27 variables in his account of how democratic regimes
emerge and consolidate, state capacity can be only one among many factors.
The demonstrations conducted in this article allow us to reach three basic
conclusions about the importance of state capacity in explaining regime out-
comes. First, the qualitative methodology of necessary and sufficient causa-
tion provides strong evidence that a certain level of state capacity appears to
be a necessary condition for democracy. The scope of the challenge postcom-
munist countries faced after the demise of state socialism was extremely
broad. It involved economic transformation, democratic transition, state
reconstruction, the resurgence of civil society, the formation of political par-
ties, a complete reorientation of foreign and security policies and, for some,
even a reconfiguration of national identities (Sakwa, 1999). As the results
presented in this article make clear, some countries were more or less suc-
cessful at implementing these changes because the capacity of the state in the
early years of transition, its strength or weakness, has set the limits of reform
agendas that they could reasonably undertake (Roberts & Sherlock, 1999).
The varying capacities of states in the early years after their release from
communism gave way to different constraints on policy makers about the
available options for conducting both economic and political reforms.
Second, multivariate quantitative analyses also demonstrate the existence
of a strong and clear empirical relationship between state capacity and
democracy among postcommunist countries over time. In that regard, this

Downloaded from cps.sagepub.com at LUND UNIVERSITY LIBRARIES on December 28, 2012

922 Comparative Political Studies 45(7)

article’s findings run counter to Remington’s (2006) assertion that young

democracies should perform worst at protecting property and contractual
rights. All the democratic regimes contained in this study are nested within
the strongest states, to be exact, the best protectors of property and contract
rights. Conversely, there is greater variability of state capacity in semiauto-
cratic and authoritarian states. Stronger states are not linearly associated with
better democracy scores in such categories, nor do young autocracies per-
form better at protecting property rights and enforcing contracts.
The state “is crucial in constituting social order, in enabling regular and
peaceful private relations among groups and individuals” (Przeworski, 1995,
p. 110) because it makes interactions among groups and individuals more
predictable. Consequently, the less capacity a state has at its disposal, the
more difficult it becomes to perform the tasks associated with modern state-
hood (Easter, 2002). When the institutions of the state are not able to enforce
rights and obligations, the state’s claims on the monopoly of violence and
resources can be more easily challenged. Furthermore, when the state cannot
fulfill its obligations, individuals turn to alternative channels to satisfy their
needs, thus explaining the rise of private protection rackets in Russia to cope
for the state’s inability to guarantee security for individuals in the early 1990s
(Volkov, 2002). Therefore, in states where leaders most successfully accom-
plish the tasks associated with statehood, the introduction of a rational tax
collection administration, infrastructure reform, keeping corruption under
control, and protecting property rights, democracy levels are generally the
highest. However, unlike the findings of Cheibub (1998), who hypothesized
that democracies are more likely than authoritarian regimes to assemble the
conditions necessary for an effective tax system, the present findings suggest
that the causal arrow might point the other way as well: Democracy is more
likely to have emerged in countries with the most effective tax systems within
a few years from independence.
Related to this point, the third and final conclusion is that the direction
of the causality between these variables cannot be established in only one
direction, which indicates that state capacity and levels of democracy have
mutually constitutive relationships. In other words, strong state capacity is
associated with democracy, whereas high democratic performance is also
driving higher state capacity scores. Although Remington (2006) argues
that “democratization does not necessarily improve institutional capacity”
(p. 289), results from this group of countries show that high levels of democ-
racy did in fact have positive impacts on capacity. Nevertheless, I concur
with Remington that a weak state is more susceptible to capture and corrup-
tion: The nuance I want to bring is that a weak state most likely will not be

Downloaded from cps.sagepub.com at LUND UNIVERSITY LIBRARIES on December 28, 2012

Fortin 923

democratic. In the end, it is perhaps reassuring that the direction of causality

cannot be established beyond doubt simply through statistical modeling;
social reality is more disorderly than the simplified representations these
models offer. Because the statistical tests performed in this article fall short
of establishing clear temporally causal relationships between variables of
interest, further examination through detailed qualitative case studies would
likely be necessary to unearth the deeper causal underpinnings behind these
empirical relationships.

Declaration of Conflicting Interests

The author declared no potential conflicts of interest with respect to the research,
authorship, and/or publication of this article.

Funding
The author disclosed receipt of the following financial support for the research,
authorship, and/or publication of this article by a postdoctoral grant from the Fond
Québécois de Recherche sur la Société et la Culture and conducted at the Center for
the Study of Democracy at Leuphana University of Lüneburg.

