Journal of Comparative Economics 32 (2004) 720–744
www.elsevier.com/locate/jce
Public support for creating a market economy
in Eastern Europe
Bernd Hayo
Philipps-University Marburg, Department of Economics (FB 02), D-35032 Marburg, Germany
ZEI, University of Bonn, Germany
Received 11 September 2002; revised 22 July 2004
Hayo, Bernd—Public support for creating a market economy in Eastern Europe
Employing two large databases, we analyze the determinants of public support for the creation
of a market economy in Eastern Europe. From a macroeconomic perspective inflation, unemploy-
ment, privatization, and enterprise restructuring reduce this support; alternatively, democratization,
the creation of working financial markets, and foreign aid per capita increase support for the market.
Across countries, higher inequality undermines market support. From a microeconomic perspective,
labor market status, both the objective and the subjective economic situations of a person, political
orientation, and the socio-demographic background of the respondent affect support for the market
economy. For example, unemployed, relatively poor, older, female, and less-educated respondents
living in rural areas are less inclined to favor the creation of a market economy. Journal of Compara-
tive Economics 32 (4) (2004) 720–744. Philipps-University Marburg, Department of Economics (FB
02), D-35032 Marburg, Germany; ZEI, University of Bonn, Germany.
 2004 Association for Comparative Economic Studies. Published by Elsevier Inc. All rights re-
served.
JEL classification: P1; P2; O52
Keywords: Economic transition; Market economy; Eastern Europe; Public support
E-mail address: hayo@wiwi.uni-marburg.de.
0147-5967/$ – see front matter  2004 Association for Comparative Economic Studies. Published by Elsevier
Inc. All rights reserved.
doi:10.1016/j.jce.2004.07.003

---

B. Hayo / Journal of Comparative Economics 32 (2004) 720–744
721
1. Introduction
Public support for the creation of a market economy is key to a successful transforma-
tion program. Williamson (1994) identifies strong political support for the reformer as one
of the three most important ingredients.1 Although not a sufficient condition, public sup-
port appears to increase the chance that reforms will be feasible and successful (Stokes,
1996). Assessments of the actual progress in the Eastern European economic transitions,
e.g. EBRD (1999) or IMF (2000), do not provide much information on determinants of
public support for a market economy. On the theoretical side, Rodrik (1996), Drazen (2000)
and Roland (2002) survey the growing body of work on economic policy reform, in which
public support for the reforms, either directly or indirectly, is crucial. However, most the-
oretical models simply assume that certain variables, especially unemployment and per
capita income, are important determinants of public support for reforms, without provid-
ing the empirical evidence to support these assertions.
Our analysis investigates the determinants of public attitudes towards creating a market
economy by employing survey data collected in Eastern European countries, which are in
the process of transforming their centrally planned economies into Western-style market
economies. While no general theory of transition is available, Blanchard (1997) presents a
partial equilibrium model of economic transformation. His model consists of a transition
economy in which decision-makers have perfect foresight to capture the stylized fact of
a U-shaped evolution of output and employment in most Eastern European economies.
The author assumes that public support for economic reforms is also U-shaped. Blanchard
provides survey evidence from Poland regarding the determinants of the perception of both
the current and the expected economic situations of people and shows that unemployment
and output do affect attitudes. However, if the democratic political process were to be
endogenized, Blanchard (1996) demonstrates that the chosen reform path might not be
achievable.
Fidrmuc (2000) investigates public support for the market in Eastern Europe empiri-
cally, but indirectly. He uses election results from four countries to analyze political support
by explaining voting shares for reform and non-reform parties using various regressors,
e.g., unemployment, entrepreneurial activity, and demographic factors. Actual votes may
indicate true revealed preferences rather than only the intentions reported by respondents
in surveys. However, voting for a party cannot be attributed easily to only one policy issue.
Therefore, it is not clear whether voters prefer a party because of its stance on economic
reform or because of its position on maintaining order, improving democracy, or national
defense. Warner (2001) investigates the relationship between the intensity of market re-
forms and electoral support for reform parties in Russian regions from 1990 to 1995. He
finds that reform parties garnered more support in the 1995 parliamentary elections in those
regions where greater progress was made in structural reforms.
In principle, public attitudes towards creating a market economy can be measured di-
rectly from opinion polls. Early attempts by Shiller et al. (1991, 1992) to compare Eastern
European attitudes with those of Western countries are interesting but limited to studying
1 The other two listed conditions for successful reforms are a visionary leadership and a coherent economic
team.

