15

Click here to load reader

Testing for Ongoing Convergence in Transition Economies, 1970 to 1998

Embed Size (px)

Citation preview

Page 1: Testing for Ongoing Convergence in Transition Economies, 1970 to 1998

Journal of Comparative Economics29, 677–691 (2001)doi:10.1006/jcec.2001.1736, available online at http://www.idealibrary.com on

Testing for Ongoing Convergence in TransitionEconomies, 1970 to 19981

Saul Estrin

London Business School, London NW1 4SA, United Kingdom

Giovanni Urga

City University Business School, London EC2Y 8HB, United Kingdom

and

Stepana Lazarova

London School of Economics, London WC2A 2AE, United Kingdom

Received June 28, 1999; revised July 30, 2001

Estrin, Saul, Urga, Giovanni, and Lazarova, Stepana—Testing for Ongoing Conver-gence in Transition Economies, 1970 to 1998

In this paper, the authors use a time-varying parameters procedure to test for a commongrowth path in the ex-Communist bloc, both pre- and postreform. They test whether there hasbeen convergence within the bloc or between the bloc as a group and the West. Surprisingly,there is little evidence of convergence within the bloc, which brings into question theeffectiveness of policies to reduce differentials in income per capita under the Communists.There is also little evidence of convergence with respect to the West, either during the periodfrom 1970 to 1990 or if the reform years are included (i.e., 1970 to 1998).J. Comp. Econ.,December 2001,29(4), pp. 677–691. C© 2001 Elsevier Science

Journal of Economic LiteratureClassification Numbers: O40, C22, C23, C15.

1 The paper has benefited from comments from participants at the Society of Economics Dyna-mics 1999 conference held in Alghero (Sardinia, Italy) and from presentations at the New EconomicSchool in Moscow, in Hamburg, and at Nuffield College in Oxford, UK. We thank Alastair McAuley,Michael Funke, Mario Nuti, Victor Polterovich, and in particular Steve Hall for helpful commentsand suggestions on various earlier versions of the paper. We thank three anonymous referees and theEditor, John Bonin, for their valuable comments and suggestions that helped to improve the paper.

677 0147-5967/01 $35.00C© 2001 Elsevier ScienceAll rights reserved.

Page 2: Testing for Ongoing Convergence in Transition Economies, 1970 to 1998

678 ESTRIN, URGA, AND LAZAROVA

1. INTRODUCTION

A large empirical literature on economic growth is based on the seminal works bySolow (1956), Romer (1986), and Lucas (1988). In this paper, we follow time-seriesmethods advocated recently in a number of studies (e.g., Bernard and Durlauf,1995, 1996; Durlauf and Johnson, 1995; Evans and Karras, 1996; Lee, Pesaran, andSmith, 1997) for testing whether the growth paths of different economies converge.In particular, we adopt a method first proposed in the context of convergence byHall, Robertson, and Wickens (1992). The procedure is based on time-varyingcoefficients that allow for flexible dynamic specifications of parameters. We applythis approach to the issue of growth and convergence in the Communist economiesof Central and Eastern Europe during the period from 1970 to 1990 and alsoincluding the postreform period from 1990 to 1998 to examine the impact of theearly years of transition.

The first important issue focuses on the characteristics of the Communist bloc.Communist governments prided themselves on policies that transferred resourcesfrom more developed regions to less developed ones and between countries withinthe Council for Mutual Economic Assistance (CMEA, also known as COMECON)trading group (Ellman, 1988, van Brabant, 1989). With the collapse of commu-nism, the Soviet Union, Czechoslovakia, and Yugoslavia disintegrated, creating18 newly independent countries. Using gross domestic product (GDP) data forthese new countries back to 1970, we address formally the question as to whetherCommunist policies were in fact ensuring convergence of per capita GDP. We findlittle evidence of convergence, either within the region as a whole or for particularlocal groupings, except within the former Yugoslavia.

Our second test concerns convergence with Western economies. The ideologyof the Communist regimes from Berlin to Vladivostok stressed economic growthbased on high investment shares, rapid expansion of the manufacturing sector, andpriority to heavy industry (Gregory and Stuart, 1986; Wiles, 1962; Ellman, 1988).An objective, outlined explicitly in policy statements from the Communist Party(Ellman, 1988), was to catch up and overtake development levels in the West. Froma Marxist perspective, socialism was a higher stage of development than capitalism,while communism, the ultimate goal of the system, would entail the eradication ofscarcity. A necessary condition for the attainment of this fundamental objective wasthat the socialist economies converge with, and ultimately overtake, the per capitaincome levels existing in the most developed countries of the West.2 Our second

Nevertheless, the usual disclaimer applies. We are grateful to Manuela Angelucci for data collectionand to George Tian for assistance provided in preparing an earlier version of the paper. S. Lazarovaand G. Urga acknowledge the financial support from ACE Phare Project no. P97-8663-F and Projectno. P96-6095-R, respectively.

