Politics and privatization in Central and Eastern …Presidentialism, on the other hand, has not had...
Transcript of Politics and privatization in Central and Eastern …Presidentialism, on the other hand, has not had...
Politics and privatization in Central and Eastern Europe:
A panel data analysis
Christian BjørnskovAarhus School of Business, Aarhus University
Niklas Potrafke1
University of Konstanz
This version: December 27, 2008
Abstract
This paper examines how government ideology has affected privatization in Central and Eastern
Europe after the end of the Cold War. We analyze a panel of 19 Central and Eastern European
countries in the period 1990 to 2007. Privatization is measured by the indices of the European
Bank for Reconstruction and Development (EBRD). Regarding the political determinants, we
consider the special characteristics of the young democracies and semi-presidential systems in
Central and Eastern Europe and construct a new set of political variables. The results
demonstrate that privatization was forced by rightwing or rather market liberal governments
while nationalist governments retarded this process. Moreover, taking into account the rapid
transition process in the beginnings of the 1990s in particular, leftist governments stuck to public
ownership even much stronger than later on.
Keywords: Central and Eastern Europe, ideology, privatization, panel data
JEL Classification: P20, P30, C23
1 Bjørnskov: Aarhus School of Business, Aarhus University, Department of Economics, Frichshuset, Hermodsvej 22,
DK-8230 Åbyhøj, Denmark. Email: [email protected]. Potrafke: University of Konstanz, Department of Economics,
Box 138 D-48457 Berlin, Germany, Email: [email protected]. We are grateful for comments on
earlier versions from Viktor Brech, Mogens Kamp Justesen, Brian Mandau and Bernd Potrafke and the participants
of the Research Seminar of the University of Groningen in August 2008 and of the Brown Bag Seminar of the
University of Konstanz in November 2008. The usual disclaimer naturally applies.
2
1. Introduction
Since late 1989, when communism collapsed and the Soviet Union withdrew both militarily and
politically from Central and Eastern Europe, most countries in that region have gone through
considerable restructuring of their economies, including major privatization. But they often
adopted contrasting strategies: some took a gradual approach to transition while others
implemented a “big-bang” approach. A few of them have yet to restructure their centrally-
planned economies.
Why do some countries privatize rapidly and others more slowly? Ultimately, why do
certain Central and Eastern European countries continue to carry a large government sector?
These questions remain contested, as do the relative merits of gradual transitions and shock
therapies (cf. Fish and Choudhry, 2007). In this paper, we do not aim to answer this second
question or the many other pertaining to the effects of privatization.2 Instead, we explore
whether the political ideology of incumbent governments in Central and Eastern Europe has
affected their privatization efforts, i.e. we look for a political explanation of these differences. We
argue that the European transition from communism to market economy offers a unique setting
to study the potential effects of partisan politics, as this group of countries initially had a set of
similar institutions, which had been forced upon them after World War II.
Most of these countries were members of the Warsaw Pact and almost all took part in the
COMECON, implying that they not only had domestically planned economies but also partook
in economic planning at a regional level. With the exogenous shock of the Soviet collapse, these
countries had the opportunity to fundamentally restructure their economic and political
institutions. While partisan politics in the Western World is usually about gradual and relatively
small changes, the postcommunist transition offered a unique opportunity of a wholesale
2 On the many effects of privatization as well as its different forms, we refer interested readers to the comprehensive
survey in Megginson and Netter (2001). In the specific context of Central and Eastern Europe, Berkowitz and
DeJong (2003), for example, show that large-scale privatization has been clearly associated with faster growth rates in
Russian regions.
3
transformation decided upon by political actors, and a choice between either a fast transition,
based on what has become known as the “Washington Consensus” associated with the political
right, or a gradual transition towards a society in which the government continues to play a
central role more amenable to traditionally leftwing politics.
Our paper is, to our knowledge, the first to build a measure of political ideology in these
transition countries. Focussing on the role of political ideology in these countries, we find
evidence to suggest that the ideology of the incumbent government was associated with the
speed of privatization in a window of opportunity in the first years of transition, but not
afterwards. Nationalist-led governments were associated with slower transition speeds.
Presidentialism, on the other hand, has not had a clear impact on privatization efforts in Central
and Eastern Europe.
The remainder of the paper is structured as follows. Section 2 first describes the political
changes and difficulties, and the process of privatization in Central and Eastern Europe since
1989. Section 3 describes our data, and in particular the measure of ideology employed in the
paper. Section 4 outlines the results, which we discuss in the concluding section 5.
2. Privatization and ideology in Central and Eastern Europe since 1989
When Hungary opened its borders to Austria on September 11, 1989, the action was effectively
the beginning to the end of Soviet domination of Central and Eastern Europe. The German
Democratic Republic tried to stem the flow of its citizens visiting West Germany by closing its
borders to Czechoslovakia on October 3, but the final damage was done and the communist bloc
in Central and Eastern Europe effectively began to collapse. Although very few commentators
had seen it coming, the East German regime relatively quickly gave in and opened its borders to
the other part of Germany on November 9, the Polish government decided to form a coalition
government with the opposition movement Solidarnosc, and the Czech Velvet Revolution also
dismantled the socialist system rapidly. Within months, communism had collapsed in the region
4
and countries that had effectively been governed from Moscow since the late 1940s once again
gained full sovereignty. Political socialism was gone, and the next logical step for most of these
countries seemed to be to get rid of its economic manifestation, government economic planning.
Even before the postcommunist transition, Hungary began to allow some foreign direct
investments and private sector activity in response to a deepening economic crisis in the 1980s.
For example, Levi’s established a production facility outside Budapest under a new joint venture
law, and a two-tier banking system was already in development. While limited, these steps may
have facilitated the Hungarian transition and might have revealed a preference for more private
activity already existing in the political system pre-transition. After the collapse of communism,
the Central European economies were rapidly transformed with Hungary one of the forerunners.
The private sector share of Hungarian GDP rose from five percent in 1989 to 50 percent four
years later; starting from the same initial level, the Czech private sector increased to 47 percent of
GDP (EBRD, 2007). Poland, which from the onset had a substantially larger private sector
accounting for around 30 percent of GDP, reached the same level as Hungary in 1993 while
Estonia, which presumably started at the Soviet private sector share of five percent, reached a
privatization degree of no less than 61 percent in four years after regaining independence in 1991.
The privatization process took comparatively longer time in other countries. For example,
the private sector share of GDP in Ukraine took until 1996 to reach the 50 percent level, while in
Serbia it remained at 40 percent as late as 2002. At the extreme end of the transition and
privatization experience, the private sector share of GDP in Turkmenistan reached 25 percent by
1997 and has stagnated since, while Belarus – all but in name a communist dictatorship – took
until 2002 to reach the same rather modest level of private activity. However, we treat these
extremes as outliers for two reasons. First of all, Belarus started privatization in the democratic
years following independence, yet started to reverse the process when Lukashenko took over the
presidency in July 1994 and rapidly moved the country towards the model of a communist
dictatorship. The same pattern applies to Turkmenistan, which suggests that these countries are
5
on a fundamentally different trajectory than the remaining countries of the region. In statistical
terms, these countries are therefore properly treated as outliers, not only due to their communist
ideology but also due to the possibly different mechanisms underlying privatization and political
decision-making. We therefore exclude them from the following analysis.
Our simple figures document both the rapid transition in some countries and the large
variation across the entire postcommunist region despite the relatively similar initial situation of
these countries. In order to gain a fuller understanding of the transition, the set of indices
developed at the European Bank for Reconstruction and Development (EBRD) provides a
broader measure of privatization by weighing seven key variables: 1) privatization revenues
(cumulative, in percent of GDP); 2) private sector share in GDP (in percent); 3) private sector
share in employment (in percent); 4) budgetary subsidies and current transfers (in percent of
GDP); 5) the share of industry in total employment (in percent); 6) the change in labour
productivity in industry (in percent); and 7) investment/GDP (in percent). The data, which we
employ in the following analyses, cover three indices: an overall index of enterprise reform,
denoted EBRD-All, an index of small-scale privatisation (EBRD-Small), and an index of large-
scale privatization (EBRD-Large). All three indices are distributed from 1 (no privatization) to 4.3
(maximum transition to private market economy). Figure 1 tracks these three indices of the 19
countries included in our comparison, while Figure 2 illustrates the overall development in the
average country. Throughout this paper, we measure privatization by these indices.
