The Perils of Polarization: Economic Performance in the Postcommunist World Author(s): Timothy Frye Source: World Politics, Vol. 54, No. 3 (Apr., 2002), pp. 308-337 Published by: Cambridge University Press Stable URL: http://www.jstor.org/stable/25054190 Accessed: 18/12/2009 09:12
Your use of the JSTOR archive indicates your acceptance of JSTOR's Terms and Conditions of Use, available at http://www.jstor.org/page/info/about/policies/terms.jsp. JSTOR's Terms and Conditions of Use provides, in part, that unless you have obtained prior permission, you may not download an entire issue of a journal or multiple copies of articles, and you may use content in the JSTOR archive only for your personal, non-commercial use.
Please contact the publisher regarding any further use of this work. Publisher contact information may be obtained at http://www.jstor.org/action/showPublisher?publisherCode=cup.
Each copy of any part of a JSTOR transmission must contain the same copyright notice that appears on the screen or printed page of such transmission.
JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of content in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms of scholarship. For more information about JSTOR, please contact [email protected].
Cambridge University Press is collaborating with JSTOR to digitize, preserve and extend access to World Politics.
http://www.jstor.org THE PERILS OF POLARIZATION Economic Performance in the Postcommunist World
ByTIMOTHY FRYE*
visions of the politics of economic reform dominate the liter TWO ature. The most widespread approach relies on the logic of the re J-curve,1 which holds that the central dilemma of reform is temporal: to can forms promise generate large economic gains in the future but be To overcome resis achieved only by imposing painfiil reforms today. tance as from groups losing from reform in the short term, such labor, state-sector to concen pensioners, and employees, governments need trate are to power in executives who ideologically committed reform, backed by international financial organizations, and insulated from it popular pressure. Although many have criticized this view recendy, on still guides many academic and policy debates economic reform.2 A second view holds particular relevance for the postcommunist world. Argued most persuasively by Joel Hellman, the "partial reform" to are view is that the main obstacles economic transformation the early winners from distortions in the transition economy who then use their
* I would like to thank Andrew Barnes, David Bearce, Carles Boix, Valerie Bunce, Greg Caldeira, Joel Hellman, Marcus Kurtz, Ed Mansfield, Anthony Mughan, Phil Roeder, Kira Sanbonmatsu, comments. ver Kevin Sweeney, Kuba Zielinski, and three anonymous reviewers for excellent Previous were at at sions of this essay presented the Davis Center for Russian Studies Harvard University, the annual meeting of the American Political Science Association, Cambridge, Mass., February 2001, and the annual meeting of the American Association for the Advancement of Slavic Studies, Washington, D.C., November 2001. 1 Przeworski, Democracy and theMarket: Political and Economic Reforms in Eastern Europe and Latin America (New York Cambridge University Press, 1991). Przeworski cogendy describes the logic of a the J-curve but does not endorse this view. John Williamson, "In Search of Manual for Technopols," in John Williamson, ed., The Political Economy of Policy Reform (Washington D.C.: Institute for Inter national Economics, 1994); Joan Nelson, "Introduction: The Politics of Economic Adjustment in De veloping Countries," in Joan Nelson, ed., Economic Crisis and Policy Choice: The Politics ofAdjustment in the Third World (Princeton: Princeton University Press, 1990); Stephan Haggard, Pathways from the Pe in N.Y.: riphery: The Politics of Growth Newly Industrializing Countries (Ithaca, Cornell University Press, 1990). 2 see Most prominendy, Joseph Stiglitz, "Whither Socialism? Ten Years of the Transition" (Manu script, World Bank, Washington, D.C., 1999); Adam Przeworski, ed., Sustainable Democracy (New York: Cambridge University Press, 1995).
WorldPolitics 54 (April 2002), 308-37 PERILS OF POLARIZATION 309 gains to block further reform.3According to this view, in the short term economic to reform tends produce concentrated benefits for well interest costs positioned groups while dispersing the of transformation rest to throughout the of society. The greatest threat economic reform comes therefore from a narrow group of industrial managers, bankers, and resource on state natural tycoons bent capturing the for their per sonal benefit. This view suggests that robust political competition and diverse are to governing coalitions essential prevent the early winners from state taking control of the and sidetracking further reform. an treats This article advocates alternative approach. It economic per formance as a reflection of the political struggle between ex-communist a war over eco and anticommunist factions engaged in of attrition nomic and political resources.4 Specifically, political polarization be tween ex-communist a and anticommunist factions has had devastating effect on economic these growth. Where competing political factions to over eco have had roughly equal power and continue struggle the nomic rules or of the game?as in Bulgaria Ukraine?economic growth has been slow. Where either ex-communist or anticommunist factions scene?as or have dominated the political in Estonia Uzbekistan?eco nomic performance has been considerably better. two Political polarization has shaped economic outcomes in ways. it First, has heightened uncertainty about future economic conditions because a turnover in to businesses expect potential government bring sharp swings in policy. Facing this possibility, businesses have shied from asset away productive long-term investments, preferring instead stripping, intensive lobbying of state officials and highly profitable but semilegal business deals. More broadly, political polarization has made to it difficult for governments make credible commitments to respect existing and future property rights. Second, political polarization has led to awar of attrition inwhich ex-communist and anticommunist factions have failed to agree on co measures to coun herent address the economic crisis. In the polarized anticommunist some tries, factions have attempted version of neoliberal reforms, and traditional ex-communist factions have at some version of to tempted gradual reform, but neither has been able impose its preferred policy. The resulting war of attrition has led to in coherent policy and slow growth.
3 Hellman, "Winners Take All: The Politics of Partial Reform in Postcommunist Transitions," World Politics 50 1998). 4 (January Alberto Alesina and Allen Drazen, "Why Are Stabilizations Delayed?" American Economic Review 81 (December 1991). 310 WORLD POLITICS
on an in Based analysis of economic growth twenty-five postcom munist countries during the period 1990-98,1 present three findings. First, political polarization has hindered economic growth, controlling a for range of factors. Second, economic growth in polarized countries as in coun has followed the electoral calendar: elections approach such a rates tries, the odds of change in economic policy increase and growth of elections plummet. In nonpolarized countries, by contrast, the effect on growth is absent. Third, political polarization is associated with a is consistent with the greater policy volatility, finding that argument. These findings have implications for studies of the postcommunist eco transformation, the political business cycle, and the politics of more nomic reform generally. on eco This article differs from existing literature the politics of nomic reform in the postcommunist world in three respects. First, it measures Lib economic growth rather than economic liberalization.5 an is not an eralization is important component of transformation but on end in itself. Second, while many works rely cross-sectional analysis or focus on a subset of years within the last decade, this article takes full that data over a time advantage of available data and pools ten-year pe an riod.6 Finally, scholars have developed impressive body of literature outcomes in often lack explaining particular countries, but these works a comparative perspective.7 In I a In Section I, I develop the argument. Section II, conduct quantitative analysis of the effects of political polarization and the on I electoral calendar economic growth. In Section III, examine the impact of political polarization on policy volatility. Section IV concludes.