2 The issue of convergence of economic systems was discussed widely in the comparative sys-tem literature (see Bergson, 1971). This paper addresses the original Bergson (1971) question usingcontemporary econometric methods.

Page 3: Testing for Ongoing Convergence in Transition Economies, 1970 to 1998

CONVERGENCE IN TRANSITION ECONOMIES 679

objective is to provide an empirical test of this hypothesis. A null hypothesis canbe derived from the growth slowdown literature (Gomulka, 1994). The socialisteconomies began to grow more slowly beginning during the 1970’s (Weitzman,1979), and most countries stagnated during the 1980’s (Kornai, 1995; Blanchardet al., 1991). Hence, it is not surprising that we find almost no evidence of con-vergence between any socialist countries and the leading Western economies,represented by Germany, the United States, the Organization for Economic Co-operation and Development (OECD) average, and the European Community 153

average during the period from 1970 to 1990 or when the early years of transitionto 1998 are taken into account.

Our analysis is based on a new important data set, comprising per capita outputfor 26 countries of the Communist bloc over the period from 1970 to 1998. Thepaper is structured as follows. Section 2 summarizes the concept of convergenceand introduces the Kalman filter procedure for testing ongoing convergence. InSection 3, we report the empirical results, and Section 4 concludes.

2. THE CONCEPT OF CONVERGENCE AND TIME-VARYINGPARAMETERS PROCEDURE

There is a wide variety of measures of convergence in the literature. The oldestand most widely used approach, pioneered in the classical works of Barro andSala-i-Martin (1991, 1992), exploits the cross-sectional variation in the data. In adifferent strand of literature, the time variation of income levels is examined (e.g.,Bernard and Durlauf, 1995, 1996; Evans and Karras, 1996; Bernard and Jones,1996a,b). To use information contained in both time and cross-section dimensions,panel data techniques are employed in assessing convergence (e.g., Lee et al.,1997; Evans and Karras, 1996; Islam, 1995). The possibility that economies maynot converge to a common steady state even after controlling for initial conditions,but that they may converge to one of several steady states, was analyzed by Quah(1996a,b) and Durlauf and Johnson (1995).

The possibility offered by the time-series approach of testing for pairwise con-vergence makes it our preferred method. The idea behind testing for this type ofconvergence is that if two economies converge, the difference between their in-come levels in logarithms approaches zero. Thus, in the stochastic environment,the difference between the levels in the steady state follows a stationary process.The limit stationarity of the pairwise differences allows also for deterministic orstochastic trends in the levels of incomes. If such trends are present, then theymust be shared by both series. In the case of integrated series, they must be coin-tegrated with a cointegrating vector (1,−1). In the case of trend stationary series,the deterministic trends must be identical in both series.

3 The countries are Belgium, Germany, Greece, Spain, France, Ireland, Italy, Luxembourg, theNetherlands, Austria, Portugal, Finland, Denmark, Sweden, and the United Kingdom.

Page 4: Testing for Ongoing Convergence in Transition Economies, 1970 to 1998

680 ESTRIN, URGA, AND LAZAROVA

Various procedures can be used to test for stationarity of the pairwise differences.However, there is a problem with such testing procedures. While convergence isdetermined by the limiting behavior of the series, stationarity is a property ofthe entire time history of the series (Hall, Robertson, and Wickens, 1993). Ifdata are taken from economies that are far from their steady states, then the nullhypothesis of convergence may be accepted erroneously when stationarity tests areused (Bernard and Durlauf, 1996). Hence, time-series techniques appear to applymore naturally to data characterized by steady state dynamics. The analysis ofstationarity reveals only convergence that can be informally qualified as achieved.

Because we are concerned with assessing convergence in transition countriesthat may be in a disequilibrium state, we propose another test. To allow for thechanging economic environment, we adopt a time-varying parameters approach.The test has been applied to study convergence by Hall, Robertson, and Wickens(1992, 1993). The procedure considers the possibility of parameters evolving overtime and, thus, affords a dynamic interpretation of convergence. The model allowsfor convergence that starts at any point in time, a property of particular importancein the analysis of transition economies.

To carry out the test, we first formulate our problem in state-space form. Theseries under consideration is the difference in logarithms of output per capita levelsin countriesi and j :

dt = logGDPit − logGDPjt .

The observed differencedt consists of an unobserved componentht and a noisecomponentut :

dt = ht + ut ,

where the unobserved component follows a random walk process given by

ht = ht−1+ vt .

The noiseut is assumed to be a serially uncorrelated, normally distributed processwith time-invariant varianceσ 2:

ut ∼ N(0, σ 2).

The disturbancevt is serially uncorrelated, uncorrelated withut , and normallydistributed with variance that may change with time.