Insert Figure 1 about here
Insert Figure 2 about here
It is quite evident from the data and the figures that a process of rapid privatization, as
indicated above by the simpler private sector share data, was chosen in Czechoslovakia, Estonia,
Hungary, and Poland. In addition, Figure 2 clearly illustrates that the bulk of the privatization
6
efforts took place in the period 1990 to 1994, which we will treat as a specific window of
opportunity (see below). Of the countries that had reached or surpassed an index value of 3 by
2007, Hungary was the first in 1992, followed by the newly-formed Czech Republic, Estonia, and
Poland in the year after, while Slovakia – despite registering a value of 3 as part of
Czechoslovakia in 1993 – followed as an independent country in 1999. Lithuania and Slovenia
reached that level in 2002, Latvia in 2003, and Croatia in 2004. This development is also reflected
in the 2007 data, in which four countries – Estonia, Hungary, Poland and Slovakia – score 3.7,
the Czech Republic scores 3.3, while a third group consisting of Croatia, Latvia, Lithuania and
Slovenia have reached a level of 3.0. At the other end of the scale, Tajikistan and Uzbekistan
remain stuck with an extensively government-controlled and planned economy at 1.7, and
Belarus and Turkmenistan receive the score 1 as they continue to operate Soviet-style communist
economies (cf. Åslund, 2007). We do not consider these countries since they remain
fundamentally unreformed.
Some countries chose shock therapy, others succeeded in implementing more gradual
reforms, while a few either took no reform measures or rolled reforms back swiftly after
beginning to implement them. In the following, we do not consider which was the best option
but instead look at why countries chose different speeds of privatization with emphasis on
potential ideological reasons. Fish and Choudhry (2007) provide a survey of the debate on the
best way to liberalize the postcommunist economies, with a special focus on the impact of this
choice on subsequent democratization. They stress that the dispute over gradualist transitions
versus shock therapy liberalization was and is highly ideological. For example, “frequent resorts
to the term neoliberalism usually signals gradualist sympathies” (Fish and Choudhry, 2007, 255).
They therefore term the two conflicting views on the desirable speed and degree of liberalization
as ‘the Washington Consensus’ and ‘the Social Democratic Consensus’.
The Washington consensus, which is also referred to as the Big Bang approach or shock
therapy, is most often associated with a fast transition towards market economy with substantial
7
and deep liberalization of all markets. In addition, many commentators also associate this
approach with an acceptance of relatively large short-run transitional costs, including significant
social costs in the form of unemployment and uncertainty.3 The Social Democratic Consensus,
also known as the gradualist approach, on the other hand favours a relatively slow reform pace–
and only gradual liberalization that maintains some state enterprises and government regulation.
As such, while the adherents of this approach typically stress their concern for minimizing the
social costs of transition, it also implies a higher degree of respect for existing power structures
and a continuation of political control over the economy. Overall, if policies were consistent with
ideology, we would expect that rightwing governments in Central and Eastern Europe were more
likely to chose shock therapy and privatize their economies faster.
However, the prospects of implementing ideological policies may have varied over the
course of transition. First, immediately following the collapse in 1989-90, expectations were high
and large parts of the populations of the postcommunist countries were therefore ready to accept
even large changes in society. In the years that followed, reality crept in on the voters as
substantial parts of the economies proved unsustainable and were dismantled with mass
unemployment as a natural consequence.
Second, in most countries the transition meant a relatively clean break with vested political
interests, and may thereby have created a unique situation in which the political influence of such
interests from established industries was at a minimum. Conversely, it is often argued that one of
the main problems of established democracies is the effect which Olson (1982) termed
‘institutional sclerosis’, that special interests are embedded in even minor policymaking decisions.
3 Traditionally, many rightwing economists following the tradition of Friedrich Hayek would argue for a gradual
approach to institutional change instead of large-scale experiments of ‘social engineering’. It may therefore seem
confusing that the Big Bang approach has come to be associated with the political right wing. However, the answer
should probably be sought in the recognition that too slow reforms might give special interests time to ‘regroup’,
that a fast transition would be towards a well-known system and thus not an exercise in typical social engineering,
and that the belief that the flexibility of a full market economy itself would ensure minimal transitional costs.
8
Consequently, policies that would affect one or more interests adversely are not undertaken. For
these reasons, one could therefore argue that both rapid privatization per se and partisan politics
were more likely to occur in an early phase of transition, i.e. within a window of opportunity in
the first years after the communist collapse when voters were willing to support large-scale
reforms and special industry influence was dispersed.4
In the following, we explore the role of the partisan ideological preferences of the
incumbent governments during the period since 1989, i.e. we attempt to explain the patterns in
Figure 1. We furthermore explore the potentially differential privatization effects within a
window of opportunity in the first years following transition. We draw data from a number of
different sources as well as developing our own measure of government ideology in the transition
period. The following section describes these data.
3. Further data and estimation strategy
First of all, we potentially include 20 Central and Eastern European countries in our dataset:
Albania, Belarus, Bosnia and Herzegovina, Bulgaria, Croatia, the Czech Republic, Estonia,
Hungary, Latvia, Lithuania, Macedonia, Moldova, Montenegro, Poland, Romania, Russia, Serbia,
the Slovak Republic, Slovenia, and Ukraine. Belarus is excluded from the analysis, as it has
approximately 80 percent independents in parliament, which has been controlled by a communist
dictator, Alexander Lukashenko. There is also only a limited number of observations for Bosnia
and Herzegovina in particular. This brings our dataset to 19 countries and a maximum of 342
observations for which we have data on the three privatization indices outlined in the previous
section; Table 1 summarizes the data.
4 Although we are unable to test the hypotheses, one might also argue – following Olson (1982) – that the effects of
breaking the bond between politics and industry interests would be larger with rightwing governments. Many
leftwing governments in Central and Eastern Europe immediately after transition consisted partly of reformed
communists, and thereby had close connections to vested interests, while the first rightwing governments did not
suffer this problem.
9
Insert Table 1 about here
Our dependent variables in the following are the changes of the EBRD-All, EBRD-Large
and EBRD-Small indices, as we are interested in the intensity of the privatization process, i.e. the
speed of the transition to market economy. The use of changes in the privatization index to some
extent can also be argued to be less sensitive to endogeneity problems. If election outcomes are
affected by the support for ‘capitalism’ in some broad sense, the use of levels of privatization
could proxy for how far – and potentially too far – the process of transition has advanced, yet
using index changes does not necessarily entail the same problem.
3.1. The ideology data
Using index changes thus ameliorates the standard problem of causality. However, on top of the
‘usual’ problems, placing parties and governments on an ideological scale presents additional
difficulties.5 First of all, the political systems were naturally in flux during the first years after
communism collapsed. A number of former communist parties refitted themselves as Western
European style Social Democrat parties while their place in the political system was taken over by
new leftwing parties. It is therefore not easy to decide when to code the new parties and
governments as ‘reformed’ and when to treat them as effectively identical to Soviet communist
parties. On a broader level, in particular in the early phases of development after communism
collapsed, one would expect ideology in Central and Eastern Europe to be less precisely
measured due to this type of problem. Yet, by evaluating official party programs, a rough
5 The problems of coding ideology are discussed at length in both Poole and Rosenthal (2006) and the March 2007
issue of Electoral Studies, which included a special symposium on “Comparing Measures of Party Positioning: Expert,
Manifesto, and Survey Data”.
10
codification of parties in Central and Eastern Europe is possible, while keeping possible
problems in mind.
Second, a number of Central and Eastern European political systems have come to include
at least one strongly nationalist party. For example, the nominally conservative party in power in
Albania in 1992 may also be coded as a classically nationalist and populist party. This problem
recurs in a number of other countries and not least in the formerly Yugoslavian countries where
Croatia was governed by former general in its early years. As such, the data
potentially include a second ‘nationalist’ dimension to politics in addition to the traditional left-
right scale, which we develop in the following. We nevertheless ignore this problem in the main
analysis as the straightforward solutions would be to either code parties on a left-right scale
according to their actual policies instead of their nominal placement, in which case we would risk
confounding the privatization effort with the ideology coding and thus obtain entirely spurious
results.6 Alternatively, we could include a second nationalist dimension in the data although this
risks being highly subjective. We therefore only use this as a part of the robustness tests since the
existing literature provides no clear guidelines for how to identify such governments.