5 Hellman (fn. 3); M. Steven Fish, "The Determinants of Economic Reform in the Post-Commu and David A. nist World," East European Politics and Society 12 (Winter 1998); Jeffrey S. Kopstein Politics 53 Reilly, "Geographic Diffusion and the Transformation of the Postcommunist World," World (October 2000). 6Anders Aslund, Peter Boone, and Simon Johnson, "How to Stabilize: Lessons from Post-Com on no. 1 munist Countries," Brookings Papers Economic Activity, (1996); Fish (fn. 5); Stanley Fischer, Ratna Sahay, and Carlos Vegh, "Stabilization and Growth inTransition Economies: The Early Expe Blan rience," Journal ofEconomic Perspectives 10 (Spring 1996); Kopstein and Reilly (fn. 5); Olivier chard, The Economics of Post-Communist Transition (Oxford: Oxford University Press, 1997). 7 to a MIT Jeffrey Sachs, Poland's Jump Market Economy (Cambridge: Press, 1995); Anders Aslund, a How Russia Became Market Economy (Washington, D.C.: Brookings Institution, 1995); Laszlo Bruszt and David Stark, Post-Socialist Pathways: Transforming Politics and Property in East Central Europe Tran (Cambridge: Cambridge University Press, 1997); David Bartlett, The Political Economy ofDual sitions: Market Reform and Democratization inHungary (Ann Arbor: University of Michigan Press, a Economic in 1997); Andrei Shleifer and Daniel Treisman, Without Map: Political Tactics and Reform Russia (Cambridge: MIT Press, 2000). PERILS OF POLARIZATION 311
I.The Political Roots of Economic Performance
in more over Few issues comparative politics have received attention wave the last fifteen years than the politics of economic reform. The of neoliberal reforms that swept Latin America, Western Europe, Africa, and eventually the postcommunist world sparked renewed interest in the on impact of politics economic performance. The postcommunist countries, with their diverse institutional arrangements and economic a outcomes, offer rich environment for exploring the topic. Indeed, de spite the homogenizing effects of Soviet-style socialism and economic across globalization, the differences in economic performance post communist countries are to striking.8 According the European Bank for size in Reconstruction and Development, the of the economy Hun gary increased by 2.5 percent annually from 1992 to 1998; inMoldova it fell by 8.5 per year during the same period?a contraction of historic proportions.9 Many countries have defied predictions. Given Ukraine's to its proximity Europe, large market, educated workforce, and balance of agriculture and industry,Deutsche Bank rated it the Soviet republic in saw vast with the greatest economic potential 1990. But Ukraine over economic decline the subsequent decade.10 Meanwhile, Poland, case as a the economic basket of the 1980s, emerged regional economic power in the 1990s. on economic to a The literature the politics of reform points range of factors that may influence economic performance. The J-curve view are to suggests that presidential power and elite partisanship central are com economic performance. By insulating strong executives who to mitted reform and backed by the international community, countries to may withstand pressure change policy exerted by groups bearing the costs short-term of reform. In contrast, the partial reform view empha sizes the importance of democracy and dispersed political power for it economic performance. These features make difficult for the short term winners from distortions in the transitional economy to capture state to state in the and derail reforms. Others have pointed spending,
8 Valerie Bunce, "The Political Economy of Post-Socialism," Slavic Review 58 (Winter 1999). 9 on See Appendix 1 for descriptive statistics polarization and growth. The EBRD's Transition Report rates notes some coun 1999 notes laconically that growth in the region "can lack precision." It also that rates. tries incorporate estimates of the size of the informal economy into their growth The growth data are presented here likely the best available and have been widely used. European Bank for Reconstruction In I and Development, Transition Report (London: EBRD, 1999), 188. the quantitative analysis, introduce a correction to the growth data that attempts to account for the size of the informal economy. See fh. 60. 10 Gertrude Schroeder, "Economic Transformation in the Post-Soviet Republics," in Bartlomiej Kaminski, ed., Economic Transition in Russia and theNew States of Eurasia (Armonk, N.Y.: M. E. Sharpe, 1997). 312 WORLD POLITICS
and ties to international stitutional legacies, geography, organizations, as as of such the European Union and the IMF, important determinants economic performance in transition societies. I adopt a different approach inwhich I analyze how the balance of economic political power in postcommunist countries affects perfor mance. More specifically, I argue that political polarization exacerbates the distributional effects and credible commitment problem of eco to nomic transformation and leads slower growth. Before developing the argument, I briefly discuss my treatment of political polarization. term in Scholars have used the "political polarization" many ways.11 as in the two factions in Conceptually, used this article, major political are ex-communist and anticommunist polarized political systems fac tions. Leaders of the former have typically held responsible positions state or com within the apparatus communist party, have been largely to a state mitted supporting dominant sector, have campaigned with the backing of the ex-communist party, and have emphasized the pos to itive side of communist policy prior 1989. In contrast, the leaders of anticommunist parties have either left or never held positions in the communist party prior to 1989, have been highly critical of the activi to run ties of the communist party prior 1989, have against the largest a ex-communist party, and have favored dominant role for the private sector. it diver This cleavage is particularly important because reflects over structure gent views the of the economy. For example, during the past decade in Russia, the traditional ex-communist party, the Com state own munist Party of the Russian Federation, promised continued ership of land and a leading role for the state in finance and heavy a mas industry; the anticommunists led by President Yeltsin promised on sive restructuring of the economy based private property, liberalized prices, and curtailed state subsidies for industry.12Not only do these
11 a Giovanni Sartori defines polarized party system by the ideological distance between parties, but measures are world. Others measure such precise unavailable in the postcommunist polarization using assume are indices of social cleavages, such as ethnic fractionalization. These indices that cleavages po not movements. ex litically salient, but many cleavages do translate into political The salience of the cases seems to communist/anticommunist cleavage in the postcommunist be quite high. Other types of as not na cleavages, such ethnic divisions, also do translate readily into economic policy. For example, tionalist rhetoric aside, Prime Minister Meciar continued the construction of a market economy in William Slovakia. Sartori, Parties and Party Systems (New York: Cambridge University Press, 1976); Easterly and Ross Levine, "Africa's Growth Tragedy: Policies and Ethnic Divisions," Quarterly Journal and ofEconomics 82 (November 1997); Philip Keefer and Stephen Knack, "Polarization, Politics, Prop a these see erty Rights" (Manuscript, World Bank, Washington, D.C., 2000). For critique of indices, David Laitin and Daniel Posner, "The Implications of Constructivism for Constructing Ethnic Frac APSA-CP tionalization Indices," Newsletter for the Organized Section in Comparative Politics oftheAPSA 12 (Winter2001). 12 Gennady Zyuganov, My Russia: The Political Autobiography ofGennady Zyuganov, ed. Vadim Medish (Armonk, N.Y.: M. E. Sharpe, 1997), pt. 4; Aslund (fn. 7). PERILS OF POLARIZATION 313
competing platforms have little in common, but they also produce dif ferent winners and losers.