We are interested in determining whether such variance declines with time. Wedefine the variance as

varvt = Ät ,

Page 5: Testing for Ongoing Convergence in Transition Economies, 1970 to 1998

CONVERGENCE IN TRANSITION ECONOMIES 681

so that

vt ∼ N(0, Ät ).

We may then test whether the variance follows the process:

Ät = φ2Ät−1 with Ä0 given.

If φ < 1, thenÄt < Ät−1 and, in the long run, the difference between the two seriesdt will become stationary. This is a measure of ongoing convergence. To test thenull hypothesis of convergence (H0 : φ = 1) against the alterantive (H1 : φ < 1),we make use of thet statistics as proposed by Hall, Robertson, and Wickens (1992,1993). On the other hand,φ > 1 implies divergence.

To account for the structural change after the beginning of the transition process,we allow for change in magnitude of variance of the state variableht :

Ät = φ2tÄ0+ δst ,

wherest is a structural step dummy that is equal to 1 for years after the break inthe growth process associated with economic reform.

3. EMPIRICAL RESULTS

In this section, we first provide a detailed description of the way in whichthe data have been constructed. Then we describe briefly the growth process inour 26-country sample from 1970 to 1998, including the recession that occurredthroughout the region during the early 1990’s (European Bank for Reconstructionand Development [EBRD], 1999; World Bank, 1996).4 We investigate empiricallywhether the economies converge using the Kalman filter procedure.

3.1. The Data Set

The data are derived from two sources. For the period after reform, which isdated between 1988 and 1992 depending on the country, each individual transitioncountry adopted a conventional system of national accounting measuring grossdomestic product and its components. These data were collected from the EBRDTransition Reportfor the relevant years. However, the transition economies didnot use the United Nations (UN) System of National Accounts (SNA) during theCommunist era. Rather, they adopted the material product system (MPS), whichdiffers in method, coverage, and classification.

To use consistent series for the long pretransition period in our data sets, wedraw on the work of Paul Marer and his colleagues (Marer et al., 1992), whoused fixed and flexible bridges to construct the most reliable estimates to data

4 For analyses of the causes of the output drop, see Portes (1993) and Gomulka (1994).

Page 6: Testing for Ongoing Convergence in Transition Economies, 1970 to 1998

682 ESTRIN, URGA, AND LAZAROVA

of GDP and its components for the 1980’s in every former Communist country.This work addresses the critical issues of relative price differences, repressedinflation, and the valuing of trade in nonconvertible currencies as well as dealingwith the problems of accounting conventions. Their study provides the basis forconstructing internally consistent data series covering both the pre- and postreformperiods using consistent indicators of national income.

The data series are constructed as follows. For the economies of Central and East-ern Europe (i.e., Albania, Bulgaria, Czech Republic, Slovakia, Hungary, Poland,and Romania) during the years between 1990 and 1998, we use data for grossdomestic product and its components from national accounts and checked theseagainst the EBRDTransition Report. These data have been deflated by the Pro-ducer Price Index for 1987 and converted into dollars at the 1987 exchange rate.For the period from 1980 to 1990, we use the series for gross national product atfixed prices, converted into dollars at the 1987 exchange rate, derived from Mareret al. (1992). For the period from 1970 to 1979, we use net material product (NMP)growth rates at constant prices from national accounts in each country and applythese to the 1980 GDP figures converting to dollars in the previously explained way.The only exceptions to this rule are for Hungary, Poland, and Romania, where databack to 1975 on GDP at constant prices were available from Marer et al. (1992).

For the countries of the former Soviet Union, a similar methodology was applied.We use the SNA for these countries back to 1991 or earlier, converted into dollarsat the 1987 exchange rate. However, for the pre-independence period, we haveinformation only on the proportion of net material product by each republic, whichlater became an independent state in the Commonwealth of Independent States,for every five years from 1970 to 1990. We also have the growth rate of NMP forthe Soviet Union for each year from 1970 to 1990. From these two informationsources, we are able to construct an estimated GDP series back for 1970 on theassumption that NMP and GDP growth rates are equivalent.

3.2. A Description of the Growth Process

We consider the development of GDP per capita for the 26 transition countries inour sample, consisting of the 15 countries of the former Soviet Union, the Czechand Slovak republics, 4 of the former Yugoslav republics, and all of the othercountries of Central Europe. For comparison, we include four Western data seriescovering Germany, the United States, the OECD, and the European Community15, respectively. In contrast to the four Western growth series, the one commonexperience in the transitional economies is a remarkable output drop at the timeof reform. However, for the entire period, there is a considerable variety in growthexperiences among the former Communist countries, and the patterns are differentin Central Europe and the former Soviet Union. Table 1 reports the average growthrates of GDP per capita for each country in various subperiods.