Third, a purely practical problem is the rapid change in parties, exemplified in the broad
Hungarian coalition called the “Federation of Young Democrats”, FIDESZ, that came to power
in the 1998 election. While it started as a youth organization in 1988 that rapidly developed a
liberalist/libertarian program, the Federation may in recent years be best characterised as
conservative, yet it consists of many factions that may not all fit easily into a standard left-right
continuum and has evolved considerably since 1988 (cf. Kiss, 2002). To the extent that the
ideology of single member parties of coalitions in Central and Eastern Europe cannot be
6 The problem with confounding variables in constructing the ideology variable is pertinent in all the literature. It is
also the reason why we refrain from using the alternative ideology index from Bjørnskov (2008), as actual party
preferences are central to his ideology coding.
11
precisely identified, the index is less precise than one would ideally want, and also less precise
than similar data for Western Europe where established party systems are more predictable.7
As a consequence, one has to accept a less than ideally precise measure of political ideology. We
therefore follow the fairly simple procedure in Woldendorp et al. (2000) by measuring the
governments’ ideological positions that places the cabinet on a discrete left-right scale with values
between 1 and 5. It takes the value 1 if the share of rightwing parties in terms of seats in
government and their supporting parties in parliament is larger than 2/3, and 2 if it is between
1/3 and 2/3. The index is 3 in a balanced situation if the share of centre parties is 50 percent, or
if the leftwing and rightwing parties form a government that is not dominated by one side or the
other. Corresponding to the first two cases, it takes the values 4 and 5 by a dominance of the
leftwing parties defined likewise. Following Potrafke (2008), this indicator is consistent across
time, but does not attempt to capture differences between the party-families across countries.
Finally, the years in which there was a change of government are labelled according to the
party that was in office for a longer period, e.g. when a rightwing government followed a leftwing
government in August, we label this year as leftwing, as decisions taken by the new government
are most likely to take effect in the subsequent year. In the following, it must be noted that the
coding of the ideology variable – from right to left – implies a predicted negative association with
privatization due to the discussion in section 2.
3.2. Control variables and estimation strategy
We additionally control for the yearly percentage change in GDP ( ), the yearly change
in employment or engaged persons ( , election years, presidential systems, and
7 Coding political ideology in Bosnia and Herzegovina presents special problems. First of all, even though the
country elected a parliament in 1990, some of the heaviest fighting in the Ex-Yugoslavian civil war took place in the
capital Sarajevo, effectively preventing any form of politics. The next parliamentary elections took place in 1998.
Second, what politics took place was for rather clear reasons dominated by nationalist and religious factors outside
standard ideological concerns. We therefore cannot code ideology in Bosnia before the end of the war in late 1995.
12
the yearly change in trade ( ), which captures the success of the overall transition process
and thereby probably also popular support for an ongoing transition, and the change towards a
generally more open regime, respectively; Table 1 provides the data sources. In consecutive steps,
as noted above, we also add a dummy for the period before 1994, which is intended to capture
the ‘window of opportunity’ that opened after the Soviet collapse and the fall of the Berlin Wall
in late 1989. In subsequent tables, we instead add an alternative dummy for country-specific
windows of opportunity, such that a five-year window starts in 1990 for the Czech Republic, but
only the year after independence for countries that were not already independent in 1989. We
also try another alternative coding identical to the second measure, except for Bosnia and
Herzegovina, Croatia, Serbia and Macedonia for which the window opens starting in 1995 with
the Dayton peace settlement following the war in ex-Yugoslavia.
In a set of additional robustness tests, we exclude single countries, control for former
membership of the Soviet Union and expected EU membership, add variables capturing a
broader concept of globalisation, and allow for effects of nationalist parties. Throughout, our
choice of estimation strategy is the use of a feasible generalized least squares (FGLS) estimator
with random effects. A series of Breusch-Pagan tests provide support for the inclusion of
random effects while Hausman tests provide no support for having to include fixed effects. As a
supplement, we provide a table with the main results estimated with panel-corrected standard
errors as well as a table with first-order autocorrelated disturbance terms in order to address the
potential problems of contemporaneous correlation across countries and of serially correlated
errors, respectively.
Insert Table 1 about here
4. Results
13
As can readily be seen in Figures 1 and 2, the efforts to privatize industries in Central and Eastern
Europe have been successful in most countries, yet to different degrees and at different speeds.
In the following, we provide a set of estimates to answer our main question, whether political
ideology affected the speed and depth of the privatization efforts in the European transition
countries.
4.1. Main results
As a first, Table 2 reports the basic results using the overall privatization index, EBRD-All.
Column 1 only adds the ideology measure, which is significant with a negative coefficient,
indicating that relatively rightwing governments have been more likely to privatize. Adding the
simple window of opportunity dummy (1990-1994) in column 2 substantially increases the fit of
the model, while adding an interaction between this dummy and ideology provides further
explanatory power. These simple results first indicate that while government ideology in general
is not associated with privatization efforts, the ideology of governments within the first years of
the Central and Eastern European transition was a significant determinant of such efforts.
Adding GDP growth and the development of employment to the specification in column 4, the
estimates indicate that the latter has not been associated with privatization while the GDP result
indicates that early privatization was accompanied by rather large drops in output in the window
of opportunity after 1989.8 We find no evidence for either election year effects on privatization
efforts, or for different performance of presidential systems. However, adding the expansion of
trade reveals a significant effect on the speed of privatization in Central and Eastern Europe.
Insert Table 2 about here
8 We should stress that there may be an endogeneity issue, as early privatization after the communist collapse may
have caused an additional temporary decrease in output. However, this problem is likely to be more important when
focusing on large-scale than small-scale privatization. Either way, the evidence is broadly consistent with the U-
shaped development of average incomes during transition (Berg et al., 1999).
14
Tables 3 and 4 separate privatization into two parts, following the coding of the EBRD,
namely privatization of large-scale and small-scale industry. Focusing first on small-scale
privatization in Table 3 reveals fairly similar results as in Table 2. Privatization was substantially
stronger in the first years following the collapse of communism, and stronger in countries with
deeper transition costs. However, the ideological effects on the EBRD-Small index within the
window of opportunity are clearer in terms of both size and significance of the coefficient; they
are also unaffected by the inclusion of the additional controls in columns 4-7 even while trade
expansion again becomes significant. On the other hand, large-scale privatization as explored in
Table 4 did not depend to the same degree on the window of opportunity and only seems to
have been influenced weakly by the ideology of the government. Conversely, the development of
the EBRD-Large index tends to have been affected by the development of trade openness to a
degree, which is both statistically and economically significant. Furthermore, we also find that
large-scale privatization was not as strongly associated with the economic troughs, as captured by
GDP, as the corresponding efforts to privatize small-scale industry.9
Insert Tables 3 and 4 about here
4.2. Robustness tests
As such, the estimates show that the ideological influence within the first years of transition
appears to have worked entirely on the efforts to privatize small-scale industry and business.
However, we have so far assumed that this five-year window of opportunity was the same for all,
even though some of the 19 countries in our sample were not independent at the beginning of
9 One important issue is that large-scale industry may tend to have long-term contracts. The privatization of large-
scale firms is likely to take longer if such contracts have to be satisfied before restructuring. However, we note that
large-scale privatization does not seem to be better explained by lagged government ideology, which one would
expect if such privatization takes longer time than small-scale privatization.
15
our period in 1990. In addition, even though certain countries did gain their independence,
privatization was far from taking priority in the Ex-Yugoslavian countries that fought bloody civil
wars following formal independence. Table 5 provides an overview of independence dates and
the duration of civil wars. The years in which independence occurred are labelled corresponding
to the circumstances that were present for a longer period, e.g. when independence occurred in
August, we label this year as not independent, although inferences do not change when the five
year window starts in the year of independence.
Insert Table 5 about here
As a solution to the problem of asymmetrical independence dates, Tables 6 and 7 perform
a first type of robustness test by including one of the two dummies for alternative windows of
opportunity that are measured not in terms of the years 1990-1994, but in five years following the
first opportunity to begin transition to a privately-driven market economy. As such, the
alternatives allow for a set of countries – Lithuania, Estonia, Latvia and Slovenia, Bosnia and
Herzegovina and Croatia, and Montenegro – to begin transition substantially later than the rest of
Central and Eastern Europe, following the dates in Table 5.10 In the measure in Table 6, the
window of opportunity takes its beginning at independence while the measure used in Table 7
10 Montenegro provides a particularly difficult case to handle, as it gained its formal independence only in June 2006,
following a popular referendum, but may have had de facto independence well before that date. For example, after
control over its economic policies and even adopted the German Mark as its official currency. For the Ex-
Yugoslavian countries, we count transition onset as the earliest data at which the independence war was effectively
over. Slovakia may be another troublesome case as it both benefitted from the joint Czechoslovakian transition until
the breakup of the country in 1993, but followed a different path after independence. However, we code transition
onset in 1989 for both the Czech Republic and Slovakia.