Not all ex-communist parties fall into the "traditional" category. John Ishiyama and others have argued that the Hungarian, Polish, are more Lithuanian, and Slovenian ex-communist parties far commit to are ted market-oriented reforms than other ex-communist parties in economic to the region.13 These parties adopted platforms roughly akin and often faced from European social-democratic parties competition more left-wing ex-communist factions. Given their histories, however, these parties lacked credibility, particularly with private business. Here two I treat these reform-oriented ex-communist parties in ways. Ini I treat as tially, them distinct from "traditional" ex-communist parties. in one I then place them group with other ex-communist parties. My treatment of political polarization is similar to that of Stephan use Haggard and Robert Kaufman, who antisystem parties, which they describe as "left and populist parties that have historically mobilized or as around anti-capitalist anti-oligarchic protests," indicators of po true for traditional ex-communist larization. This depiction rings par as in ties.14 Political polarization used here may bring sharp swings policy. In addition, however, it raises the possibility of change in the underlying structural features of the economy. The War of Attrition
to economic in at Political polarization tends undermine performance least two ways. Alberto Alesina and Allen Drazen argue that in politi cally polarized settings it is difficult for politicians to agree on economic welfare but distributional costs on policies that promote social impose The costs of transformation be in a specific groups.15 may paid variety taxes on state or end of ways: by levying capital, by sacking workers, by to sectors. is ing subsidies loss-making The key that each group prefers are to that some other group pay these costs. Where these costs be divided among competing political groups that have very different pol these do not know who will concede icy preferences and where groups the political battle by agreeing to pay the costs of transforming the
13 creates successor and dem Ishiyama three categories of communist parties?standpatter, liberal, areas: ocratic reformist?based on their policy positions in three the economy, the communist past, and in the value of democratic competition. Ishiyama, "Communist Parties Transition: Structures, Leaders, and Processes of Democratization in Eastern Europe," Comparative Politics 27 (January 1995); idem, or Com "The Sickle the Rose: Previous Regime Types and the Evolution of Ex-Communist Parties," Political Studies 30 (June 1997). parative14 Political Democratic Transitions Princeton Uni Haggard and Kaufman, The Economy of (Princeton: versity Press, 1995), 167. 15 Alesina and Drazen (fn. 4). 314 WORLD POLITICS
stalemate war of is to ensue. economy, political (a attrition) likely Dur ing this stalemate both parties seek to shift the costs of the new policy to a their political opponents, situation that inhibits the introduction of coherent economic to policies necessary promote growth.16 Consider a two one earns revenue country with groups: from capital and the other from labor.17Although both will benefit by transforming the economy, each prefers that the other group bear the distributional costs associated with transformation. The capital-oriented faction to the prefers end soft-budget constraint, liberalize prices, and open in to trade. a re dustry foreign The labor-oriented faction prefers gradual form with taxes on elements of protection, significant the private sector, and redistribution to workers to cushion the costs of transformation. over Where political polarization prevents agreement the distribution of these it is to costs, exceedingly difficult introduce coherent policies that promise to improve the economy. In a highly polarized setting each to a war group expects the other concede first and costly of attri tion ensues. when one faction wins the Only political struggle?for example after an we election marginalizes the loser?should expect coherent govern ment a policy, productive response by the private sector, and improved economic a performance.18 The consolidation of political forces around war roughly similar policy ends the of attrition and allows the winners to shift the costs of onto war transformation the losers. Ending the of attrition is essential for the quickly adoption of coherent policies and the resumption of growth. The of awar of attrition has in a set logic special bite postcommunist Given the ting. scope of economic change, the potential gains for win ners and costs for losers are In because especially high. addition, political institutions in transition economies are in often flux, the early winners 16 Alberto Alesina and Howard Rosenthal, Partisan Politics, Divided Government and the Economy York: Morris (New Cambridge University Press, 1995); Fiorina, Divided Government (Boston: Alwyn Bacon, 1996). 17 More two over an asset to or not formally, players bargain and choose fight fight. If neither fights, each receives 0. is so must a a w. Fighting costly each player pay cost, c, and the winner receives prize, Each is uncertain about the costs other side can as w > > the bear. As long 0 c, each party prefers to and receive rather than not Given to fight w-c, fight. these incentives, each chooses fight in hopes of shifting the costs of transformation to their opponents. The result is a war of attrition.
Player 1 Not Fight Fight Not w-c ^ Fight 0,0 -c,
S FiSht w-c, -c -c, -c
18 Alesina and Drazen (fn. 4). PERILS OF POLARIZATION 315
have an incentive to delay political consolidation around one political faction until they have amassed great wealth. Once having done so, can to in they shape political institutions lock their economic gains.19 Credible Commitment
In to war addition exacerbating the of attrition, political polarization in economic increases the probability of sharp changes policy and to a thereby undermines confidence in the government's ability make to over credible commitment property rights time.20 Potential transfers are moments of political power often decisive in polarized political sys tems because all groups expect great swings in economic policy. Losers can versa. a quickly become winners and vice At minimum, these in a changes may bring about rapid changes policy. At maximum, they to may lead structural changes in the economy. In either case, political to ex polarization increases uncertainty and leads firms discount their returns on to state as an pected investment and lobby the insurance pol icy against political change. incur Weak judicial institutions that offer few protections against state in a environment sions by the polarized postcommunist increase in can the size of potential swings policy. Incoming governments change property rights using creeping ^nationalizations, manipula tax to tions of policy, reassignments of property rights political sup to so. porters, and threats incarcerate their rivals.21 And they have done cannot to Because incoming governments credibly commit respect the property rights of their rivals in a polarized setting, the latter have little to even incentive make long-term investments. Moreover, the support ers current of the government recognize that their opportunities for fa treatment state last so as their allies vorable by officials will only long are so to a in power, they, too, have weak incentives invest. Think of in is an in businessperson polarized Russia who considering making vestment an after the introduction of economic reform program. Facing