Commencing first with Central Europe and the Balkans, we distinguish countriesthat grew relatively rapidly under communism (i.e., Bulgaria and Romania) from

Page 7: Testing for Ongoing Convergence in Transition Economies, 1970 to 1998

CONVERGENCE IN TRANSITION ECONOMIES 683

TABLE 1Average Growth Rates (changes in GDP) for Transition Economies, 1970 to 1998

al bu cz sk hu po ro ar az

1971–1998 2.4 1.0 0.5 2.0 2.8 2.7 3.0 7.7 −0.41971–1980 7.9 7.0 3.4 5.1 4.9 5.9 9.4 8.0 7.51981–1990 1.4 2.0 0.8 1.5 1.1 0.0 0.4 1.1 −0.61991–1998 −3.8 −8.8 −4.2 −1.6 2.3 1.8 −2.4 16.8 −11.5

be es ge ka ky la li mo ru

1971–1998 3.2 2.1 −1.1 0.0 1.9 0.5 2.1 −1.0 1.51971–1980 7.1 5.3 7.0 4.2 5.3 5.3 5.2 4.7 5.51981–1990 3.8 −0.1 0.1 2.3 4.2 1.5 2.6 1.6 3.31991–1998 −3.2 0.7 −14.5 −9.2 −6.3 −7.6 −3.1 −12.7 −6.8

ta tu uk uz cr ma sm sl

1971–1998 −1.1 −0.9 −0.3 2.8 1.5 0.8 0.0 3.71971–1980 5.5 4.2 4.8 6.2 5.7 6.1 5.1 5.71981–1990 0.2 1.4 2.4 4.5 −0.8 −0.6 −0.1 −1.01991–1998 −12.3 −11.5 −11.6 −4.4 −1.2 −4.7 −7.0 7.4

Note.Country names with abbreviations: Albania (al), Bulgaria (bu), Czech Republic (cz),Slovakia (sk), Hungary (hu), Poland (po), Romania (ro), Armenia (ar), Azerbaijan (az), Belarus(be), Estonia (es), Georgia (ge), Kazakhstan (ka), Kyrgyzstan (ky), Latvia (la), Lithuania (li),Moldavia (mo), Russia (ru), Tajikistan (ta), Turkmenistan (tu), Ukraine (uk), Uzbekistan (uz),Croatia (cr), Macedonia (ma), Serbia-Montenegro (sm), Slovenia (sl).

those with a more lackluster performance (i.e., Albania, Czech Republic, Slovakia,Hungary, Macedonia, and Slovenia as well as Poland, in which growth was highlycyclical even during the Communist era). When it came around 1990, the outputshock associated with economic reform was considerable. In two countries (i.e.,Macedonia and Serbia-Montenegro), the recession eliminated the growth fromthe previous 20 years entirely. Growth had resumed in most Central Europeancountries by 1993, but for the most part it has been quite modest (EBRD, 1999).By 1998, only a handful of countries (e.g., Poland, Slovenia) had achieved theirper capita income levels of 1989. However, for countries in which growth hadresumed, the path appears to be somewhat faster than during the pre-1990 period.

The situation is rather more mixed and variable in the republics of the formerSoviet Union. Most republics, with the exception of Tajikistan, Turkmenistan, andperhaps Kazakhstan, had fairly strong growth performances between 1970 and1990, typically higher than those of Central Europe. Growth in Belarus, Georgia,and Russia was particularly high. However, the output shock everywhere wasmuch larger than in Central Europe (World Bank, 1996), and it more than elim-inated the gains in per capita GDP attained since 1970 in every country except

Page 8: Testing for Ongoing Convergence in Transition Economies, 1970 to 1998

684 ESTRIN, URGA, AND LAZAROVA

Estonia, Russia, and Uzbekistan. However, with the exception of the Baltic states,positive growth was not sustained anywhere in the former Soviet Union before1998.

Therefore, the raw data provide at best patchy evidence of convergence betweenthe countries of the region, either during the Communist period from 1970 to 1990or over the whole sample period from 1970 to 1998. In Central Europe, the lessdeveloped Balkan economies grew somewhat faster under communism, and in theSoviet Union, some of the poorest republics grew very rapidly. One cannot discernmuch evidence that the Communist countries might be converging with Westerncountries during either subperiod.

3.3. Hypotheses about Convergence in the Soviet Bloc

The transition economies were not a uniform bloc even during the Commu-nist era, and the divergence in economic experience has become more markedsince 1990. The most pronounced differences are between the Central Europeaneconomies, which have a shorter history of communism and an ambition to inte-grate with the rest of the Europe, and the countries of the former Soviet Union, inwhich the preconditions were much less favorable to reform in terms of tradition,history, and institutions and for which the commitment to reform has been morehalfhearted (Estrin and Wright, 1999). Even the former Soviet Union is not a uni-form bloc in terms of traditions and institutions. The Baltic states are followingmore of a Central European reform path by seeking European Union membership(Grabbe and Hughes, 1998). For these reasons, we arrange the countries into thefollowing groups.