16
takes its beginning either at independence when independence was peaceful or after the
conclusion of the civil wars in Ex-Yugoslavia.
Insert Tables 6 and 7 about here
The results using the alternative coding of a window of opportunity are very similar, as the
estimates for large-scale privatization are in general insignificant while those for small-scale
privatization are strongly significant and of almost exactly the same size as in Table 3. It therefore
seems safe to conclude that the results are not specific to one way of classifying the window of
opportunity arising immediately after the collapse of communism. In Tables 8 and 9, we address
two other potential problems, as estimates in the preceding tables could be biased given that
there is either contemporaneous cross-country correlation or autocorrelated country-specific
residuals. The first of these problems may be particularly pertinent if privatization policies
diffused across borders such that governments learn from each other, ideologically compatible or
not, or from the same source. In that case, we would expect the problem to show up as
contemporaneous correlation, implying that assigning credit or blame for policies to specific
governments would not be viable.
To ameliorate these problems, we provide estimates obtained with panel-corrected standard
errors according to Beck and Katz (1996) and random effects estimates allowing for an AR (1)
disturbance, instead of the random effects FGLS estimator employed in the previous tables. This
again provides qualitatively similar estimates, yet the effects of political ideology within the
window of opportunity (coded as in Tables 3 and 4) are roughly a third larger than when applying
the random effects estimator.
Insert Tables 8 and 9 about here
17
In table 10, we add a dummy variable for whether the government was dominated by a
nationalist party; the nationalist dummy takes on the value one when the party of the chief
executive was nationalist and zero otherwise.11 The results pertaining to the nationalist dummy
mirror those of the ideology variable by showing that having a nationalist government within the
window of opportunity rather clearly reduced the likelihood that the government undertook
major privatizing reforms while, again, we find no evidence of nationalist effects outside the
window of opportunity. However, the inclusion of the nationalist dummy does not affect the
main result although they do contribute some interpretative nuance.
As two additional robustness tests, we replace the change in trade volume with the KOF
index of globalisation (see Dreher 2006a and Dreher et al. 2008 for details) and include a measure
of total government size in Table 11. Adding the globalization index, although the sample size
drops to 14 countries and 203 observations as this variable is not available for the former
Yugoslavian countries, arguably covers another political dimension as well as more interaction
with other European economies than trade volumes as the index also includes tariff and non-
tariff protection and flows of foreign direct investments. As before, its inclusion does not change
the main results that turn out to be robust to a number of changes. The marginal effects or trade
expansion are, nevertheless, about 2.5 times those of an equivalent change in the overall
globalization index, suggesting that the important factor is international trade, and not foreign
investments or other components of the broader globalization index.
Insert Tables 10 and 11 about here
11 Alternatively, the nationalist-dummy could have been coded referring to the largest government party if a
nationalist influence is primarily through the government and not the chief executive. We have tried both ways and
report the results using the chief executive, as we fail to find robust effects when coding nationalist government the
other way.
18
Adding the share of government expenditures as percent of GDP in columns 4-6 could
arguably be also important. While we argue that privatization efforts was a separable but central
part of the transition to market economy, one could also posit that the EBRD privatization
indices simply capture an overall shift away from an economy in which the government played a
central role to a less interventionist system, in which case one would also expect that transition
efforts would lead to smaller government expenditures. However, the estimates rather clearly
show that the effects of government ideology on privatization do not reflect a simple general
transition, but capture effects over and above changes in government size. In addition, we note
that even though the sample is down to 16 countries and 185 observations in columns 4 to 6, the
main results are basically unchanged.
We follow two suggestions in Megginson and Netter (2001, 357-358): First, as shown in
Table 12, we test for the possibility that we are simply picking up differences between nations
that were already sovereign before 1990, and those Central and Eastern European countries that
either regained independence (Estonia, Latvia, Lithuania) or emerged as new states following the
demise of communism (Russia, Ukraine). We also test for the possibility that a warranted
expectation of EU membership has driven countries towards faster transition. Second, in Table
13 we test for whether the existence of structural adjustment programmes and other
arrangements with the World Bank and the IMF have driven the privatization transition. This
might be important as an oft-repeated argument in political debates is that IMF programmes in
particular are ideologically biased, meaning that government ideology could simply proxy for the
presence of such programmes.
Insert Table 12 about here
The results in Table 12 first of all reject that our results reflect a difference between existing
and new states. However, while the inclusion of a dummy for the eight countries in the region
19
that actually achieved EU accession by 2004 (thus excluding Bulgaria and Romania) and an
interaction between this dummy and our window of opportunity dummy does not affect our
central estimates, they show weak support for accession countries having reformed faster in the
first years of transition. It should nevertheless be stressed that we cannot interpret this as
evidence of an incentive effect of prospective membership as our dummy necessarily captures the
situation ex post and does not measure any sensible ex ante factors.
Insert Table 13 about here
Finally, the results in Table 13 suggest that the presence of IMF arrangements has lead to
faster privatization in Central and Eastern Europe, all other things being equal. World Bank
presence, on the other hand, has not affected the transition speed. However, controlling for the
effects of such involvement does not alter any of our main conclusions, as it only seems to have
affected the privatization efforts of large-scale industry. In other words, while the IMF
contributed to a faster privatization of large-scale industries, whether in its capacity as a direct
partner or a political scapegoat, it has not affected the privatization of small-scale industry.
A set of further robustness tests are not reported in tables, particularly including a set in
which we check for the sensitivity of the results to individual countries. To rule out this
possibility, we performed the regressions again, excluding one country at a time. Overall, the
inferences are so robust that they are not sensitive to the inclusion of particular countries.
However, the impact of the ideology variables in the period till 1994 on the overall index declines
when the Czech Republic and to some extent Moldova are excluded, but remains significant on a
1 percent level in the basic scenario. On the other hand, the impact becomes stronger when
Lithuania is excluded. Coding the window of opportunity due to the countries’ individual
independence, the effects are remarkably weaker excluding the Czech Republic as well as
remarkably stronger excluding Lithuania. The impacts on the small-scale privatization index are
20
notably sturdy and the inferences regarding the large-scale privatization index also do not depend
on particular countries. As our results thus survive a set of robustness tests, we proceed to a final
discussion.
5. Discussion and conclusions
Most countries in Central and Eastern Europe have moved from a communist planned economy
to market economy since communism collapsed in late 1989. The speed of the transition has
nevertheless varied widely across these countries, with some of them adopting a gradualist
approach that leaves room for government intervention and state enterprises, while others took a
big-bang approach moving rapidly towards being in line with the Washington Consensus. Only
very few countries retained economic planning.
In this paper, we seek a political explanation to why countries took one path instead of
another. Noting that a gradual approach is more amenable to traditional leftwing positions, we
argue that the transition in Central and Eastern Europe offers a unique example in which to
study partisan politics, as all countries started from comparable situations, following the
exogenous shock of the collapse of Soviet communism. Furthermore, in the first years of
transition, a window of opportunity for partisan change existed as vested interests lost most of
their political influence and new political actors and parties emerged. While effects of ideology in
established democracies are often either absent or in the ‘wrong’ direction, with leftwing
governments more likely to implement reforms (cf. Cukierman and Tomassi, 1998), the special
circumstances in the transition countries would a priori make a more standard ideological
situation likely for the first years of transition.
To answer the question whether political ideology affected the speed of privatization in
Central and Eastern Europe after 1989, we first develop a measure of ideology in 19 transition
countries to form a panel dataset. Our results demonstrate that privatization was mainly
marshalled by rightwing or rather market liberal governments, although this effect is only clear
21
for the window of opportunity during the first few years of transition. Exploring the data further
reveals that rightwing governments in the postcommunist countries mainly forced the
privatization of small and medium-scale industries while we find no significant ideological
differences when it comes to the privatization of large industries. The size of the estimate is not
only statistically significant but also of economic and political significance, indicating that the
difference of the privatization of small-scale industry between a moderate rightwing and a
moderate leftwing government within the first years of transition in total amounted to almost one
point on the four-point EBRD index employed in this paper.