19 Hellman (fn. 3); Timothy Frye, "Presidents, Parliaments and Democracy: Insights from the Post communist World," in Andrew Reynolds, ed., The Architecture ofDemocracy: Constitutional Design, Management, and Democracy (New York: Oxford University Press, 2002). Conflict20 Alberto Alesina and Guido Tabellini, "A Positive Theory of Fiscal Deficits and Government 57 and Debt," Review ofEconomic Studies (July 1990); Jakub Svensson, "Investment, Property Rights, no. 7 Political Instability: Theory and Evidence," European Economic Review 42, (1998). 21 to a One anecdote highlights this dynamic. A Russian oligarch recounted journalist that in early at Economic told him: February 1996 the Davos meetings of theWorld Forum, George Soros "'Boys, a now was your time is over. You've had few good years but your time is up.' His [Soros's] argument to that the communists were definitely going win. We Russian businessmen, he said, should be careful we to to our in time not lose our lives." that managed get jets and Chrystia Freeland, The Sale of the to York: Random 192. Century: Russia's Wild Ride from Communism Capitalism (New House, 2000), 316 WORLD POLITICS
current to a the probability that the government may give way govern ment with far different preferences thatwill put any such investment at risk, he or she will be reluctant to invest.22 a in In contrast, think of businessperson nonpolarized Slovenia who can not invest with the confidence that economic policy will change dramatically, regardless of changes in the political environment.23 This to confidence is central investment decisions in particular and support more for economic reform generally.24 The effects of political polarization are likely to be especially strong to a prior elections. As elections approach, the probability of large swing in policy increases, making it especially difficult for the govern ment to sector to convince the private invest in productive activities. over Heightened uncertainty future economic policy may lead busi nesses to abroad until the election results are in. Politi park their funds an cians may seek to promote growth in election year, but their ability to a sector in a actually generate progrowth response from the private polarized setting will be limited. In polarized political systems, we to in the therefore expect economic growth be highest years farthest an in from election and lowest election years.25 we to In sum, should find that political polarization leads slower rates in is of economic growth, that growth polarized political systems to to next inversely related the time the election, and that political po to more larization leads volatile policy. cases of the In Particular suggest the plausibility argument. polarized anticom Bulgaria the ex-communist Bulgarian Socialist Party and the two in coun munist United Democratic Front, the leading factions the over try the last decade, have proposed very different economic reforms. to an Neither, however, has gained sufficient political power sustain over time that it could the costs of trans economic policy such impose formation on the other, so each faction has instead used its time in
22 see as an On Russia, Aslund (fh. 7); Vladimir Mau, Russian Economic Reforms Seen by Insider: Suc cess orFailure? (London: Chatham House, 2000); Jerry Hough, The Logic ofEconomic Reform in Russia (Washington, D.C.: Brookings Institution, 2001). 23 see to On Slovenia, Sabrina Petra Ramet, "Slovenia's Road Democracy," Europe-Asia Studies 45, no. 5 (1993); Joze Mencinger, "The Slovene Economy," Nationalities Papers 21 (Spring 1993); Jeffrey D. Sachs and Boris Pleskovic, "Political Independence and Economic Reform in Slovenia," in Olivier Blanchard, Kenneth A. Froot, and Jeffrey D. Sachs, eds., The Transition in Eastern Europe, vol. 1, Studies (Chicago: University of Chicago Press, 1994). Country24 Przeworski (fn. 1). 25 see For a treatment of the political business cycle in Russia, Daniel Treisman and Vladimir Gim or the of British pelson, "Political Business Cycles and Russian Elections, Manipulations Chudar," also and Edward is Journal of Political Science 31 (April 2001). See Timothy Frye Mansfield, "Timing in the Post-Communist World" Ohio Everything: Elections and Trade Liberalization (Manuscript, State University and University of Pennsylvania, November 2001). PERILS OF POLARIZATION 317
TABLE 1 The Cases at a Glance
Polarized Countries* Countries Nonpolarized Rate -7.9% Rate -1.2% Average Growth (1.1) Average Growth (. 71)
Albania Armenia
Belarus Azerbaijan Bulgaria Croatia KyrgysstanCzech Republic Moldova Estonia
Romania Hungary Russia Kazakhstan Ukraine Latvia Lithuania Macedonia Poland Slovakia Slovenia Tajikistan Turkmenistan Uzbekistan
= N 249; T= 4.65; standard deviation in parentheses at half the in aThese countries have had polarized political systems for least years the sample.
resources own office to redistribute economic to its supporters rather to than promote growth. Consequendy, growth has been weak. Indeed, a as few large industrial conglomerates, such MultiGroup, have weak in ened the economy by looting the state.26 By contrast, nonpolarized countries, such as Estonia and Uzbekistan, anticommunist and ex communist factions, respectively, have dominated politics and quickly costs on imposed the economic of transformation their opponents.27 Political uncertainty about economic policy has been much lower in economic these countries, and performance much better. Some quantitative evidence is also consistent with the argument. In Table 1,1 divide the cases into countries that had polarized political sys
26 see On Bulgaria, John D. Bell, "Post-Communist Bulgaria?" in Karen Dawisha and Bruce Parrot, in York: eds., Politics, Power, and the Struggle for Democracy South-Eastern Europe (New Cambridge University Press, 1997); John Bristow, The Bulgarian Economy in Transition (Cheltenham, U.K.: Ed as ward Elgar, 1996); Venelin Ganev, "The Dorian Gray Effect: Winners State Breakers in Postcom munism," Communist andPost-Communist Studies 34 (January 2001). 27 see On Uzbekistan, Gerald M. Easter, "Preference for Presidentialism: Postcommunist Regime Change in Russia and the NIS," World Politics 49 (January 1997); Asad Alam and Arup Bannerji, "Uzbekistan and Kazakhstan: A Tale of Two Transition Paths" (Manuscript, World Bank, Washing on see ton, D.C., 2000); Estonia, Ole Norgaard and Lars Johannsen, The Baltic States after Independ ence (Northampton, Mass.: Edward Elgar, 1999); Terry Cox and Bob Mason, Social and Economic Transformation in East Central Europe (Cheltenham, U.K.: Edward Elgar, 1999). 318 WORLD POLITICS
terns for at least half the period under study and those that did not. I rates two then compare the of economic growth for these groups.28 Av rates over eraging growth and polarization time forfeits information are because many countries polarized for only part of the decade. was Nonetheless, average annual economic growth -7.8 percent in the in countries. If polarized countries and -1.2 percent the nonpolarized we count as countries with reform communist factions polarized, these were figures -5.9 percent for the polarized countries and -2.2 percent for the In either these differences are sta nonpolarized countries. case, at are to tistically significant the .05 level. Thus, there grounds investi more gate the relationship between polarization and growth closely.