Notes
1. Both Bäck and Hadenius (2008) and Charron and Lapuente (2010) use a two-
item composite index of state capacity combining “bureaucratic quality” and
“level of perceived corruption” from Political Risk Services’ International Coun-
try Risk Guide data.
2. Central and Eastern European countries also underwent radical border modi-
fications at the end of the Second World War. With the Treaty of Paris of
1947, changes were enacted between the Hungarian–Slovak and Romanian–
Hungarian borders (Transylvania was given back to Romania from Hungary,
where it remained only a few years after the Ribbentrop–Molotov Pact). With
the Yalta Accords, Poland saw its borders shifted more than 200 kilometers to
the west, so that large amounts of territory were redistributed among Ukraine,
Belarus and Lithuania, whereas its western boundaries were also relocated far-
ther to the west, taking in former German territory on the Oder–Neisse Line.
3. For an in-depth discussion of index construction, particularly the issues of valid-
ity and reliability, see Fortin (2010).
4. In this study, both measures were transformed to vary in the same direction, from
autocracy to democracy, and standardized for increments to be on the same scale.
Pearson’s r between both indices is .9, and Cronbach’s α is .95.
5. When both the working and the alternative hypotheses are set at 50%.

Downloaded from cps.sagepub.com at LUND UNIVERSITY LIBRARIES on December 28, 2012

924 Comparative Political Studies 45(7)

6. This posterior probability given the data was calculated using the All Cases
Design P (WH|D) = P (WH) / (P (WH) + P (AH) * 1 / (a + c + d + 3)).
7. An F test indicated that the null hypothesis of no effects should be rejected. The
addition of country dummy variables should account for the remaining possibility
of intercept differences across units. It is important to note that the addition of these
variables can bias downward the coefficients of those variables whose effects are
partly cross-sectional.
8. Although no strict minimum rule exists for T, Nathaniel Beck (2001, p. 274)
cautions practitioners to be suspicious when time-series cross-sectional (TSCS)
methods are used when T < 10, although the rule can be more flexible for the
number of cross-sectional observations.
9. Five completed data sets (in the form of TSCS) were created, thus imputing miss-
ing data five times, all using independent draws, in an attempt to approximate the
true distributional relationship between these missing data and the information
that is already present in the data set. About 25% of values were missing for tax
revenue and contract intensive money at t = 0, whereas practically no values
were missing after the first year.
10. To detect serially correlated errors, a Lagrange multiplier test was performed.
This was accomplished by an ordinary least squares (OLS) regression equation
of the fully specified model and then regressing the residuals on all the inde-
pendent variables and the lagged residuals. Since the coefficient on the lagged
residual was statistically significant, I had to reject the null hypothesis of inde-
pendent errors. From this perspective, it becomes safest to assume that we are in
the presence of first-order autocorrelation AR(1).
11. Although some question the linearity of these findings (Mungiu-Pippidi, 2010).
12. These were partially addressed by the addition of a lagged dependent variable.
Because of almost perfect collinearity, the measure of distance from each coun-
try’s capital to Brussels cannot be used together with the least square dummy
variable and is not included in Model 3.
13. Further confirming these results, a cross-sectional OLS regression (not shown)
explaining the difference in levels of democracy between t = 0 and 2005 with
the level of democracy at t = 0 and difference in state capacity between t = 0
and 2005 yielded an adjusted R2 of .41. With the addition of GDP per capita, the
adjusted R2 rose to .48. In all models, the variable depicting the change in state
capacity scores remained statistically significant.
14. However not all agree that modernization is the principal explanatory variable
for levels of democracy; see Arat (1988) and Gonick and Rosh (1988).
15. The variance inflation factor used to quantify the severity of multicollinearity is
highest for GDP per capita in Model 3 (24.19). Also, the large condition number
(indicator of the global instability of regression coefficients) of 17, indicates

Downloaded from cps.sagepub.com at LUND UNIVERSITY LIBRARIES on December 28, 2012

Fortin 925

many collinear relations among the regressors of Model 3 (Fox, 1997, pp. 350-
351). However, even in the face of multicollinearity, OLS estimates remain unbi-
ased (Berry & Feldman, 1985). Hence, it follows that the inflated standard errors
should typically lead us to underestimate the significance of the estimators, not
overestimate them, lending additional confidence to the results presented in
Table 2.
16. Pearson’s r between the two variables is .77.
17. These tests were conducted on all five imputed data sets. Similar results were
obtained for all data sets with significant F test results up to two lags.
18. Taken at face value, the results of the Granger causality tests involving state
capacity seem disappointing. One way to interpret these findings is that Granger
causality is limited to linear change. Since we already know from graphical evi-
dence that the relationship between our variables of interest is not linear, it is
probable that the tests overlook much of the nonlinear variance at work. Second,
this test can be performed only on pairs of variables and may produce misleading
results when more than two variables are known to have an impact, such as in
the present case. Last, our measurements are imperfect depictions of broad and
abstract concepts. We cannot hope to achieve the same kind of precision with
such crude measures as with detailed economic data gathered on a weekly basis
over long periods of time.