Biais and Perotti (2002) provide a first theoretical explanation for our findings. Their model
is based on the idea of class voting: capital owners tend to vote for the right wing and ‘working
class’ voters tend to vote for the left wing. Incumbent market-liberal governments can therefore
raise their chances of becoming reelected by making more voters capital owners through
introducing their privatization programmes. However intuitively appealing the part of their story
on re-electing market-liberal governments after privatization may seem, we note that it does not
fit the empirical evidence. Poland as well as Hungary are important counterexamples to this
theory. Moreover, the probability of having a rightwing government elected in the second
election after transition began does not differ between countries choosing fast and slow overall
privatization.12
It thus remains an open question why we find such clear ideological effects in the first years
of transition and not later on. A popular explanation instead relies on the disillusionment of the
12 The simplest way of testing the assumption underlying Biais and Perotti (2002) is to split the sample in a group
that privatized quickly and a group that privatized at a slower rate. Given their argument, one would expect rightwing
parties in countries in the first group to, on average, be more successful in second elections after transition, all other
things being equal. Regardless of whether we split the sample according to the progress in the first few years after
transition of the overall index, the small-scale or the large-scale index, we find no evidence to support Biais and
Perotti (2002). Indeed, using the small-scale index for which we find ideological effects, a two-tailed t-test comes out
with a probability p<.84.
22
public in certain countries such as Russia, where Megginson and Netter (2001, 326) stress that
privatization efforts “became unpopular […] because of the largely correct perception that they
were robbery by the old elite and the new oligarchs”. Åslund (2007) provides an alternative
potential explanation in his defence of shock therapy, in that he stresses that such major
institutional reversals are substantially easier before a new set of special interests have become
entrenched and intertwined with the political system. This explanation, consistent with Olson’s
(1982) notion of institutional sclerosis that comes about through the institutionalization of special
interests in mature democracies, also seems consistent with the finding that the ideological effects
were only significant for the privatization of small-scale industry, and total effects within the
window of opportunity in the first years after the collapse of communism were in general larger
for this type of industry. Put simply, insiders of old large-scale industries that had enjoyed a
perversely symbiotic relation with communist regimes embedded in the command economies,
had strong incentives to oppose privatization, as stressed by the early work of Blanchard and
Aghion (1996). These insiders and large-scale firms were also immediately after transition in a
situation where they had special interest influence on policymaking and could withstand
ideologically motivated privatization efforts not backed by a large majority of parties in
parliament. Furthermore, the short-run employment effects of restructuring or dismantling large
industries may have been too readily visible and prohibitively large without broad political
consensus backing it.
For small and medium-scale industries, on the other hand, the political connections are
likely to have been virtually embryonic in the first years of transition and thereby did not offer
them a similar protection against restructuring and privatization. Under such conditions, partisan
effects were much more likely to emerge, not least since it can be argued that traditional socialist
economics, which most politicians on the reformed leftwing were brought up with, never had a
real notion of the importance of small and medium-scale industry. On the other hand, many
rightwing politicians in the transition countries were influenced by either libertarianism or the
23
Chicago School of Milton Friedman, which argued for the importance of the underwood of small
business. However, these can only be suggestions for interpretation that must be judged by more
careful historical and political case studies.
A further potential worry associated with interpreting the findings of this paper could be
that the speed of privatization was not influenced directly by the political ideology of the first
governments, but that the choice of these governments was affected by a political-ideological
culture that had lain dormant since World War II (WWII), as indicated by the Lipset-Rokkan
theory (Lipset and Rokkan, 1967).13 In other words, the countries initially electing rightwing
governments may also have been those with a strong pre-war tradition for capitalism. However,
examples from some of the key countries seem to indicate otherwise. For example,
Czechoslovakia, which was one of the fast reformers, was a democracy with the political power
divided in the period 1936-1939 between the socialist president Edvard Benes and Prime Minister
Milan Hodza, who represented the Czechoslovak Agrarian Party as Prime Minister. Hungary,
another fast reformer, tied its economy to Nazi Germany in the years preceding WWII under the
nationalist Prime Minister Gyula Gömbös and Poland, which was an autocracy like Hungary, was
ruled by a centrist government. Only Estonia fits this pattern of consistence between interwar
popular opinion and postcommunist political choice, as its Prime Minister Konstantin Päts
abolished democracy in 1934 after the strongly anti-communist VAPS movement won a popular
referendum by 72 percent. Hence, this potential alternative explanation seems improbable, as has
also been argued by Rivera (1996).
In summary, we find that the first few years of transition offered an environment in which
partisan political differences clearly materialized in actual policies, exemplified here by
privatization efforts, yet the influence of ideology disappeared as new democratic polities were
13 The overall problem with estimating effects when countries start from potentially different initial conditions is
well-known in the literature. However, we believe that the initial conditions were so similar that it should not
represent a practical problem. Even if it was, the solution of the problem is anything but straightforward, and
discussed in de Haan and Sturm (2003).
24
established. Our findings consequently shed some light on how fast new polities find what may
be termed a “consensual balance” in which both left and right follow basically similar economic
policies, as we find no evidence of partisan effects outside a window of opportunity in the first
years of transition. Our findings may thus hold further implications for transition and
democratizing countries outside Central and Eastern Europe.
References
Beck, Nathaniel and Jonathan N. Katz (1996). Nuisance vs. substance: Specifying and estimating
time-series cross section models. Political Analysis 6, 1-36.
Beck, Thorsten, George, Clarke, Alberto Groff, Philip Keefer and Patrick Walsh. (2001). New
tools in comparative political economy: the database of political institutions. World Bank Economic
Review 15 (1): 165-176.
Berg, Andrew, Eduardo Borensztein, Ratna Sahay and Jermin Zettelmeyer. 1999. The Evolution
of Output in Transition Economies: Explaining the Differences. IMF working paper 99/73.
Berkowith, Daniel and David N. DeJong. 2003. Policy Reform and Growth in Post-Soviet
Russia. European Economic Review 47: 337-352.
Biais, Bruno and Enrico Perotti. 2002. Machiavellian Privatization. American Economic Review 92
(1): 240-258.
Bjørnskov, Christian. 2008. Political Ideology and the Structure of National Accounts in the
Nordic Countries, 1950-2004. Paper presented at the annual meeting of the European Public
Choice Society, Jena 27-30 March 2008.
Blanchard, Olivier and Philippe Aghion. 1996. On Insider Privatization. European Economic Review
40, 759-766.
Boockmann, Bernhard and Axel Dreher. 2003. The Contribution of the IMF and the World
Bank to Economic Freedom, European Journal of Political Economy 19, 3: 633-649.
25
Cukierman, Alex and Marianno Tomassi. 1998. When Does It Take a Nixon to Go to China?
American Economic Review 88 (1): 180-197.
De Haan, Jakob and Jan-Egbert Sturm 2003. Does more democracy lead to greater economic
freedom? New evidence for developing countries. European Journal of Political Economy 19: 547-563.
Dreher, Axel (2006a). Does globalisation affect growth? Evidence from a new index of
globalisation. Applied Economics, 38 (10), 1091-1110
Dreher, Axel (2006b). IMF and economic growth: The effects of programs, loans and
compliance with conditionality. World Development, 34 (5), 769-788
Dreher, Axel, Gaston, Noel and Pim Martens (2008). Measuring globalisation – Understanding its
causes and consequences. (Berlin: Springer)
EBRD, 2007. Transition Report Update 2007. European Bank for Reconstruction and
Development, London. Data available at
http://www.ebrd.com/country/sector/econo/stats/index.htm (accessed May 13, 2008).
Fish, M. Steven and Omar Choudhry. 2007. Democratization and Economic Liberalization in the
Postcommunist World. Comparative Political Studies 40: 254-282.
Heston, Alan, Robert Summers and Bettina Aten. 2006. Penn World Tables, Version 6.2, Center
for International Comparisons of Production, Income and Prices at the University of
Pennsylvania.
Kiss, Csilla. 2002. From Liberalism to Conservatism: The Federation of Young Democrats in
Post-Communist Hungary. East European Politics and Societies 16: 739-763.
Lipset, Seymour Martin and Stein Rokkan. 1967. Cleavage Structures, Party Systems, and Voter
Alignments: An Introduction. Pp. 1-64 in Seymour Martin Lipset and Stein Rokkan (eds.), Party
Systems and Voter Alignment: Cross-National Perspectives. New York: Free Press.