---

722
B. Hayo / Journal of Comparative Economics 32 (2004) 720–744
whether Eastern European opinions during times of transformation contain an attitudinal
legacy from Communist times.2 Concentrating on the labor market and taking into ac-
count a time dimension, Blanchflower abd Freeman (1997) compare the market attitudes
of respondents in Eastern and Western Europe using the large International Social Sur-
vey Programme (ISSP) database. In contrast to Shiller et al., these authors find evidence
of an attitudinal legacy from the Communist times.3 In comparing the development in
Poland and Russia, Shleifer (1997) incorporates possible differences in trust and partici-
pation in civic activities, which he computes using the World Values Survey. In his view,
the emerging differences in social capital are not a key determinant of differences in eco-
nomic performance. Eble and Koeva (2002) investigate attitudes towards market reforms
in Russia after the crisis in 1998 using survey data.4 They find that individual attitudes
towards reforms are affected by the personal experiences of the respondents during the
transition process. In addition, groups that are distinguished by education and age appear
to be more flexible in adjusting to new situations. Finally, these authors conclude that re-
spondents from Russian regions in which a high percentage of workers have received no
wage payments within the last month are not supportive of market reforms.
Our paper investigates public support for creating a market economy empirically us-
ing two representative survey databases, namely the Central and Eastern Eurobarometers
(CEEB) and the New Democracy Barometers (NDB). Both surveys ask about the respon-
dent’s attitude towards the creation of the market but not about particular reform strategies
that will lead to a market system. We employ static and dynamic panel data methods in the
macroeconomic level analysis and cross-section regressions in the microeconomic level
analysis. The paper is structured as follows. In Section 2, the database and econometric
methodology are described. Theoretical reasons for an aggregate analysis of support for a
market economy are summarized in Section 3 and the corresponding empirical results are
shown in Section 4. Section 5 presents microeconomic level analyses using the NDB and
CEEB data. The final section concludes with policy implications.
2. The databases
The Central and Eastern Eurobarometers (CEEB) database, which is collected on behalf
of the European Commission, is used for macroeconomic analysis because the number of
countries and the collection of repeated cross-sections over time make it possible to con-
struct a panel data set using average national values. In general, about 1000 people in
each country were selected randomly for a personal interview in the autumn of the respec-
tive year. The first surveys in 1990 were undertaken only in Czechoslovakia, Hungary and
Poland, while the most extensive survey from 1996 provides data on twenty countries.5
2 This idea is related to the importance of initial economic and political conditions to the success of the
transformation (de Melo et al., 2001).
3 Frentzel-Zagorska and Zagorski (1993) provide survey evidence from Poland on this issue.
4 Finifter and Mickiewicz (1992) document survey evidence from the former USSR.
5 The primary data are available from the Zentralarchiv für Empirische Sozialforschung (ZA) in Cologne
(http://www.za.uni-koeln.de/index-e.htm). This survey project ended in 1997.