References
Achen, C. H. (2000, August). Why lagged dependent variables can suppress the
explanatory power of other independent variables. Paper presented at the annual
meeting of the Political Methodology Section of the American Political Science
Association, Los Angeles, CA.
Allison, P. D. (2000). Multiple imputation for missing data: A cautionary tale. Socio-
logical Methods and Research, 28, 301-309.
Arat, Z. F. (1988). Democracy and economic development: Modernization theory
revisited. Comparative Politics, 21, 21-36.
Bäck, H., & Hadenius, A. (2008). Democracy and state capacity: Exploring a J-shaped
relationship. Governance, 21, 1-24.
Beck, N. (2001). Time-series cross-section data: What have we learned in the past few
years? Annual Review of Political Science, 4, 271-293.
Beck, N., & Katz, J. N. (1995). What to do (and not to do) with time-series cross-
section data. American Political Science Review, 89, 634-647.
Beck, N., & Katz, J. N. (1996). Nuisance vs. substance: Specifying and estimating
time-series cross-section models. Political Analysis, 6, 1-36.
Berry, W. D., & Feldman, S. (1985). Multiple regression in practice. Beverly Hills,
CA: SAGE.

Downloaded from cps.sagepub.com at LUND UNIVERSITY LIBRARIES on December 28, 2012

926 Comparative Political Studies 45(7)

Bollen, K. A. (1979). Political democracy and the timing of development. American

Sociological Review, 44, 572-587.
Bollen, K. A., & Jackman, R. W. (1985). Political democracy and the size distribution
income. American Sociological Review, 50, 438-457.
Braumoeller, B. F., & Goertz, G. (2000). The methodology of necessary conditions.
American Journal of Political Science, 44, 844-858.
Brunk, G. G., Caldeira, G. A., & Lewis-Beck, M. S. (1987). Capitalism, socialism,
and democracy: An empirical inquiry. European Journal of Political Research,
15, 459-470.
Bunce, V. (2000). Comparative democratization: Big and bounded generalizations.
Comparative Political Studies, 33, 703-734.
Burkhart, R. E., & Lewis-Beck, M. S. (1994). Comparative democracy: The eco-
nomic development thesis. American Political Science Review, 88, 903-910.
Cameron, D. R. (2007). Post-communist democracy: The impact of the European
Union. Post-Soviet Affairs, 23, 185-217.
Charron, N., & Lapuente, V. (2010). Does democracy produce quality of government?
European Journal of Political Research, 49, 443-470.
Cheibub, J. A. (1998). Political regimes and the extractive capacity of governments:
Taxation in democracies and dictatorships. World Politics, 50, 349-376.
Dimitrovna, A., & Pridham, G. (2004). International actors and democracy promotion
in Central and Eastern Europe: The integration model and its limits. Democratiza-
tion, 11(5), 91-112.
Dion, D. (1998). Evidence and inference in the comparative case study. Comparative
Politics, 30, 127-145.
Easter, G. M. (2002). The politics of revenue extraction in post-communist states:
Poland and Russia compared. Politics & Society, 30, 599-627.
Firebaugh, G. (1980). Cross-national versus historical regression models: Conditions
of equivalence in comparative research. In R. F. Tomasson (Ed.), Comparative
social research (Vol. 3, pp. 333-344). Greenwich, CT: JAI.
Fortin, J. (2010). A tool to evaluate state capacity in post-communist countries,
1989–2006. European Journal of Political Research, 49, 654-686.
Fox, J. (1997). Applied regression analysis, linear models, and related methods.
Thousand Oaks, CA: SAGE.
Freedom House. (2008). Annual survey of Freedom House country scores, 1972–73
to 2007. Washington, DC: Author.
Fukuyama, F. (2004). State-building: Governance and world order in the 21st cen-
tury. Ithaca, NY: Cornell University Press.
Ganev, V. I. (2007). Preying on the state: The transformation of Bulgaria after 1989.
Ithaca, NY: Cornell University Press.