Megginson, William L. and Jeffry M. Netter. 2001. From State to Market: A Survey of Empirical
Studies on Privatization. Journal of Economic Literature 49: 321-389.
26
Olson, Mancur. 1982. The Rise and Decline of Nations: Economic Growth, Stagflation, and Social Rigidities.
New Haven, CT: Yale University Press.
Poole, Keith T. and Howard Rosenthal. 2006. Ideology and Congress: A Political Economic History of
Roll Call Voting. New York: Transaction Publishers.
Potrafke, Niklas. (2008). Public Health expenditures in OECD countries: Does policy matter?
Paper presented at the annual meeting of the European Public Choice Society, Jena 27-30 March
2008.
Rivera, Sharon Werning. 1996. Historical Cleavages or Transition Mode? Influences on the
Emerging Party Systems in Poland, Hungary and Czechoslovakia. Party Politics 2: 177-208.
Woldendorp, Jan, Hans Keman and Ian Budge. (2000). Party government in 48 democracies (1945-
1998): composition, duration, personnel. (Dordrecht: Kluwer Academic Publishers)
Åslund, Anders. 2007. How Capitalism was Built: The Transformation of Central and Eastern
Europe, Russia, and Central Asia.
27
Figure 1. Development of the EBRD indices in the period 1990 to 2007.
12
34
12
34
12
34
12
34
1990 1995 2000 2005
1990 1995 2000 2005 1990 1995 2000 2005 1990 1995 2000 2005 1990 1995 2000 2005
Albania Bosnia and Herzegovina Bulgaria Croatia Czech Republic
Estonia Hungary Latvia Lithuania Macedonia
Moldova Montenegro Poland Romania Russia
Serbia Slovak Republic Slovenia Ukraine
ebrdall ebrdsmallebrdlarge
year
Graphs by country
Figure 2. Development of the EBRD indices in the period 1990 to 2007. Averages of the 19 countries.
12
34
1990 1995 2000 2005 2010year
Overall SmallLarge
28
Table 1. Descriptive statistics
Variable Obs Mean Std. Dev. Min Max Source
Overall Privatization Index 342 2.20 0.81 1 3.7European Bank for Reconstruction and Development (EBRD)
Small ScalePrivatization Index 342 3.45 0.98 1 4.3
European Bank for Reconstruction and Development (EBRD)
Large ScalePrivatization Index 342 2.67 1.02 1 4
European Bank for Reconstruction and Development (EBRD)
Ideology 336 2.91 0.91 1 5 Own collectionDummy till 1994 342 0.28 0.45 0 1 Own collectionWindow ofOpportunity 342 0.27 0.44 0 1 Own collection
Window ofOpportunity(Civil War)
342 0.27 0.44 0 1 Own collection
GDP per capita(real) 317 6375.30 3506.59 318.27 22523.68
The Conference Board and Groningen Growth and Development Centre, Total Economy Database, January 2008
Persons engaged(in tousands) 317 7869.00 15330.84 512.30 75324.7
The Conference Board and Groningen Growth and Development Centre, Total Economy Database, January 2008
Election year(parliamentary) 342 0.30 0.46 0 1 Own collection
Presidential System 342 0.32 0.47 0 1 Own collectionTrade (sum of importsAnd exports as a share GDP)
255 96.93 33.76 26.30 174.40United Nations Economics Commissionwww.unece.org
Nationalist Dummy 226 0.13 0.34 0 1 Based on Beck et al. (2001)KOF index of globalisation 212 57.83 13.97 20.36 85.51 Dreher (2006a), Dreher et al.
(2008)Government Expenditures(as a share of GDP) 252 27.70 6.75 8.58 48.16 Penn World Table (Heston et al.,
2006)Dummy Former Soviet State 342 0.26 0.44 0 1 Own collectionDummy Prospected EU-Membership 342 0.42 0.49 0 1 Own collection
Number of World BankProjects agreed 304 2.17 2.70 0 18 Boockmann and Dreher (2003)
IMF Arrangement ineffect for at least 5 months in a particular year
304 0.43 0.52 0 2 Dreher (2006b)
IMF Arrangement agreed 323 0.24 0.44 0 2 Dreher (2006b)
29
Table 2. Regression Results. Dependent variable: all privatization. FGLS with random effects.
(1) (2) (3) (4) (5) (6) (7)
Ideology -0.025** -0.033*** -0.012 -0.027** -0.011 -0.010 -0.011[2.54] [3.51] [1.05] [2.53] [0.86] [0.79] [0.84]
Dummy till 1994 0.122*** 0.321*** 0.319*** 0.323*** 0.338***[6.27] [5.06] [4.64] [4.52] [4.13]
Ideology*Dummy till 1994 -0.066*** -0.065*** -0.066*** -0.068**
[3.29] [2.96] [2.90] [2.49]-0.128** -0.018 -0.017 -0.028
[2.08] [0.30] [0.27] [0.46]-0.208 -0.003 -0.002 0.003[1.29] [0.02] [0.02] [0.02]
Election-Year(parliamentary) -0.020 -0.027
[1.01] [1.26]Presidential-System 0.003 0.003
[0.17] [0.16]0.169***
[3.69]Constant 0.130*** 0.121*** 0.060* 0.137*** 0.057 0.059 0.056
[4.30] [4.21] [1.79] [4.22] [1.54] [1.51] [1.44]
Observations 336 336 336 311 311 311 234R-squared 0.02 0.12 0.15 0.04 0.15 0.16 0.22
Number of n 19 19 19 18 18 18 17
Notes: Absolute value of t statistics in brackets; * significant at 10%; ** significant at 5%; *** significant at 1%. The dependent variable is -All; estimation technique is FLS with random effects.
30
Table 3. Regression Results. Dependent variable: -scale privatization. FGLS with random effects.
(1) (2) (3) (4) (5) (6) (7)
Ideology -0.028*** -0.038*** -0.010 -0.032*** -0.010 -0.009 -0.009[2.59] [3.81] [0.88] [3.05] [0.88] [0.74] [0.77]
Dummy till 1994 0.158*** 0.415*** 0.393*** 0.412*** 0.550***[7.63] [6.25] [6.09] [6.15] [7.03]
Ideology*Dummy till 1994 -0.085*** -0.083*** -0.089*** -0.115***
[4.07] [4.04] [4.19] [4.38]-0.238*** -0.110* -0.100* -0.070
[3.98] [1.93] [1.72] [1.23]-0.388** -0.145 -0.153 -0.137
[2.46] [0.99] [1.05] [0.97]Election-Year(parliamentary) 0.018 0.019
[0.95] [0.91]Presidential-System 0.021 0.020
[1.11] [0.99]0.239***
[5.46]Constant 0.139*** 0.127*** 0.048 0.153*** 0.053 0.036 0.029
[4.24] [4.18] [1.37] [4.86] [1.52] [1.00] [0.78]
Observations 336 336 336 311 311 311 234R-squared 0.02 0.17 0.21 0.09 0.26 0.26 0.43
Number of n 19 19 19 18 18 18 17
Notes: Absolute value of t statistics in brackets; * significant at 10%; ** significant at 5%; *** significant at 1%.
31
Table 4. Regression Results. Dependent variable: -scale privatization. FGLS with random effects.
(1) (2) (3) (4) (5) (6) (7)
Ideology -0.017* -0.025** -0.008 -0.020* -0.011 -0.010 -0.014[1.68] [2.43] [0.67] [1.81] [0.80] [0.73] [1.04]
Dummy till 1994 0.107*** 0.258*** 0.238*** 0.248*** 0.263***[5.14] [3.80] [3.27] [3.26] [3.10]
Ideology*Dummy till 1994 -0.050** -0.042* -0.045* -0.031
[2.34] [1.82] [1.87] [1.09]-0.102 -0.005 -1×10-4 -0.003[1.60] [0.08] [0.00] [0.04]-0.244 -0.073 -0.077 -0.073[1.45] [0.44] [0.46] [0.48]
Election-Year(parliamentary) 0.003 0.019
[0.16] [0.84]Presidential-System 0.010 0.008
[0.47] [0.36]0.213***
[4.49]Constant 0.120*** 0.112*** 0.066* 0.128*** 0.071* 0.064 0.060
[3.80] [3.67] [1.82] [3.81] [1.82] [1.55] [1.48]
Observations 336 336 336 311 311 311 234R-squared 0.01 0.08 0.10 0.03 0.10 0.10 0.26
Number of n 19 19 19 18 18 18 17
Notes: Absolute value of t statistics in brackets; * significant at 10%; ** significant at 5%; *** significant at 1%.