---

B. Hayo / Journal of Comparative Economics 32 (2004) 720–744
723
Single CEEB surveys have been aggregated to cover the time period from 1990 to 1997
and include data on 21 countries.6 Hence, the aggregated data base contains responses from
about 120,000 people.
Using the CEEB data, Kim and Pirttilä (2003) find evidence that countries showing
greater public support for market reforms also achieve higher growth rates eventually. To
control for possible endogeneity, we use the general method of moment (GMM) estimation
technique. The CEEB data have problems concerning the consistent and regular coverage
of important questions across different surveys because many interesting questions are
eliminated over time. In addition, only a few demographic variables are collected over
time and some countries are also eliminated.
The New Democracy Barometers (NDB) database, which was initiated by the Centre
for Public Policy at Strathclyde University in Glasgow, is used for microeconomic analysis
of a cross-section in 1995. Because it contains a rich set of individual-level questions, we
can control explicitly for several potentially important factors. On the downside, the NDB
covers only seven countries, namely the Czech Republic, the Slovak Republic, Slovenia,
Hungary, Poland, Romania, and Bulgaria, and contains only one relevant round of surveys
in 1995 for our question of interest. The NDB is based on personal interviews and designed
to be a random sample of the population (Rose and Haerpfer, 1993).7 Although limited
in terms of available variables, the CEEB database can also be used in microeconomic
analysis to investigate the robustness of the results from the NBD. However, we must be
confident that we are measuring similar things in the two databases.
The CEEB database focuses on one issue, namely a person’s opinion about the creation
of a market economy. The specific question is:
Do you personally feel that the creation of a free market economy, that is one largely
free from state control, is right or wrong for (our country’s) future?
The answers to that question are coded into three categories, namely ‘wrong’ (−1),
‘don’t know’ (0), and ‘right’ (1), to form the dependent variable called SUPPORTCEEB.
To get aggregate values, we compute the mean values of SUPPORTCEEB for every coun-
try at each point in time.8 In the case of NDB, we create a variable measuring market
support that may be advantageous methodologically because it measures the concepts of
interest directly. The notion of a market economy is not defined easily so that capturing it
by asking one specific question is problematic. The NDB uses four variables related to core
aspects of a market economy. The questions investigate whether people prefer differential
or equal incomes, private or state property, high pay but job risk or a secure job, and many
goods but high prices or price controls. Considering more than one indicator increases the
6 Observations from Albania, Armenia, Belarus, Bulgaria, Croatia, Czech Republic, Estonia, Georgia,
Hungary, Kazakhstan, Latvia, Lithuania, Macedonia, Moldova, Poland, Romania, Russia, Slovakia, Slovenia,
Ukraine, Yugoslavia are included.
7 The primary data for the NDB are not accessible for researchers who are not members of the Citizens in
Transition Network (CITNET).
8 Keeping the ‘don’t know’ responses is highly desirable for maintaining representativeness of the sample.
However, forming a dichotomous dependent variable by dropping the ‘don’t know’ category yields similar results.

---

724
B. Hayo / Journal of Comparative Economics 32 (2004) 720–744
chances that we are measuring actual attitudes towards the market in general and not just
perceptions about a single aspect of the market economy.
Taking into account these considerations, we condense the information from the NDB
through factor analysis. As Appendix Table 1 in the reports, exactly one factor can be
extracted and all of the factor loadings are larger than the usual threshold of 0.5. We inter-
pret the factor extracted from the four questions as indicating public support for the market
economy and denote it SUPPORTNDB. Moreover, this factor is valid not only at the aggre-
gate level but also at the national level. The Spearman rank correlation coefficient between
the averages of SUPPORTCEEB and SUPPORTNDB based on the ten countries covered in
the NDB in 1995 is 0.72, which is significant at a 5% level. Hence, we find strong support
that the two variables from different data sources measure a similar concept.
Figure 1 plots public support for the market economy over time using averages of SUP-
PORTCEEB for the countries in the sample. These values can be interpreted as the share
Fig. 1. Public support for the market economy across countries: mean values for SUPPORTCEEB.