The first group consists of all the non-Soviet European countries in the WarsawPact plus Yugoslavia. Except for Albania, all of these were members of the CMEA,but none was a republic of the Soviet Union itself. However, they include relativelydeveloped economies such as the Czech Republic and much poorer ones suchas Albania. For the pre-1990 period, we search for two forms of convergencewithin this group. First, we seek evidence of the impact of redistribution policieswithin the multiethnic federal states of Yugoslavia and Czechoslovakia. Second,we investigate whether the allocation and trade policies of COMECON acted toequalize per capita income in the region, for example, by accelerating economicdevelopment in Bulgaria relative to Czechoslovakia. For the post-1990 period, theissue of convergence between this region as a whole and the West is an importantsymbol of the success of economic reform. Moreover, convergence with WesternEurope is of increasing policy significance in the light of proposals to admit someor all of the nations of Central Europe into the European Union (Grabbe andHughes, 1998).

The second group consists of Russia, Ukraine, Kazakhstan, and Uzbekistan.Belarus is also included because the extent of its real economic and politicalindependence from Russia is unclear. All of these countries are very large and,with the exception of Ukraine and Belarus, rich in natural resources. Hence, they

Page 9: Testing for Ongoing Convergence in Transition Economies, 1970 to 1998

CONVERGENCE IN TRANSITION ECONOMIES 685

have a common but very different development experience compared to the smaller,less resource-endowed, and more peripheral remaining republics.

In the third group are the three Baltic republics: Estonia, Latvia, and Lithuania.These countries only joined the Soviet Union during World War II and were amongthe first to declare their independence. They are relatively more developed thanmost of the former Soviet Union and have been closely associated historically intrade and commerce with the Scandinavian economies.

The fourth group consists of the remaining Soviet republics (i.e., Armenia,Azerbaijan, Georgia, Kyrgyzstan, Moldavia, Tajikistan, and Turkmenistan). Forthe most part, they are trans-Caucasian republics or on the periphery of the Sovietbloc. Most are relatively small and less developed and, with the exception ofAzerbaijan and Turkmenistan, have limited natural resource reserves.

For these four groups of countries, we use the variable parameters testing proce-dure to try to answer the following questions. Did the operation of the CMEA andthe Soviet planning system achieve its stated objective of equalizing developmentlevels (i.e., is there convergence within each of the groupings between 1970 and1990)? Did the countries within each group converge with the more developedeconomies of the West, represented by Germany, the United States, the EuropeanCommunity 15, or the OECD average, during the period from 1970 to 1990? Hasthe post-1990 reform process altered the record of convergence, either within eachgroup or between each group and Western economies?

In this paper, we do not report all possible pairwise convergence tests but ratherfocus attention on the most significant comparisons within the groups. However,to address the issue of convergence within the Soviet Union, we examine thepossible convergence between the two other Soviet groupings and Russia for theperiod from 1970 to 1990.

3.4. Results Using the Kalman Filter Procedure

In this section, we report the empirical results using the Kalman filter procedure.From the state-space formulation presented in Section 2, we estimate the parameterφML using the maximum likelihood method. Then we test the null hypothesis ofno convergence, H0 : φML= 1, against the alternative hypothesis of H1 : φML< 15

using thet distribution (tφML) suggested in the recent contributions of Hall et al.(1992, 1993) and St. Aubyn (1995), who also provides the relevant critical values.In the following sections, we reportt statistics only. While statistically significantnegativet values indicate convergence, statistically significant positivet valuesindicate ongoing divergence.

3.4.1. Central and Eastern Europe

We performed this test for the 10 countries in this group. The results for theCommunist era are shown in Table 2. We find little evidence of convergence

5 Note that if a value higher than 1 is obtained, then the dispersion explodes.

Page 10: Testing for Ongoing Convergence in Transition Economies, 1970 to 1998

686 ESTRIN, URGA, AND LAZAROVA

TABLE 2Testing for Convergence within Central and Eastern Europe, 1970 to 1990:t Statistics

bu cz sk hu po ro cr ma sm sl

al 5.282 2.652 3.002 4.997 1.167 2.008 0.823 1.084 2.652 3.500bu −0.133 0.398 0.779 0.233 0.662−0.598 1.834 2.700 3.514cz −1.236 1.361 1.544−1.002 −2.464∗ 1.010 3.031 1.997sk 1.327 −0.586 −0.503 −2.456∗ 1.149 1.533 4.170hu 0.320 −0.192 −1.208 1.816 2.704 3.493po 1.194 −1.654 −0.051 0.505 0.860ro −1.370 0.211 −0.163 −0.293cr 0.667 −4.725∗∗ −3.123∗∗ma −1.831 −0.835sm 0.505

Note.See Table 1 for abbreviations.∗ Significant at the 10% level.∗∗ Significant at the 5% level.

between the Central European countries during the years of communism. Table 2establishes that any centralized cross-country planning of activities and resourcereallocation that occurred during the final years of communism failed to equal-ize incomes per capita across countries in the non-Soviet part of COMECON.6

However, there are a few interesting exceptions to this result, related primarily tointerregional transfers within federal republics.