32
Table 5. Dates of the Countries’ Independence
Country Date Civil War
Albania March 1991 None
Bosnia and Herzegovina April 1992 (from Yugoslavia)March 1992 till November 1995
Bulgaria June 1990 None
Croatia June 1991 (from Yugoslavia)March 1992 till November 1995
Czech Republic December 1989 None
Estonia August 1991 (from the Soviet Union) None
Hungary October 1989 None
Latvia September 1991 (from the Soviet Union) None
Lithuania March 1990 (from the Soviet Union) None
Macedonia September 1991 (from Yugoslavia)March 1992 till November 1995
Moldova August 1991 (from the Soviet Union) None
Montenegro June 2006 (from Serbia)March 1992 till November 1995
Poland December 1989 None
Romania December 1989 None
Russia June 1991 (from the Soviet Union) None
Serbia April 1992 (from Yugoslavia)March 1992 till November 1995
Slovak Republic December 1989 None
Slovenia June 1991 (from Yugoslavia)June till July 1991 (tendays)
Ukraine August 1991 (from the Soviet Union) None
Note: We count independence for Albania and Bulgaria as of the first free elections.
33
Table 6. Regression Results. Individual five year Windows of Opportunity after Independence.FGLS with random effects.
(1) (2) (3) (4) (5) (6)
DependentVariable EBRD
AllEBRD
AllEBRDSmall
EBRDSmall
EBRDLarge
EBRDLarge
Ideology -0.009 -0.007 -0.007 -0.009 -0.004 -0.006[0.90] [0.58] [0.65] [0.72] [0.41] [0.50]
Window ofOpportunity 0.307*** 0.311*** 0.406*** 0.446*** 0.293*** 0.276***
[4.68] [3.98] [5.96] [5.86] [4.34] [3.50]Ideology*
Window ofOpportunity
-0.060*** -0.061** -0.079*** -0.084*** -0.047** -0.032
[2.75] [2.25] [3.50] [3.20] [2.09] [1.17]-0.037 -0.094 -0.005[0.63] [1.62] [0.09]
yment 0.032 -0.111 -0.021[0.21] [0.77] [0.14]
Election-Year(parliamentary) -0.035* 0.007 0.007
[1.65] [0.34] [0.31]Presidential-System -0.006 0.003 0.003
[0.28] [0.14] [0.15]0.153*** 0.213*** 0.191***
[3.36] [4.77] [4.14]Constant 0.049 0.047 0.032 0.032 0.041 0.036
[1.53] [1.24] [0.95] [0.84] [1.22] [0.92]
Observations 336 234 336 234 336 234R-squared 0.16 0.23 0.23 0.42 0.18 0.30
Number of n 19 17 19 17 19 17
Notes: Absolute value of t statistics in brackets; * significant at 10%; ** significant at 5%; *** significant at 1%.
34
Table 7. Regression Results. Individual five year Windows of Opportunity after Independence considering the civil war in Yugoslavia. FGLS with random effects.
(1) (2) (3) (4) (5) (6)
DependentVariable EBRD
AllEBRD
AllEBRDSmall
EBRDSmall
EBRDLarge
EBRDLarge
Ideology -0.008 -0.012 -0.007 -0.012 -0.006 -0.015[0.71] [0.94] [0.61] [0.94] [0.57] [1.13]
Window ofOpportunity(Civil War)
0.303*** 0.266*** 0.376*** 0.435*** 0.231*** 0.191**
[4.60] [3.45] [5.46] [5.79] [3.35] [2.39]Ideology*
Window ofOpportunity(Civil War)
-0.064*** -0.051* -0.074*** -0.086*** -0.035 -0.017
[2.89] [1.96] [3.23] [3.38] [1.49] [0.61]-0.056 -0.113* -0.038[0.93] [1.93] [0.61]-0.003 -0.143 -0.084[0.02] [0.98] [0.54]
Election-Year(parliamentary) -0.028 0.018 0.015
[1.27] [0.85] [0.65]Presidential-System -0.009 -0.001 -0.004
[0.44] [0.03] [0.16]0.147*** 0.203*** 0.189***
[3.16] [4.48] [3.91]Constant 0.046 0.064 0.033 0.041 0.051 0.069*
[1.39] [1.60] [0.95] [1.06] [1.49] [1.66]
Observations 336 234 336 234 336 234R-squared 0.14 0.20 0.21 0.39 0.13 0.23
Number of n 19 17 19 17 19 17
Notes: Absolute value of t statistics in brackets; * significant at 10%; ** significant at 5%; *** significant at 1%.
35
Table 8. Robustness Checks. Panel corrected standard errors.
(1) (2) (3) (4) (5) (6)
DependentVariable EBRD
AllEBRD
AllEBRDSmall
EBRDSmall
EBRDLarge
EBRDLarge
Ideology -0.012 -0.012 -0.010 -0.009 -0.008 -0.013[1.22] [1.18] [0.85] [0.83] [0.68] [1.17]
Dummy till 1994 0.316*** 0.329*** 0.415*** 0.553*** 0.268*** 0.291***[5.60] [4.55] [6.10] [6.97] [4.19] [3.76]
Ideology*Dummy till 1994 -0.064*** -0.066*** -0.085*** -0.115*** -0.052** -0.038
[3.65] [2.67] [3.82] [4.32] [2.55] [1.43]-0.038 -0.071 -0.009[0.65] [1.09] [0.14]0.011 -0.142* -0.044[0.14] [1.89] [0.41]
Election-Year(parliamentary) -0.017 0.018 0.021
[0.82] [0.90] [0.98]Presidential-System 0.003 0.021 0.007
[0.16] [1.11] [0.36]0.182*** 0.253*** 0.244***
[3.47] [5.71] [5.05]Constant 0.061** 0.058* 0.048 0.028 0.064* 0.056
[2.09] [1.83] [1.40] [0.81] [1.89] [1.58]
Observations 336 234 336 234 336 234R-squared 0.18 0.27 0.21 0.47 0.12 0.33
Number of n 19 17 19 17 19 17
Notes: Absolute value of t statistics in brackets; * significant at 10%; ** significant at 5%; *** significant at 1%.
36
Table 9. Robustness Checks. FGLS with random effects and an AR(1) disturbance.
(1) (2) (3) (4) (5) (6)
DependentVariable EBRD
AllEBRD
AllEBRDSmall
EBRDSmall
EBRDLarge
EBRDLarge
Ideology -0.012 -0.009 -0.011 -0.009 -0.009 -0.014[0.95] [0.57] [0.83] [0.76] [0.67] [0.95]
Dummy till 1994 0.327*** 0.350*** 0.414*** 0.549*** 0.250*** 0.238***[4.71] [3.80] [5.75] [6.95] [3.53] [2.61]
Ideology*Dummy till 1994 -0.067*** -0.072** -0.087*** -0.115*** -0.049** -0.026
[3.08] [2.35] [3.85] [4.34] [2.19] [0.87]-0.02 -0.07 0.004[0.32] [1.22] [0.06]-0.003 -0.136 -0.107[0.02] [0.95] [0.64]
Election-Year(parliamentary) -0.033 0.019 0.016
[1.56] [0.92] [0.74]Presidential-System 0.004 0.02 0.009
[0.15] [0.96] [0.34]0.157*** 0.236*** 0.189***
[3.43] [5.40] [4.03]Constant 0.059 0.052 0.050 0.03 0.068* 0.064
[1.58] [1.10] [1.29] [0.78] [1.78] [1.40]
Observations 336 234 336 234 336 234R-squared 0.15 0.22 0.20 0.43 0.10 0.26
Number of n 19 17 19 17 19 17
Notes: Absolute value of t statistics in brackets; * significant at 10%; ** significant at 5%; *** significant at 1%.
37
Table 10. Robustness Checks. Including Nationalist-Dummy. FGLS with random effects.