II. Quantitative Analysis
assess To the arguments, I estimate the following model, which in as cludes variables for political polarization and factors often cited de terminants of growth. = + + GDPGrowit ?0 ^Polarization? ^^Democracy^ + + $3GovernmentSpendingit fy?penness^) + + $sGDPperCapitai(t_1} ^6Warit+ $7LogInfiation i(t_1} + + + + ^GDPGrowiM) ?{$9Country? Z($10Yearit) eit (1) The is the real rate of dependent variable, GDPGrow.f, year-to-year GDP / in change in in country / in year measured 1998 U.S. dollars. The rate GDP average real of annual economic growth in in the countries economic under study is -3.8 percent. Scholars have measured perfor mance in many ways, but as Luiz Carlos Bresser-Perreira, Jose Maria Maravall, and Adam Przeworski note: "The ultimate economic crite success can a coun rion for evaluating the of reforms only be whether at of try resumed growth stable and moderate levels inflation.,,29 The main independent variable of interest is political polarization. measures seat Polarization. the share of the largest ex-communist (an ticommunist) faction when an anticommunist (ex-communist) holds in in 1994 anticommunist the executive. For example, Bulgaria the Union of Democratic Forces won 29 percent of the seats and formed to minister from ex-communist the largest party opposed the prime the its score was 29. Bulgarian Socialist Party. Thus, polarization
28 treat a as at seats a Here I polarized country having least 20 percent of the held by traditional an ex-communist (anticommunist) party when the executive is held by anticommunist (ex-commu nist) in a given year. 29 Bresser-Perreira, Maravall, and Przeworski, Economic Reforms inNew Democracies: A Social Democratic Approach (New York: Cambridge University Press, 1993). PERILS OF POLARIZATION 319
consensus on Studies from other regions have produced little there is whether regime type influences economic growth, but strong in hand with de sentiment that economic performance has gone hand in world.30 This is consistent with the mocracy the postcommunist par better than other tial reform view that democracies should perform a 1 each that Freedom countries. Democracy\ takes value of for year rates a as a on House country "free," that is, lor 2 its 7-point scale of a political rights; otherwise it takes value of 0.31 High levels of government spending may cushion the distributional en effects of free markets and thereby promote growth.32 Similarly, dogenous growth theorists argue that targeted government spending levels may generate growth.33 However, high of government spending more resources to at the ex may also give rulers pursue personal wealth of economic is the ratio of pense performance. GovernmentSpendingit as a of domestic government spending percentage gross product.34 to as to economic Openness the world economy is often cited critical statistical model includes an an growth.35 To capture this notion, the nual index of external liberalization compiled byWorld Bank econo . from 0 for a mists and country experts. Openness ranges completely to an to autarkic economic system 100 for economic system fully open foreign trade.36 I include a that 1 for each that a dummy variable, War^ equals year a war. country is involved in I control for the wealth of each country by including the gross domestic product per capita, GDPper Capita.,.. 30 Karen L. Remmer, "Democracy and Economic Crisis: The Latin American Experience," World the Third Politics 42 (April 1990); Joan M. Nelson, "The Politics of Economic Transformation: Is Politics and World Experience Relevant in Eastern Europe?" World 45 (April 1993); Aslund, Boone, Johnson (fn. 6); Hellman (fn. 3). 31 a I 1 This threshold, although convention, is nonetheless somewhat arbitrary. estimate model after as 4 lower. so does not alter the redefining the threshold for democracy (a) 3 and lower, (b) and Doing results. Data are available at Freedom House, Annual Survey of Freedom Country Scores, are 1972/73-1998/99 (www.Freedomhouse.org). Similar results obtained using updated POLITY III scores are as a the for democracy, which then treated dummy variable with technique advocated by and Third Wave with the Keith Jaggers and Ted Robert Gurr; Jaggers Gurr, "Tracking Democracy's Polity III Data," Journal of Peace Research 32 (November 1995). 32 as a Bresser-Pereira, Maravall, and Przeworski (fn. 29). That government spending portion of content or GDP is relatively high indicates nothing about the beneficiaries ofthat spending. 33 MIT Phillipe and Peter Howitt, Endogenous Growth Theory (Cambridge: Press, 1998). 34 Aghion EBRD (fn. 9). 35 of Jeffrey D. Sachs and Andrew Warner, "Economic Reform and the Process Global Integration," on no. 1 Robert Determinants Economic Growth Brookings Papers Economic Activity 1, (1995); Barro, of David Cause Growth?" (Cambridge: MIT Press, 1997); Jeffrey A. Frankel and Romer, "Does Trade American Economic Review 89 (June 1999). 36 re Aslund, Boone, and Johnson (fn. 6) argue that economic performance in the former Soviet states to structural publics may differ from that in other in the region due "different underlying factors, as on a such the greater reliance military-industrial production, longer history of communism, greater zone over reliance on trade within the communist bloc, and membership in the ruble when control money creation disintegrated." 320 WORLD POLITICS
to One might expect wealthier economies exhibit stronger perfor a to would mance.37 Alternatively, neoclassical approach growth predict rates that poorer countries would exhibit higher of growth.381 also in cluded a value of the GDPGrow., to ac lagged dependent variable, _1} count for temporal dependence in the data.39 The sample includes twenty-five postcommunist countries during the period 1990-98.40 After pooling these data, I report an ordinary least rate economic squares (OLS) model for the annual of growth.41 Be cause are the sample has fewer than 250 observations and the data in a tests are on a arranged panel, statistical based heteroskedasticity as consistent covariance matrix (HCCM) suggested recently by J. Scott Long and Laurie Ervin.42 To control for unmeasured exogenous economic conditions, I also in as clude dummy variables for each year. This is important, countries in the were to from the interna region exposed similar exogenous shocks to tional economy. To control for unmeasured factors specific individual as the econ countries, such institutional legacies and the composition of omy, I also add a dummy variable for each country. Including fixed-effect concerns dummy variables reduces for omitted variable bias and thereby are not re gives greater confidence in the results. These fixed effects are ported but available from the author. The economic variables in the are a to reverse causation. model lagged by year reduce the likelihood of Results
Results from model 1 in Table 2 are consistent with the argument. a Controlling for range of factors, Polarization. is negatively and signif
37 and Howitt (fn. 33). 38Aghion Barro (fn. 35). 39 on Christopher Achen critiques this strategy for addressing temporal dependence, the grounds that the lagged endogenous variable may substantially deflate the impact of other independent vari are ables. The results of the argument stronger when the lagged dependent variable is dropped. Achen, "Why Lagged Dependent Variables Can Suppress the Explanatory Power of Other Independent Vari ables" (Paper presented at the annual meeting of the Political Methodology Section of the American Political Science Association, Los July 2000). 40 Angeles, The former Soviet countries enter the data set in 1992 after the fall of the Soviet Union. 41 more not correc Because the units outnumber the years in the data by than 2:1,1 do employ the to to tion suggested by Nathaniel Beck and Jonathan Katz; Beck and Katz, "What Do (and Not Do) Review 89 with Time-Series-Cross-Section Data in Comparative Politics," American Political Science 1995). (September42 Long and Ervin show that in samples of fewer than 250 observations Ordinary Least Squares errors. a more (OLS) regression may produce inconsistent standard They propose correction that is ap are more propriate for small samples. Results from this analysis slighdy stronger when opting for the errors on am to an traditional approach of using robust standard with clustering country. I thankful me anonymous reviewer for alerting to this article. Long and Ervin, "Using Heteroscedasticity Con sistent Standard Errors in the Linear Regression Model," American Statistician 54 (August 2000). Table 2 Polarization and Growth1
Model
Polarization -.26** continuous (.12) Polarization continuous -.23** AllCP (.10)
Polarization -7.36** -9.66**
dummy (2.73) (3.07)
Polarization -4.31** dummy all CP (2.17) 1.98 2.69 Democracy 1.98 2.47 2.22 (2.28) (2.25) (2.33) (2.29) (2.32)
Government .03 .03 .07 -.03 -.06 spending (.11) (.11) (.11) (.10) (.10) 1.48 .52 2.61 1.83 3.06 Openness (5.75) (5.70) (5.12) (5.32) (5.06)