Table 2 suggests that the strong cross-republican resource transfer policies intro-duced after 1972 in the former Yugoslavia (Estrin, 1983) may have been effectivein equalizing incomes between three of the republics of the federation (i.e., Croatiaand Serbia–Montenegro) as well as between Croatia and Slovenia. These resultsrun counter to the standard interpretation of the disintegration of Yugoslavia, whichargues that the rising tide of nationalism and separatism was underpinned by in-creasing economic differentiation. Rather, our evidence is consistent with the viewthat the costs of subsidies from Slovenia to Croatia and from Croatia to Serbia,and perhaps also to poorer southern states such as Montenegro, may have givenSlovenia and Croatia an economic motive to secede from the federation.

Of the 55 possible pairwise convergencies within Table 2, apart from the 3within the former Yugoslavia, there are only 2 others that are significant. Bothinvolve Croatia and the former Czechoslovakia. Although we do not find significantconvergence between the two parts of the former Czechoslovakia, both constituentrepublics were themselves converging with Croatia. This seems unlikely to be aconsequence of policy intervention and may reflect regional integration betweenrelatively developed and industrialized Communist neighbors.

6 We do not report the table on convergence within Central and Eastern Europe for the full period.There appears to be slightly more significant evidence that these countries are converging in that 14 ofthe pairwise comparisons are statistically significant.

Page 11: Testing for Ongoing Convergence in Transition Economies, 1970 to 1998

CONVERGENCE IN TRANSITION ECONOMIES 687

TABLE 3Testing for Convergence within the Group of Large Soviet

Republics, 1970 to 1990:t Statistics

Belorussia Ukraine Uzbekistan Kazakstan

Russia −0.454 1.344 0.172 0.632Belorussia 0.766 −1.057 −1.479Ukraine −4.056∗∗ −0.596Uzbekistan 5.743

∗∗ Significant at the 5% level.

3.4.2. Former Soviet Republics

The group of the large Soviet republics does not exhibit any sign of convergence,as can be seen from Tables 3 and 4, the only exception being Ukraine andUzbekistan, which show significant convergence at the 5% level. The growthprocess shows a break in 1991, which is captured by the step dummy. Perhapsthese republics were less reliant on natural resources than others in the group and,therefore, faced similar conditions for growth in the Soviet system. Hence, we donot find evidence that policies to equalize income per capita in the Soviet Unionwere effective, either during the Communist era or when the transition period isincluded. This supports the view that the general slowdown in growth in the SovietUnion during the final years of communism was very uneven, as it was affectedby the geographical distribution of natural resource and high-technology sectorsrather than by development levels. This regional diversity of growth experiencehas extended to the transition period.

Although not reported, the results for the Baltic republics group are similar forboth the Communist era and the period from 1970 to 1998. In both samples, thehypothesis of ongoing convergence is rejected. The relative pace of developmentpre- and postreform has been associated not with initial levels of GDP per capitabut rather with regional trade integration and success in attracting foreign direct

TABLE 4Testing for Convergence within the Group of Large Soviet

Republics, 1970 to 1998:t Statistics

Belorussia Ukraine Uzbekistan Kazakstan

Russia 2.473 1.008 −1.340 −0.529Belorussia 7.327 −1.051 −1.450Ukraine −4.157∗∗ −0.459Uzbekistan 5.729

Note.The step dummy assumes a value of 1 in 1991 for thosecountries.∗∗ Significant at the 5% level.

Page 12: Testing for Ongoing Convergence in Transition Economies, 1970 to 1998

688 ESTRIN, URGA, AND LAZAROVA

TABLE 5Testing for Convergence between the Four Groups and Western Economies,

1970 to 1990:t Statistics

EuropeanGermany United states OECD community 15

Russia −1.320 −1.001 −1.112 −1.093Lithuania 4.496 1.113 7.682 5.404Albania 6.493 4.111 5.877 6.592Bulgaria 1.310 −1.373 0.422 1.363Czech Republic −0.615 −3.347∗∗ −0.934 0.589Slovakia −5.878∗∗ −4.671∗∗ −6.431∗∗ −4.465∗∗Hungary −1.558 −4.537∗∗ −2.499∗∗ −1.099Poland −0.665 −1.359 −1.005 −0.259Romania −1.168 −1.181 −1.322 −1.501Croatia −1.488 −1.517 −1.061 2.240Macedonia 4.342 2.138 3.913 4.211Serbia–Montenegro 5.071 1.856 4.589 5.923Slovenia 1.100 −1.363 0.578 1.936

∗∗ Significant at the 5% level.

investment. Moreover, the results of the Kalman filter test for the other countriesfrom the Soviet group are quite simple. If the whole sample is considered, thenmost of the pairs of countries seem to exhibit a tendency to some degree of ongoingdivergence. The evidence of nonconvergence becomes even stronger during thepostreform period. Although these tests are not reported, they are available onrequest from the authors.