Downloaded from cps.sagepub.com at LUND UNIVERSITY LIBRARIES on December 28, 2012

Fortin 927

Gonick, L. S., & Rosh, R. M. (1988). The structural constraints of the world-
economy on national political development. Comparative Political Studies,
21, 171-199.
Granger, C. W. J. (1969). Investigating causal relations by econometric models and
cross spectral methods. Econometrica, 37, 424-438.
Granger, C. W. J. (1980). Testing for causality: A personal viewpoint. Journal of Eco-
nomic Dynamics and Control, 2, 329-352.
Grzymala-Busse, A. (2007). Rebuilding leviathan. Party competition and state exploi-
tation in post-communist democracies. Cambridge, UK: Cambridge University
Press.
Grzymala-Busse, A., & Jones Luong, P. (2002). Reconceptualizing the state: Lessons
from post-communism. Politics & Society, 30, 529-554.
Hashim, M. (2005). Putin’s etatization project and limits to democratic reforms in
Russia. Communist and Post-Communist Studies, 38, 25-48.
Hellman, J., Jones, G., & Kaufmann, D. (2000). Seize the state, seize the day: State
capture, corruption and influence in transition (Policy Research Working Paper
2444). Washington, DC: World Bank.
Hendley, K. (1997). Legal development in post-Soviet Russia. Post-Soviet Affairs,
13, 228-251.
Hicks, A. M. (1994). Introduction to pooling. In T. Janoski & A. M. Hicks (Eds.), The
comparative political economy of the welfare state (pp. 169-188). Cambridge,
UK: Cambridge University Press.
Holmes, S. (1997). When less state means less freedom. Transitions, 4(4), 66-75.
Holmes, S. (2006). Conclusion: The state of the state in Putin’s Russia. In T. J. Colton
& S. Holmes (Eds.), The state after communism: Governance in the new Russia
(pp. 299-310). Lanham, MD: Rowman & Littlefield.
Honaker, J., King, G., & Blackwell, M. (2009). Amelia II: A program for missing
data. Retrieved from http://gking.harvard.edu/amelia
Huber, E., Rueschemeyer, D., & Stephens, J. D. (1999). The paradoxes of contem-
porary democracy: Formal, participatory, and social dimensions. In L. Anderson
(Ed.), Transitions to democracy (pp. 168-192). New York, NY: Columbia Uni-
versity Press.
Huntington, S. S. (1991). The third wave: Democratization in the late twentieth cen-
tury. Norman: University of Oklahoma Press.
Jackman, R. W. (1973). On the relation of economic development to democratic per-
formance. American Journal of Political Science, 17, 611-621.
Judge, G. C., Griffiths, W. E., Carter Hill, R., Lutkepohl, H., & Lee, T.-C. (1985). The
theory and practice of econometrics (2nd ed.). New York, NY: John Wiley.
Kaufman, R. R. (1999). Approaches to the study of state reform in Latin American
and postsocialist countries. Comparative Politics, 31, 357-375.

Downloaded from cps.sagepub.com at LUND UNIVERSITY LIBRARIES on December 28, 2012

928 Comparative Political Studies 45(7)

Keefer, P., & Stasavage, D. (2000). Checks and balances, private information, and the
credibility of commitments. International Organization, 56, 751-774.
King, G., Honaker, J., Joseph, A., & Scheve, K. (2001). Analyzing incomplete politi-
cal science data: An alternative algorithm for multiple imputation. American
Political Science Review, 95, 49-69.
Kittel, B. (1999). Sense and sensitivity in pooled analysis of political data. European
Journal of Political Research, 35, 225-259.
Kittel, B., & Winner, H. (2005). How reliable is pooled analysis in political economy?
The globalization-welfare state nexus revisited. European Journal of Political
Research, 44, 269-293.
Kmenta, J. (1971). Elements of econometrics. New York, NY: Macmillan.
Kopstein, J., & Reilly, D. A. (2003). Postcommunist spaces: A political geography
approach to explaining postcommunist outcomes. In G. Ekiert & S. E. Hanson
(Eds.), Capitalism and democracy in Central and Eastern Europe (pp. 120-154).
Cambridge, UK: Cambridge University Press.
Linz, J. J., & Stepan, A. (1996). Problems of democratic transition and consolidation.
Baltimore, MD: Johns Hopkins University Press.
Lipset, S. M. (1959). Some social requisites of democracy: Economic development
and political legitimacy. American Political Science Review, 53, 69-105.
Mann, M. (1993). The sources of social power, Vol. 2: The rise of classes and nation
states, 1760–1914. Cambridge, UK: Cambridge University Press.
Marshall, M. G., & Jaggers, K. (2009). Polity IV project: Political regime character-
istics and transitions, 1800–2007. The Polity IV dataset. Retrieved from http://
www.cidcm.umd.edu/inscr/polity/
Munck, G., & Verkuilen, J. (2002). Conceptualizing and measuring democracy: Eval-
uating alternative indices. Comparative Political Studies, 35, 5-34.
Mungiu-Pippidi, A. (2008, November). The learning curve revisited: State-building
lessons from postcommunist Europe. Paper presented at the Conference on New
Public Management and the Quality of Government, Gothenburg, Sweden.
Mungiu-Pippidi, A. (2010). When Europeanization meets transformation: Lessons
from the unfinished Eastern European revolutions. In V. Bunce, M. McFaul, &
K. Stoner-Weiss (Eds.), Democracy and authoritarianism in the postcommunist
world (pp. 59-81). Cambridge, UK: Cambridge University Press.
O’Donnell, G. (1993). On the state, democratization and some conceptual problems:
A Latin American view with glances at some postcommunist countries. World
Development, 21, 1355-1369.
Przeworski, A. (Ed.). (1995). Sustainable democracy. Cambridge, UK: Cambridge
University Press.