GDP -.0047** -.0046** -.0038** -.0041** -.0040** per capita (.0017) (.0017) (.0015) (.0016) (.0015)
War -6.86* -6.85* -7.86** -7.52* -7.81** (4.11) (4.73) (3.69) (4.1) (3.54)
Inflation (log) -1.41** -1.59** -1.28** -1.52** -1.28** (.60) (.64) (S5) (.62)
GDPGrow Lag .22** .22** .17* .23** .15 (.09) (.10) (.10) (.11) (.10)
Time to election -.31 (.69)
to 2.11* Time election* polarization (1.21)
Constant 18.31** 16.96** 20.08** 15.76** 20.01** (7.94) (7.57) (7.79) (7.34) (7.60)
N 183 184 184 184 184
R2 .71 .71 .72 .72 .73
*p<.10;**p<.05;***p<.01 are are not aFixed effects for countries and years included in the analysis but reported. rate of annual in GDP in Dependent variable is the real change U.S. $1998. 322 WORLD POLITICS
to economic A in icantly related growth.43 ten-point rise the polariza tion index yields a 2.6 percent decrease in growth, indicating that the are as as results fairly large, well statistically significant. The wealth of a as is country, proxied by GDPperCapita.(tl)y nega a tively associated with growth, result that supports neoclassical models rate of growth. A10 percent increase in wealth decreases the of growth are to eco by 2 percent. War and high inflation also negatively related nomic is unrelated to economic once growth. Democracy.t growth proper are to controls introduced. Economic openness is positively related but the coefficient is as indicated growth, insignificant, by Openness., 1}. not GovernmentSpendingitis positively associated with growth but is As the on significant. expected, coefficient the lagged endogenous vari is and but not indi able, GDPGrow.{tljy positive significant very large, cating that growth is fairly volatile in this period. In the preceding analysis, I have included only the "traditional" ex-communist in the measure In parties of polarization.44 model 2 I adopt amore expansive definition of polarization by including the four "reformed" ex-communist as well. parties Adopting this broader defi nition on has little impact the substantive significance of polarization. 3 a Model presents the results using dummy variable for polarization on once an movement a cer relying the logic that opposition reaches a to tain threshold, it becomes viable alternative the ruling party, while not. 3 a those below this threshold do In model polarization takes value of 1, when either an anticommunist controls the executive and the largest traditional ex-communist party holds at least 20 percent of seats in or a execu parliament; where traditional ex-communist controls the at tive and the largest anticommunist parties control least 20 percent of seats in are the parliament, and 0 otherwise. Such parties almost always the largest opposition party and, particularly with the help of other par are to to costs trans ties, well placed frustrate attempts impose the of on formation primarily their supporters.45 Results from model 3 rate in indicate that the average annual of growth polarized countries is 7.3 in countries. percent lower than nonpolarized To test the robustness of this indicator, I also included all ex-com more seats munist parties that held than 20 percent of the in parliament or held control measure not of the executive. This does distinguish be
43 to not Dropping this threshold 15 percent does alter the results presented in columns 3 and 4 in Table 2. The measure retains its and with this new 44 polarization sign significance coding. Ishiyama (fn. 13). 45 a an or Haggard and Kaufman (fn. 14) consider party system polarized if antisystem anticapital seats ist party has 15 percent of the in parliament. PERILS OF POLARIZATION 323
tween reformed and traditional ex-communist parties. Model 4 in illustrates are to a Table 2 that the results robust range of plausible cod ings of polarization. As expected, the impact of the dummy variable more treatment using the expansive of polarization is somewhat in is weaker?growth polarized countries 4.3 percent lower than in other countries?but Polarization. retains its statistical significance. In this analysis the inclusion of the four reformed communist parties had a on to mild ameliorative effect growth when compared previous mod not els. Nonetheless, this recoding does dramatically change the results.
The Electoral Calendar, Political Polarization, and Growth
Model 5 examines the impact of the electoral calendar in polarized and an nonpolarized settings by including interaction term, TimeToElec that the number of until the next tions?Polarization.t, multiplies years national election by 1 if the political system is polarized and by 0 oth in a erwise.46 As elections approach polarized setting and uncertainty course one increases about the of future policy, would expect politicians to more to and businesses be likely engage in economic opportunism to than in strategies promote growth. on The significant and negative coefficient the variable Polarization. in model 5 suggests that in election years (when time to elections is 0), economic growth is particularly low in polarized countries. The posi on term tive and significant coefficient the interaction TimeToElection* a a Polarization. indicates that in polarized setting the farther country is an a normal from election, the higher the growth rate.47 Consider In four-year electoral cycle. polarized countries, economic growth de on to an clines average by 3.3 percent three years prior election, by 5.4 two to an percent years prior election, by 7.5 percent in the year pre an in ceding election, and by 9.6 percent in election years. Thus, polar an ized countries economic growth is highest in years farthest from as not election and declines the election approaches. It is surprising that on non the electoral cycle has little effect economic growth in the as polarized countries, indicated by the statistically insignificant coeffi cient on TimeToElection.. it The results of this analysis demonstrate that the electoral calendar a on in also has large effect growth polarized countries; they suggest 46 a National elections here include parliamentary elections in parliamentary system and presiden a tial elections in presidential system. 47 rates an Declining growth in parliamentary systems may hasten calls for election, thus suggesting seems the potential endogeneity of growth and elections. However, it is notable that declining growth no on to have effect elections in nonpolarized countries. 324 WORLD POLITICS
that growth is reduced because of the potential swings in policy that surround elections in such countries. Most important, they provide fur ther confirmation of the argument. Potential Omitted Variables
consensus on for Since there is little the proper specification growth models, I also address concerns for omitted variables that may influence with the I include economic performance.48 Consistent J-curve view, To measure variables for presidential power and elite partisanship. pres use idential power, I Matthew Shugart and John Carey's scale of presi cases dential power as adapted slightly for the postcommunist by Frye, Hellman, and Tucker.49 Given the difficulty of measuring elite parti two first measures the of sanship, I created variables. The percentage seats The second the held by the largest ex-communist party.50 probes in a partisanship of the head of government?a president presidential a minister in a system and prime parliamentary system?by creating the variable which is based on the executive's career ElitePartisanship^ to the ex-communist prior 1989 and his relationship with party.51 I added Consistent with the partial reform argument, Fragmenta which for the cases Noriel Roubini and tion.^ adapts postcommunist measure extent to which actors can Jeffrey D. Sachs's of the partisan block policy change.52This variable has been used by Hellman and the EBRD measures and the number of political factions and institutional players whose assent is needed to make a change in policy.53 48 As Ross Levine and David Renelt note: "There does not exist a consensus theoretical framework on not the variables that to guide empirical work growth, and existing models do completely specify on should be held constant while conducting statistical inference the relationship between growth and "A of the primary variables of interest." See Levine and Renelt, Sensitivity Analysis Cross-Country Nazrul "Growth Growth Regressions," American Economic Review 82 (September 1992), 943; Islam, 110 Empirics: A Panel Data Approach," Quarterly Journal ofEconomics (November 1995). 49 Electoral Shugart and Carey, Presidents and Assemblies: Constitutional Design and Dynamics (New "Data-Base York: Cambridge University Press, 1992); Timothy Frye, Joshua Tucker, and Joel Hellman, on Political Institutions in the Post-Communist World" (Data set, Columbus, Ohio, 2001). Herbert a on eco Kitschelt and Edward Malesky offer conditional theory of the effects of presidential power that with weak economic nomic performance in the postcommunist world. They argue countries is initial prospects choose strong presidencies, and thus the effect of presidential power conditional upon seem to a the one here. prospects. Testing the argument would require model different from proposed Reform" See Kitschelt and Malesky, "Constitutional Design and Post-Communist Economic (Paper pre sented at the annual meeting of theMidwest Political Science Association, Chicago, April 2000). 