3.4.3. Comparison with Western Economic Blocs

We now consider whether selected countries in each group achieved convergencewith the West. When we take the subsample from 1970 to 1990 (Table 5), thereis no evidence of convergence with any of the Western countries for the largestCommunist country, Russia, or for countries that started from low GDP per capita(e.g., Bulgaria, Romania, republics of the Yugoslav federation). These results bringinto question the traditional view of the Communist system as being most effectivein mobilizing resources and, therefore, in generating growth at an early stageof development (Ellman, 1988). The only cases of convergence with the Westinvolve the two richest countries of Central and Eastern Europe: Hungary andCzechoslovakia. However, in the latter, evidence for convergence with the West isactually stronger for the poorer republic, Slovakia.

The negative result is even stronger for the sample period from 1970 to 1998,as reported in Table 6. The growth process shows a break in different years fordifferent countries, as indicated in the note in Table 6. Only in Hungary do we findevidence of convergence with any of the groups of Western countries. Hungary

Page 13: Testing for Ongoing Convergence in Transition Economies, 1970 to 1998

CONVERGENCE IN TRANSITION ECONOMIES 689

TABLE 6Testing for Convergence between the Four Groups and Western Economies,

1970 to 1998:t Statistics

EuropeanGermany United States OECD Community 15

Russia 9.751 7.607 7.651 10.749Lithuania 4.467 0.858 3.350 5.390Albania 7.481 3.966 6.716 6.666Bulgaria 1.034 −1.364 0.419 1.357Czech Republic 12.679 11.901 13.142 14.046Slovakia 10.070 7.837 9.808 10.887Hungary −1.295 −4.563∗∗ −2.337∗ −0.822Poland −0.424 −1.141 −1.132 −0.460Romania 0.577 −1.828 −0.034 −0.722Croatia 1.159 −0.103 0.302 1.262Macedonia 6.137 5.105 5.745 6.904Serbia–Montenegro 15.221 10.895 14.676 16.301Slovenia 0.061 −1.224 0.255 1.388

Note.The step dummy assumes a value of 1 in 1993 for the Czech Republic,Slovakia, Croatia, Macedonia, and Serbia–Montenegro; in 1992 for Poland; in 1990for Slovenia; and in 1991 for the remaining countries.∗ Significant at the 10% level.∗∗ Significant at the 5% level.

was the only COMECON country to abandon central planning prior to the begin-ning of our sample period; it began the reform process in 1968. The absence ofgrowth convergence to any Western country or group of countries for nearly everytransition country up to 1998 highlights the disappointing growth record duringthe early years of transition across the region. However, it is too early to drawconclusive inferences on this matter.

4. CONCLUSIONS

This paper has considered a long-standing question in the comparative eco-nomics literature, namely the convergence of economic systems. Using contem-porary econometric methods, it confirms assertions made during the 1970’s and1980’s that there was no convergence between the capitalist and socialist sys-tems in terms of economic performance. Moreover, using an admittedly shortdata set, it suggests that the sustained difference in GDP per capita between thecountries of Central and Eastern Europe and the more developed economies inthe OECD has not yet changed much in the early process of transition to a mar-ket economy. The lack of convergence to date between the countries of CentralEurope and those of the European Union suggests that there may be serious dif-ficulties in implementing the accession of the 10 applicant countries from theregion.

Page 14: Testing for Ongoing Convergence in Transition Economies, 1970 to 1998

690 ESTRIN, URGA, AND LAZAROVA

The paper highlights the failure of the reallocation mechanism within the so-cialist bloc. Policies to redistribute resources from richer regions to poorer oneswithin the national entities (e.g., Yugoslavia, Soviet Union, Czechoslovakia) werestated clearly. In a rigorous way, our study shows that this reallocation did notlead to convergence in per capita GDP, with the possible exception of Yugoslavia.Given the importance attached to such redistributive policies, this is a significantindictment of the socialist system. Perhaps the failure to achieve this policy ob-jective facilitated the centrifugal political forces that led to the disintegration ofseveral supranational entities in the region.

REFERENCES

Barro, Robert J., and Sala-i-Martin, Xavier, “Convergence across States and Regions.”BrookingsPapers Econ. Activity0, 1:107–158, Jan.–Mar. 1991.

Barro, Robert J., and Sala-i-Martin, Xavier, “Convergence.”J. Polit. Econ.100, 2:223–251, April1992.

Bergson, Abram, “Development under Two Systems: Comparative Productivity Growth since 1950.”World Polit.23,4:579–617, July 1971.

Bernard, Andrew B., and Durlauf, Steven N., “Convergence in International Output.”J. Applied Econ.10,2:97–108, Apr.–June 1995.