(1) (2) (3) (4) (5) (6)
DependentVariable EBRD
AllEBRD
AllEBRDSmall
EBRDSmall
EBRDLarge
EBRDLarge
Ideology -0.012 -0.010 -0.010 -0.004 -0.015 -0.014[0.84] [0.70] [0.82] [0.31] [0.95] [0.94]
Nationalist-Dummy -0.002 -0.002 -0.019 -0.004 -0.015 0.013[0.07] [0.06] [0.60] [0.12] [0.37] [0.34]
Dummy till 1994 0.361*** 0.266** 0.528*** 0.489*** 0.263*** 0.224**[4.02] [2.46] [6.95] [5.43] [2.76] [1.99]
Ideology*Dummy till 1994 -0.066** -0.053 -0.106*** -0.106*** -0.041 -0.027
[2.31] [1.56] [4.40] [3.75] [1.35] [0.77]Nationalist-Dummy*
Dummy till 1994 -0.051 0.065 -0.245*** -0.204*** -0.027 0.029
[0.62] [0.77] [3.53] [2.91] [0.31] [0.34]-0.266 -0.501*** -0.174[1.44] [3.25] [0.91]0.274 0.148 -0.345[0.81] [0.52] [0.98]
Election-Year(parliamentary) -0.024 0.028 -0.006
[0.95] [1.31] [0.23]Presidential-System -0.005 0.007 0.003
[0.20] [0.31] [0.11]0.127** 0.119*** 0.181***[2.31] [2.61] [3.19]
Constant 0.057 0.068 0.054 0.042 0.091** 0.078*[1.30] [1.49] [1.45] [1.10] [1.97] [1.65]
Observations 225 188 225 188 225 188R-squared 0.19 0.21 0.35 0.46 0.13 0.25
Number of n 17 16 17 16 17 16
Notes: Absolute value of t statistics in brackets; * significant at 10%; ** significant at 5%; *** significant at 1%.
38
Table 11. Robustness Checks. KOF Index of Globalisation and Government Expenditures.FGLS with random effects.
(1) (2) (3) (4) (5) (6)
DependentVariable EBRD
AllEBRD
AllEBRDSmall
EBRDSmall
EBRDLarge
EBRDLarge
Ideology -0.005 -0.007 0.001 -0.004 -0.008 -0.014[0.34] [0.51] [0.10] [0.34] [0.48] [0.95]
Dummy till 1994 0.333*** 0.433*** 0.350*** 0.380*** 0.239** 0.229**[3.49] [4.24] [4.29] [4.11] [2.40] [2.10]
Ideology*Dummy till 1994 -0.064** -0.108*** -0.078*** -0.078** -0.051 -0.016
[2.06] [3.12] [2.94] [2.48] [1.57] [0.44]-0.213 -0.443** -0.780*** -0.733*** -0.511*** 0.074[1.12] [2.09] [4.80] [3.82] [2.58] [0.33]0.317 0.45 0.187 -0.145 -0.082 -0.271[0.94] [1.40] [0.65] [0.50] [0.23] [0.79]
Election-Year(parliamentary) -0.054** -0.029 -0.006 -0.006 -0.019 0.011
[2.17] [1.19] [0.28] [0.27] [0.75] [0.41]Presidential-System -0.017 0.006 0.01 -0.007 -0.013 0.024
[0.59] [0.24] [0.43] [0.29] [0.44] [0.87]
Globalisation 0.472* 0.426** 0.244
[1.87] [1.97] [0.93]0.014 0.254*** 0.189***[0.22] [4.48] [2.83]
Expenditures 0.057 0.039 0.043
[0.52] [0.40] [0.37]Constant 0.053 0.071 0.037 0.058 0.085 0.063
[1.04] [1.63] [0.85] [1.46] [1.61] [1.34]
Observations 203 185 203 185 203 185R-squared 0.23 0.41 0.25 0.44 0.19 0.23
Number of n 14 16 14 16 14 16
Notes: Absolute value of t statistics in brackets; * significant at 10%; ** significant at 5%; *** significant at 1%.
39
Table 12. Robustness Checks. Dummy Former Soviet States and Dummy Prospected EU-Membership. FGLS with random effects.
(1) (2) (3) (4) (5) (6)
DependentVariable EBRD
AllEBRD
AllEBRDSmall
EBRDSmall
EBRDLarge
EBRDLarge
Ideology -0.009 -0.011 -0.009 -0.008 -0.013 -0.013[0.69] [0.83] [0.74] [0.65] [1.00] [1.00]
Dummy till 1994 0.362*** 0.325*** 0.536*** 0.483*** 0.261*** 0.197**[4.32] [3.54] [6.69] [5.54] [2.99] [2.08]
Ideology*Dummy till 1994 -0.067** -0.066** -0.118*** -0.106*** -0.032 -0.022
[2.41] [2.36] [4.47] [4.00] [1.11] [0.75]-0.039 -0.028 -0.063 -0.071 -0.001 -0.004[0.65] [0.47] [1.09] [1.25] [0.02] [0.06]0.007 0.004 -0.132 -0.12 -0.072 -0.062[0.04] [0.03] [0.93] [0.85] [0.46] [0.41]
Election-Year(parliamentary) -0.028 -0.027 0.02 0.017 0.019 0.018
[1.33] [1.26] [0.97] [0.84] [0.85] [0.80]Presidential-System 0.004 -0.003 0.019 0.029 0.008 -0.007
[0.18] [0.10] [0.91] [1.02] [0.34] [0.24]0.172*** 0.169*** 0.238*** 0.236*** 0.213*** 0.211***
[3.75] [3.67] [5.43] [5.41] [4.46] [4.47]Dummy Former
Soviet State 0.013 0.001 0.001
[0.57] [0.06] [0.04]Dummy till 1994*Dummy Former
Soviet State-0.081 0.063 0.013
[1.63] [1.32] [0.25]Dummy Prospected
EU-Membership -0.011 -0.005 -0.035
[0.39] [0.17] [1.19]Dummy till 1994*
Dummy Prospected EU-Membership
0.014 0.083* 0.078
[0.29] [1.79] [1.54]Constant 0.048 0.064 0.028 0.025 0.059 0.081*
[1.14] [1.42] [0.71] [0.59] [1.36] [1.75]
Observations 234 234 234 234 234 234R-squared 0.23 0.22 0.44 0.44 0.26 0.27
Number of n 17 17 17 17 17 17
Notes: Absolute value of t statistics in brackets; * significant at 10%; ** significant at 5%; *** significant at 1%.
40
Table 13. Robustness Checks. Number of World Bank Projects and IMF Programs agreed. FGLS with random effects.
(1) (2) (3) (4) (5) (6)
DependentVariable EBRD
AllEBRD
AllEBRDSmall
EBRDSmall
EBRDLarge
EBRDLarge
Ideology -0.0082 -0.0055 -0.0055 -0.0065 -0.014 -0.0071[0.57] [0.41] [0.40] [0.52] [0.93] [0.52]
Dummy till 1994 0.2901*** 0.2913*** 0.4754*** 0.4824*** 0.2411** 0.2307***[3.17] [3.34] [5.45] [5.87] [2.52] [2.60]
Ideology*Dummy till 1994 -0.0626** -0.0651** -0.1118*** -0.1127*** -0.0294 -0.0276
[2.11] [2.30] [3.94] [4.23] [0.95] [0.96]-0.3907** -0.2757 -0.5947*** -0.5952*** -0.0505 -0.0607
[2.11] [1.61] [3.37] [3.70] [0.26] [0.35]0.3226 0.0814 -0.2266 -0.1465 -0.2578 0.0477[0.98] [0.42] [0.72] [0.80] [0.75] [0.24]
Election-Year(parliamentary) -0.0346 -0.029 0.0157 0.0142 0.0243 0.023
[1.40] [1.26] [0.67] [0.65] [0.94] [0.98]Presidential-System -0.0193 -0.0084 0.0011 0.0026 0.0086 0.0034
[0.73] [0.36] [0.04] [0.12] [0.31] [0.14]0.1435*** 0.1427*** 0.2052*** 0.2093*** 0.1971*** 0.1886***
[2.85] [2.95] [4.28] [4.59] [3.75] [3.83]Number of World Bank
Projects agreed 3×10-5 -0.0004 -0.0072
[0.01] [0.10] [1.54]IMF Arrangement in
effect for at least 5 months in a particular year
0.0458** 0.0025 0.0406*
[2.00] [0.11] [1.70]IMF Arrangement agreed 0.0470** -0.012 0.0862***
[2.06] [0.56] [3.71]Constant 0.0502 0.0478 0.0486 0.0546 0.0632 0.028
[1.08] [1.12] [1.09] [1.36] [1.30] [0.65]
Observations 200 217 200 217 200 217R-squared 0.26 0.24 0.46 0.46 0.27 0.30
Number of n 16 17 16 17 16 17
Notes: Absolute value of t statistics in brackets; * significant at 10%; ** significant at 5%; *** significant at 1%.