Downloaded from cps.sagepub.com at LUND UNIVERSITY LIBRARIES on December 28, 2012

Fortin 929

Ragin, C. C. (1998). Comparative methodology, fuzzy sets and the study of sufficient
causes. APSA-CP: Newsletter of the Comparative Politics Section of the APSA,
9, 18-22.
Ragin, C. C. (2000). Fuzzy-set social science. Chicago, IL: University of Chicago
Press.
Remington, T. (2006). Democratization, separation of powers, and state capacity. In
T. J. Colton & S. Holmes (Eds.), The state after communism: Governance in the
new Russia (pp. 261-298). Lanham, MD: Rowman & Littlefield.
Risse, T., Green Cowles, M., & Caporaso, J. (2001). Europeanization and domestic
change: Introduction. In T. Risse, M. Green Cowles, & J. Caporaso (Eds.), Trans-
forming Europe: Europeanization and domestic change (pp. 1-20). Ithaca, NY:
Cornell University Press.
Roberts, C., & Sherlock, T. (1999). Bringing the Russian state back in: Explana-
tions of the derailed transition to market democracy. Comparative Politics, 31,
477-498.
Sakwa, R. (1999). Postcommunism. Buckingham, UK: Open University Press.
Seewright, J. (2002). Testing for necessary and/or sufficient causation: Which cases
are relevant. Political Analysis, 10, 178-193.
Sharlet, R. (1998). Legal transplants and political mutations: The reception of consti-
tutional law in Russia and the newly independent states. Eastern European Con-
stitutional Review, 7(4), 107-123.
Smith, P. H. (2005). Democracy in Latin America: Political change in comparative
perspective. Oxford, UK: Oxford University Press.
Spruyt, H. (1994). The sovereign state and its competitors. Princeton, NJ: Princeton
University Press.
Suleiman, E. (1999). Bureaucracy and democratic consolidation: Lessons from East-
ern Europe. In L. Anderson (Ed.), Transition to democracy (pp. 141-167). New
York, NY: Columbia University Press.
Tilly, C. (2007). Democracy. Cambridge, UK: Cambridge University Press.
Vachudova, M. A. (2005). Europe undivided: Democracy, leverage, and integration
after communism. Oxford, UK: Oxford University Press.
Vachudova, M. A. (2010). Democratization in postcommunist Europe: Illiberal
regimes and the leverage of the European Union. In V. Bunce, M. McFaul, &
K. Stoner-Weiss (Eds.), Democracy and authoritarianism in the postcommunist
world (pp. 82-104). Cambridge, UK: Cambridge University Press.
Volkov, V. (2002). Violent entrepreneurs: The use of force in the making of Russian
capitalism. Ithaca, NY: Cornell University Press.
Wang, X. (1999). Mutual empowerment of state and society: Its nature, condition,
mechanisms, and limits. Comparative Politics, 31, 231-249.

Downloaded from cps.sagepub.com at LUND UNIVERSITY LIBRARIES on December 28, 2012

930 Comparative Political Studies 45(7)

Way, L. A. (2005). Authoritarian state building and the sources of regime competi-
tiveness in the fourth wave: The cases of Belarus, Moldova, Russia, and Ukraine.
World Politics, 51, 231-261.
Weingast, B. R., & Moran, M. (1983). Bureaucratic discretion or congressional con-
trol? Regulatory policymaking by the Federal Trade Commission. Journal of
Political Economy, 91, 765-800.

Bio
Jessica Fortin is a research associate at GESIS—Leibniz Institut für Sozial-
wissenschaften in Mannheim, Germany.

Downloaded from cps.sagepub.com at LUND UNIVERSITY LIBRARIES on December 28, 2012