Bernard, Andrew B., and Durlauf, Steven N., “Interpreting Tests of the Convergence Hypothesis.”J. Econ.71,1–2:161–173, Mar.–April 1996.

Bernard, Andrew B., and Jones, Charles I., “Technology and Convergence.”Econ. J.106,437:1037–1044, July 1996b.

Bernard, Andrew B., and Jones, Charles I., “Productivity across Industries and Countries: Time SeriesTheory and Evidence.”Rev. Econ. Stat.78,1:135–146, Feb. 1996a.

Blanchard, Olivier, Dornbush, Rudiger, Krugman, Paul, Layard, Richard, and Summers, Lawrence,Reform in Eastern Europe.Cambridge, MA: MIT Press, 1991.

Durlauf, Steven N., and Johnson, Paul A., “Multiple Regimes and Cross-Country Growth Behaviour.”J. Applied Econ.10,4:365–384, Oct.–Dec. 1995.

Ellman, Michael,Socialist Planning.Cambridge, UK: Cambridge Univ. Press, 1988.Estrin, Saul,Self-Management.Cambridge, UK: Cambridge Univ. Press, 1983.Estrin, Saul, and Wright, Mike, “Corporate Governance in the Former Soviet Union: An Overview.”

J. Comp. Econ.27,3:398–421, Sept. 1999.European Bank for Reconstruction and Development,Transition Report.London: EBRD, various

issues, 1994–2000.Evans, Paul, and Karras, Georgios, “Convergence Revisited.”J. Monetary Econ.37,2:249–265, Apr.

1996.Gomulka, Stanislaw, “Economic and Political Constraints during Transition.”Europe–Asia Stud.46,

1:89–106, Jan. 1994.Grabbe, Heather, and Hughes, Kirsty,Enlarging the EU Eastwards.London: Royal Institute of Inter-

national Affairs, Chatham House Papers, 1998.Gregory, Paul R., and Stuart, Robert C.,Soviet Economic Structure and Performance.New York:

Harper, 1986.Hall, Stephen, Robertson, Donald, and Wickens, Michael, “Measuring Convergence of the EC

Economies.”Manchester School Econ. Social Stud.60,0:99–111, supplement, 1992.Hall, Stephen, Robertson, Donald, and Wickens, Michael, “How to Measure Convergence with an

Application to the EC Economies.” Discussion Paper no. 19–93. London: Centre for EconomicForecasting, London Business School, June 1993.

Page 15: Testing for Ongoing Convergence in Transition Economies, 1970 to 1998

CONVERGENCE IN TRANSITION ECONOMIES 691

Islam, Nazrul, “Growth Empirics: A Panel Data Approach.”Quart. J. Econ.110,4:1127–1170, Nov.1995.

Kornai, Janos,Highway and Byways: Studies on Reform and Post-Communist Transition.London:CEU Press, 1995.

Lee, Kevin, Pesaran, Hashem H., and Smith, Ron, “Growth and Convergence in a Multi-CountryEmpirical Stochastic Solow Model.”J. Applied Econ.12,4:357–392, July–Aug. 1997.

Lucas, Robert E., Jr., “On the Mechanisms of Economic Development.”J. Monetary Econ.22,1:3–42,July 1988.

Marer, Paul, Arvay, Janos, O’Connor, John, Schrenk, Martin, and Swanson, Daniel,HistoricallyPlanned Economies: A Guide to the Data.Washington, DC: World Bank, 1992.

Portes, Richard, Ed.,Economic Transformation in Eastern Europe.London: Centre for EconomicPolicy Research, 1993.

Quah, Danny T., “Empirics for Economic Growth and Convergence.”Eur. Econ. Rev.40,6:1353–1375,June 1996a.

Quah, Danny T., “Twin Peaks: Growth and Convergence in Models of Distribution Dynamics.”Econ.J. 106,437:1045–1055, July 1996b.

Romer, Paul M., “Increasing Returns and Long-Run Growth,”J. Polit. Econ.94, 5:1002–1037, Oct.1986.

Solow, Robert M., “A Contribution to the Theory of Economic Growth.”Quart. J. Econ.70,1:65–94,Feb. 1956.

St. Aubyn, Miguel, “Evaluating Tests for Convergence of Economic Series Using Monte-Carlo Methodswith an Application to Real GDP’s per Head.” Ph.D. thesis, London Business School, 1995.

van Brabant, Jozef,Economic Integration in Eastern Europe.New York: Routledge, 1989.Weitzman, Martin L., “Technology Transfer to the USSR: An Econometric Analysis.”J. Comp. Econ.

3, 2:167–177, June 1979.Wiles, Peter,The Political Economy of Communism.Oxford, UK: Basil Blackwell, 1962.World Bank, World Development Report 1996: From Plan to Market. Oxford, UK: Oxford Univ. Press

(for the World Bank), 1996.