---

B. Hayo / Journal of Comparative Economics 32 (2004) 720–744
725
Fig. 2. The aggregate evolution of support for the market economy.
of net supporters, i.e. supporters minus opponents in relation to all respondents, of the cre-
ation of a market economy.9 In the next section, the apparent differences across countries
are investigated statistically, conditional on the influence of other explanatory variables.
To generate information about the overall time trend for market support in Eastern Europe,
two regressions are run to analyze market support first by year dummies only and then
by year and country dummies. The resulting coefficients for the year dummies are plotted
in Fig. 2; both series indicate that support was at a minimum in 1994. Although the co-
efficients obtained from the regression having only year dummies are consistent with the
prediction of Blanchard (1997) that support for the market is U-shaped over time, the series
based on the regression including year and country dummies is multi-peaked.10
3. Theoretical determinants of support for the market at the aggregate level
The relevant variables for investigating public support at the aggregate level are listed in
Table 1 with descriptive statistics and expected effects included. These variables are chosen
based on the literature and divided into categories of indicators, namely, macroeconomic
indicators, standard of living, fiscal policy, transition progress, external influences, political
institutions, and political business cycles. Lipton and Sachs (1990) claim that fundamental
economic reform involves three core elements, namely, macroeconomic stabilization, eco-
nomic liberalization, and privatization of state enterprises. Williamson (1994) provides a
similar list, denoting it as the Washington Consensus. In addition, Drazen (2000) stresses
the importance of political institutions and processes in economic policymaking to justify
the inclusion of political institution and indicators of the political business cycle.
9 Standard deviations to measure dispersion are also computed to investigate the consensus of opinion within
society. The average of SUPPORTCEEB and its standard deviation are negatively correlated with a coefficient of
−0.60 that is significant at a 1% level. Hence, the lower is the degree of support for market reforms, the larger is
the disagreement in society on this issue.
10 Because not all Eastern European countries started economic reforms at the same time, it may be more
meaningful to include a proxy for transformation or stabilization time. Therefore, we constructed transformation
time trends that count years since the start of economic stabilization programs (Fischer et al., 1998). However,
a transformation trend variable and its square are both less significant than the year dummies (Hayo, 1999b).

---

726
B. Hayo / Journal of Comparative Economics 32 (2004) 720–744
Table 1
Aggregate variables: sources, descriptive statistics, and priors
Variable name
Definition
Source
Mean
Std. dev.
Expected
sign
Dependent variable
SUPPORTCEEB
Country average of individual
CEEB
0.12
0.28
level attitudes towards creating
a market economy
Macroeconomic indicators
DGDP
GDP growth, in %
EBRD
−0.22
7.69
+
UNEMP
Unemployment rate, in %
EBRD
10.14
5.18
−
INFLAT
Inflation rate, in % p.a.
EBRD
260.20
780.14
?
FXRATE
Change in log of exchange rate
IMF
192.65
676.00
?
against SDR, in %
Standard of living indicators
GDPCAP
GDP per capita, in constant
WDI
2830.30
2157.90
+
US dollars (base: 1995)
RADIOS
Number of radios in the
WDI
473.75
178.68
+
country (per 1000 people)
LIFEEXP
Life expectancy
WDI
70.08
2.47
+
GINI1
Gini coefficient
Milanovic
34.36
11.07
−
GINI2
Gini coefficient (adding WDI
Milanovic,
32.35
8.26
−
Gini coefficients of countries
WDI
not in GINI1)
Fiscal policy indicators
GOVGDP
Ratio of government
EBRD
41.67
8.59
?
expenditure to GDP, in %
GOVDEF
Ratio of government surplus to
EBRD
−4.95
6.95
?
GDP, in % (deficit = negative
value)
Transition indicators
ENTERPRISE
Average of large-scale
EBRD
2.68
0.76
+
privatization, small scale
privatization, and enterprise
restructuring
MARKETS
Average of price liberalization,
EBRD
2.84
0.47
+
trade and foreign exchange
system, and competition policy
FINANCE
Average of banking reform and
EBRD
2.16
0.63
+
interest rate liberalization, and
securities markets and
non-bank financial institutions
External indicators
OPENNESS
Ratio of trade in goods and
WDI
92.39
34.33
−
services to GDP, in %
AIDGNP
Foreign aid in relation to GNI,
WDI
2.04
3.59
+
in %
AIDCAP
Foreign aid per capita, in US
WDI
26.30
20.87
+
dollars
FDICAP
Net foreign direct investment
EBRD
571.00
833.08
?
per capita, in US dollars
(continued on next page)