50 a on see Barbara For discussion of elite policy preferences based their political base, Geddes, in in E. "Douglass C. North and Institutional Change Contemporary Developing Countries," James and Conversations with Nobelists Alt, Margaret Levi, and Elinor Ostrom, eds., Competition Cooperation: about Economics and Political Science (New York: Russell Sage Foundation, 1999). 51 see For more detail on the definition of these and other variables, Appendixes 2 and 3. 52 in Industrial Roubini and Sachs, "Government Spending and Budget Deficits the Countries," Determinants Deficits in Economic Policy 8 (April 1989); and idem, "Political and Economic of Budget the Industrial Democracies," European Economic Review 33 (1989). 53 Hellman (fn. 3); EBRD (fn. 9). PERILSOF POLARIZATION 325 I included variables that account for the influence of the Interna tionalMonetary Fund, the European Union, and geographic proximity to Western Europe.541 also included other variables that scholars have to as attributed long-run economic growth, such the percentage of the GDP in produced agriculture, levels of education, the size of the popu lation, life expectancy, and initial GDP per capita circa 1989.55Adding not these variables does change the results in any substantive fashion: none or of these variables achieved statistical significance changed the measure. are es significance level of the polarization Indeed, the results sentially unchanged from the base model. Robustness Checks
I in I assessed the robustness of the results several ways. dropped each one at a to country from the analysis time and reestimated the model on a case. determine whether the results depended single I then split the into two to see were sample five-year periods whether the results driven by rapid declines early in the decade.56Neither of these changes or affected the sign significance of the variables of interest.57 I also ran the results dropping the fixed effects for countries, for on years, then for both countries and years, and report the results lines so 1, 2, and 3, respectively, of Table 3.58 Doing only strengthened the measure. results for the polarization In each of these specifications, the or only variable that changes sign significance is openness, which be comes in positive and significant each model. Thus, the impact of openness is sensitive to the inclusion of fixed effects.When I dropped I a mem the country-specific fixed effects, added dummy variable for on was bership in the former Soviet Union.59 The coefficient FSU{ 54 The International Monetary Fund 2000 (www. Imf.org/external/np/tre/tad); Randall Stone, Lending Credibility: The International Monetary Fund and the Post-Communist Transition (Princeton: Princeton University Press, 2002). 55 to Martha DeMelo, Cevdet Denizer, and Alan Gelb, From Plan Market: Patterns of Transition, ver World Bank Policy Research Paper (Washington, D.C., 1996); Vladimir Popov, "Shock Therapy sus Gradualism: The End of the Debate (Explaining the Magnitude of The Transformational Reces no. 1 sion)," Comparative Economic Studies 42, (2000). to to 561 had drop the fixed effects for countries and years due the small sample sizes in these esti mations. 57 are not are Measures of corruption and institutional quality directly included in the model but a often correlated with the wealth of country. In addition, polarization may foster corruption and weak institutions by incentives to to prevent in from affecting your firm. 58 heightening lobby changes policy There is great debate about the inclusion of fixed effects?dummy variables for countries or years?in models such as this. Some argue that including fixed effects reduces the potential for omit ted variable bias, while others argue that there is little theoretical basis to include fixed effects and that cure worse on the is usually than the disease. See Symposium Research Design and Methods in Inter national Relations, International Organization 55 (Spring 2001). 59 The is a rather crude indicator of the institutional of a Soviet dummy variable FSUit legacy polity. set are Many aspects that the former Soviet republics apart captured in other variables, for example, 326 WORLD POLITICS
Table 3 Robustness Checks51
1. Polarization, continuous -. 16*** (without country-specific fixed effects) (.06)
2. Polarization, continuous -.28** (without year-specific fixed effects) (.12)
3. Polarization, continuous -.18*** (without either country or year fixed effects) (.05)
4. continuous Polarization, lagged -.23* (excluding reformed communist party (.12) lagged by a year)
5. CP -.15* Polarization, all lagged (including reformed communist party (.09) lagged by one year)
6. Polarization, moving average -.57*** (average polarization for three previous years (.19) excluding reformed communist party)
7. of on -.004 Effect lagged growth polarization
_(.088)_
*p<.10;**p<.05;***p<.01 a Entries 1-6 are OLS estimates for various measures of with polarization heteroskedas consistent errors in is rate ticity standard parentheses. The dependent variable the real of GDP in 1 year-on-year change in 1998 $US. Note that the remaining variables in model of 2 are to conserve are Table included in these analyses, but space their parameter estimates not on presented. Entry 7 reports the effect of lagged growth polarization including the in dependent variables inmodel 1 of Table 2.
negative and lay just beyond the bounds of significance (p=.105). In had no on the of cluding FSU.t impact significance polarization.60 I then examined the possibility that slow growth leads to political versa. polarization rather than vice I lagged both of the continuous measures one re of polarization in equation 1 by year and report the
GDP so on. per capita, miles from Vienna, government spending, and The effects of membership in the former Soviet Union should also be captured by the country-specific fixed effects. One problem with the variable is that it is constant over time. somewhat more measure a FSU.t A refined includes dummy are zone. variable for each year that countries in the ruble This variable is not significant when added to the base model. 60 are a The size and significance of the results essentially unchanged using correction for growth rates that takes into account the size of the informal economy. Marcelo Selowsky and Ricardo Martin calculate rates after the GDP the fraction FSU or growth increasing by xt/3 (for countries) X./10 (for = GDP. other countries), where xit the share of private sector output in They take their estimates for the size of the informal economy from the EBRD's Transition Report (fn. 9). See Selowsky and Martin, "Pol icy Performance and Output Growth in the Transition Economies," American Economic Review Papers and Proceedings 87 (May 1997). PERILS OF POLARIZATION 327
suits on lines 4 and 5 inTable 3. Doing so did not materially affect the as on measures retain results, the coefficients the lagged of polarization their significance. I also measured polarization by taking the average score over /- /- polarization the three previous years (year /, 1, 2). as on Again, this variable is significant, is reported line 6 of Table 3. That the lagged value of polarization is a fairly good predictor of growth rates to is evidence that causation flows from polarization growth. In addition, I lagged the value of growth and placed it on the right 1 measures on hand side of equation and placed the polarization the left hand-side of equation 1 and report the results on line 7 of Table 3. The insignificant coefficient on the lagged value of growth also sug not gests that growth is causing polarization.61 Thus, there is strong empirical evidence that political polarization drives growth. This direction of causation is quite plausible. The polar measure measures ization is taken from elections in years preceding the of economic growth. Moreover, the argument that bad economic per to formance necessarily leads political polarization may be less appro cases. voters a priate for these Susan Stokes argues that if in transition economy believe that the economy must get worse before it gets better, in that then economic decline early the transformation may indicate current on not encour policy is track.62 Thus, economic decline need voters to age support extremist candidates. Beyond transition some as it economies, scholars, such Samuel Huntington, argue that is to vice In rapid growth that leads polarization, rather than versa.63 sum, roots are and are not re the of political polarization complex readily to duced economic performance.