---

B. Hayo / Journal of Comparative Economics 32 (2004) 720–744
727
Table 1 (continued)
Variable name
Definition
Source
Mean
Std. dev.
Expected
sign
IMFSUPPORT
Dummy variable: IMF
IMF
0.38
0.49
+
program running at the time of
the survey
Political institutions indicators
EXREC
Executive recruitment
Polity
7.45
1.25
+
EXCONST
Executive constraints
Polity
5.95
1.48
+
POLCOMP
Political competition
Polity
8.04
1.88
+
POLCONV
Number of independent
Henisz
0.54
0.29
−
branches of government with
veto power over policy change
including the judiciary and
sub-federal units
Political business cycle indicators
ELECTYEAR
Dummy variable, taking the
Fidrmuc
0.39
0.49
+
value 1 if an election occurs in
that particular year
Notes. Only 78 observations are used to compute the descriptive statistics that correspond to those employed in
the analysis in Table 2.
Sources. CEEB: Central and Eastern Eurobarometers survey database. EBRD: Transition Report. IMF:
International Financial Statistics. IMF program constructed using information from http://www.imf.org/
external/country/. WDI: World Bank Development Indicators CD-Rom 2002. Milanovic: Branco Milanovic’s es-
timates http://www.worldbank.org/research/transition/house.htm). Polity: Polity IV database (http://www.cidcm.
umd.edu/inscr/polity/). Henisz: Henisz (2000). Fidrmuc: We extend special thanks to Jan Fidrmuc for providing
these data.
For macroeconomic indicators, we take inflation, exchange rates, unemployment, and
GDP growth. Typically, the IMF proposes stabilization programs to reduce high rates
of inflation. Moreover, in survey data, people often express serious concern about infla-
tion (Di Tella et al., 2001; Fischer and Huizinga, 1982; Hayo, 1998a, and Rose, 1998).
However, neither theoretical nor empirical evidence of negative economic effects of mod-
erate inflation is compelling, as Driffil et al. (1990) and Bruno and Easterly (1998) attest.
Nonetheless, inflation rates in excess of 1000 percent per year have been observed in some
of the transition countries, e.g., Armenia, Belarus, and Kazakhstan. Thus, we conjecture
that inflation may influence public support for the creation of a market economy in a non-
linear fashion.
For transition countries, Fischer et al. (1996a, 1996b) argue that eliminating high in-
flation is a requirement for economic growth. Yavlinsky and Braguinsky (1994) think that
alternative policies, e.g., de-monopolization, should be addressed before monetary policy
is tightened. Finally, Mondino et al. (1996) present a model of an inflation cycle, in which
inflation goes down rapidly after the adoption of a stabilization program but increases
again when the reform is abandoned. In their model, the inflation rate drives public support
for reforms and, as soon as the inflation rate is reduced significantly, support for the pro-
gram collapses. Thus, evidence of the actual effects of inflation on public opinion towards
economic reforms may contribute to an assessment of these conflicting hypotheses. For
robustness, we consider the exchange rate as another indicator of inflationary conditions.

---

728
B. Hayo / Journal of Comparative Economics 32 (2004) 720–744
Hence, we designate an ambiguous expected sign for inflation and the exchange rate in
Table 1.
Unemployment is included to capture the labor market situation, following the theoreti-
cal models of Blanchard (1997), Fidrmuc (1999), and Rodrik (1995). However, measuring
unemployment in transition countries may be unreliable (UN, 1997, p. 114f).11 Nonethe-
less, we expect higher unemployment to lead to less support for reforms so we report a
negative expected sign for this variable in Table 1. In addition, we consider GDP growth
because it contains information about the dynamic effects of income evolution in the re-
spective economies. Typically, living standard is proxied by income per capita, which is an
important component in many theoretical models of public support for reforms. Alterna-
tive indicators to national accounts data for measuring the standard of living, namely life
expectancy and the number of radios in a country, are included to assess the robustness of
our findings. The growth of GDP and the three standard of living indicators are expected
to have a positive effect on support for reforms. In addition Drazen (2000) argues that the
distributional consequences of economic reforms may lead to strong opposition to the cre-
ation of the market. Thus, we include the Gini index as a measure of income inequality and
expect the sign of the coefficient to be negative.
Although IMF programs emphasize a reduction in both the share of government expen-
ditures in GDP and budget deficits, East European countries may find it beneficial to incur
some fiscal debt during the transition period. Fernandez and Rodrik (1991) show that the
existence of ex ante uncertainty about winners and losers from the reforms may prevent
the implementation of efficiency-enhancing reforms, even though people would have sup-
ported the reforms ex post. Hence, if people can be assured that they will benefit from the
reforms, even if they turn out to be losers according to the previous distribution of income,
the likelihood of public support for the reform will be higher. Alesina and Drazen (1991)
make a similar argument based on a war of attrition. Since fiscal budgets are dominated
by social spending, a large share of government expenditures with a resulting high bud-
get deficit mitigates the adverse distributional consequences of the transformation. Due
to countervailing tendencies, we report ambiguous signs for the fiscal policy indicators in
Table 1.
The EBRD computes indices to reflect the extent of small-scale and large-scale privati-
zation, enterprise restructuring, and price liberalization. In addition, progress in liberalizing
trade and foreign exchange, the development of competition policy, and the reform of banks
and other financial institutions are measured. Stiglitz (1999) argues that privatization has
not been successful in most East European countries because the problems encountered in
this process have a negative impact on public support for market reforms. Nevertheless, we
expect improvements measured by the three aggregate EBRD transition indicators to have
a positive impact on support for reforms as Table 1 records.
Regarding external influences, Krueger (1993) shows that certain trade policies may
have asymmetric effects on different groups within an economy and, thus, affect the bal-
ance of political power. Furthermore, some groups may link economic problems to foreign
11 In an earlier working paper, we used employment rather than unemployment and the corresponding esti-
mation results were not significant. In addition, several problems arise when computing real GDP per capita but
using an index instead of an actual value did not yield very different results (Hayo, 1999b).

---

B. Hayo / Journal of Comparative Economics 32 (2004) 720–744
729
capital or international trade, as La Ferrara (1996) asserts. Although foreign direct invest-
ment (FDI) can be a catalyst in modernizing the transition economies, a high share of FDI
may indicate that foreigners have an unwelcome position of power. For these reasons, we
consider the degree of openness of the economy to have a negative impact and FDI to have
an ambiguous effect on public attitude to reform. On the other hand, foreign economic
aid may facilitate the creation of a market if it comes from market economies. Hence, we
expect foreign aid to have a positive impact on support for reforms as Table 1 reports. The
IMF plays a prominent role in Eastern Europe not only by providing guidance on macro-
economic policies and structural reforms but also by providing extensive credit programs.
Since its intervention is not always viewed as beneficial as Stiglitz (2003) argues, whether
or not the existence of such an active credit program affects public support for market re-
forms in a country is an open question but we expect these programs to have a positive
effect on balance.
With respect to political institutions, empirical evidence indicates that the introduction
of economic reforms is influenced by initial conditions and by changes in the political sys-
tem (de Melo et al., 2001; Falcetti et al., 2002; EBRD, 1999). In addition, Hayo (2001)
claims that economic developments affect public perceptions of the progress in political
reforms towards democracy. To capture the impact of political institutions, we consider
the effects of the impartiality of executive recruitment practices, the constraints facing the
executive, and the degree of political competition. The latter variable characterizes con-
straints to political participation and the expression of alternative preferences for policy
and leadership. These three variables are taken from the Polity IV database, which contains
information on regime and authority characteristics for all independent states and covers
years from 1800 to 2002. Our working hypothesis is that, if these measures are close to
those in Western-style democratic political systems, the probability of establishing a mar-
ket economy will be higher so that we record positive expected signs for their coefficients
in Table 1. Data are also available in Henisz (2000) on the degree of veto power within
the political system, which affects the ability of the political system to implement market
reforms. We expect this variable to have a negative influence on public support for reforms.
Finally, we consider the effect of a political business cycle by including a dummy variable
if an election occurs in the current year. We expect elections to have a positive impact on
public support for market reforms because politicians will try to improve the performance
of the economy so as to increase their chances of re-election.
4. The estimation results at the aggregate level
Initially, we investigate net support for market reforms at an aggregate level using an
unbalanced panel across countries and years. To avoid problems caused by missing data,
the sample starts in 1991 and excludes Moldova and Yugoslavia. In the empirical analysis,
we take into account all potentially important variables in a single model because the inclu-
sion of variables one by one, or in groups, may lead to spurious results. However, having

---

Document Outline

• Public support for creating a market economy in Eastern Europe
  • Introduction
  • The databases
  • Theoretical determinants of support for the market at the aggregate level
  • The estimation results at the aggregate level
  • Determinants of support for the market at the individual level
  • Summary and conclusion
  • Acknowledgments
  • References

---