61 reverse an To address the possibility of causation further, I conducted instrumental variable/two stage least-squares regression analysis by estimating the following system of equations: Polarization = + it ?0 ?^Populations $parliamentary System it+ eit. (1) = + + GDPGrow.t ?0 fifiovernmentSpending.^ $2GDPperCapitai{t_v + + + + + e.. (2) ?3Wari(t-i) $AGDPGr i(t-i)^Country.) Wjear.) a In the first stage I use a continuous variable for the size of the population and dummy variable for a as measure In the existence of parliamentary regime instruments for the continuous of polarization. the second stage, I include variables commonly found in models of economic growth. The effect of po on in this at the .10 level. Because litical polarization growth remains statistically significant analysis cases are (1) the number of is fairly small, (2) the determinants of growth and polarization poorly reason to understood, and (3) it is difficult to find proper instrumental variables, there is be skeptical of Vari this technique in this setting. Larry Bartels, "Instrumental Variables and 'Quasi-Instrumental' Peter A Guide to Econometrics ables," American Journal of Political Science 35 (August 1991); Kennedy, (Cambridge: MIT Press, 1993). 62 Susan Stokes, "Public Opinion and Market Reforms: The Limits of Economic Voting," Compar ative Political Studies 29 (October 1996). 63 Haven: Yale Samuel Huntington, Political Order in Changing Societies (New University Press, 1968). 328 WORLD POLITICS III. Political Polarization and Policy Volatility
The preceding analysis demonstrates that political polarization hin cases at In in dered economic growth in the hand. addition, polarized countries economic the of a growth declined with approach elections, finding that is consistent with the view that political polarization af fects growth by increasing the possibility of a change in policy. In this I some consistent section, examine ancillary evidence that should be means of the with the argument. This analysis provides another testing implications of the theory and offers an additional opportunity to iden tify the mechanisms that drive the argument. Thus far, I have emphasized how political polarization intensifies the credible commitment problem by making large swings in policy more to increases economic likely. According this view, political polarization to in uncertainty and reduces incentives engage productive economic as or we behavior, such saving investing. If this argument is correct, more to should expect countries with polarized political systems have extent greater year-on-year volatility in the of economic liberalization. assess To this argument, I estimate the following equation: = + + Policy Volatilityit ?0 ^Polarization^ ?2Democracyit + + $3GovernmentSpendingit ?40/7lV + + + $sGDPperCapitai(t_1} $6Warit ^Loglnflation^^ + + ?sOpennesSi(Hlj ?9Fragmentationit + li?wCountryit) + Z(?uY^) + <>.,(2)
The dependent variable in equation 2, Policy Volatility i?tis the absolute value of the percentage change in theWorld Bank Liberalization Index country i in year /.The index consists of three elements: the extent of lib eralization in foreign trade, the extent of liberalization in internal prices, as Bank and the size of the private sector, calculated annually byWorld a an and economists. It takes value of 0 for ideal-type command economy an 100 for ideal-type free-market economy.64 Because the dependent measures variable the absolute value of the percent change in the index, it includes both increases and decreases in economic liberalization.65
64 The Liberalization Index includes two clear policy elements, liberalization in foreign trade and to as a domestic prices. It also includes the size of the private sector, which is harder classify policy ele ment. an mea Nonetheless, the index does include progress in privatization, which is important policy sure. state over of In addition, reassertions of control property?which would affect the size the private common it is sector?have been following transitions of political power in polarized countries. Thus, to sector important include the size of the private in this index. 65 on See Appendix 1 for data annual average changes in theWorld Bank Liberalization Index. PERILS OF POLARIZATION 329
The main independent variable of interest is political polarization, which ismeasured with a dummy variable including the reformed com munist parties. Polarized countries should experience greater year-on year volatility in policy. Countries with robust democracies may experience less policy volatility because democracies place stronger institutional constraints on policymakers than do autocracies. Autocrats likely have greater op to in portunities introduce changes policy outside of formal legal mech are a anisms. Similarly, countries whose political systems marked by large number of veto points may find itmore difficult to make rapid in I is a changes policy.66 Accordingly, include Fragmentation. which version of the scale of political fragmentation developed by Roubini and Sachs that has been modified for the postcommunist world.67 at war or ex In addition, countries with high levels of inflation may in to perience greater policy volatility response these conditions. I therefore include and in 2. The wealth Warit Loginflation .(t_1} equation a one of country may also influence the volatility of policy. On the GDP hand, countries with high per capita may have stronger institu as to tions, such courts, political parties, and bureaucracies, check the ability of politicians to change policy quickly. On the other hand, wealthier countries may have greater control over resources and have to course greater capacity change the of policy. To capture the effect of a wealth on I include country's policy volatility, GDPperCapita.(t_1}. Countries with high levels of government spending may experience resources to less policy volatility because the government has the smooth the effects of the business cycle. Conversely, governments that more to spend freely may have tools and greater capacity change policy I a measures size of rapidly. thus include variable that the government as a GDP. spending percentage of I also add a dummy variable for the four countries that rely heavily as a source revenue. on upon oil of Countries whose economies depend to in oil and gas may experience fluctuations due changes the interna sector tional market for crude oil. Alternatively, the oil and gas may exert over inordinate influence the state, making changes from their to occur. to preferred policy less likely Finally, openness the world as economy may also shape policy volatility countries with liberal for are more to It is also im eign trade regimes exposed external shocks. to include because it is collinear with the portant Openness.^ highly
66 George Tsebelis, "Decision-Making in Political Systems: Veto Players in Presidentialism, Parlia mentarism, Multicameralism, and British Journal Political Science 25 (January 1995). 67 Multipartvism," of Roubini and Sachs (fn. 52); Frye, Hellman, and Tucker (fn. 49). 330 WORLD POLITICS
Table 4 volatility of economic liberalization4
Model
Polarization .62* (.35)
-.40 Democracy (.40)
Government .02 spending (.01) Oil 1.58 (.99)
GDP .0042** per capita (.0021)
War -.04 (.41)
Inflation (log) .07 (.10)
.99 Openness (2.07) .08 Fagmentation (.17) Constant .53 (1.79) N 167
R2_.53_
*p<.10;**p<.05;***p<.01 of annual inWorld Bank Liberalization "Dependent variable is absolute value change are are not Index. Fixed effects for countries and years included in the analysis but reported.
index of economic liberalization and therefore serves to control for the a existing level of economic liberalization within country. As before, I use OLS regression with the correction for heteroskedasticity consistent errors standard suggested by Long and Ervine.681 also employ year and country-specific fixed effects. The results from Table 4 suggest that countries with polarized polit variation in the extent of ical systems experience greater year-on-year economic liberalization. Polarizationit is both significant and positively related to the volatility of theWorld Bank Liberalization Index. On av erage, the policy volatility index is 62 percent higher in polarized coun
68 Long and Ervin (fn. 42). PERILS OF POLARIZATION 331
are at more tries. These results significant the .10 level with the expan measure as sive of polarization well.69 Wealthier countries experience somewhat higher levels of policy to not volatility, but openness the world economy is significantly related to a policy volatility. Nor is government spending, involvement in war, or on on inflation, dependence oil.70 Similarly, the coefficients Democ and indicate neither has a racyit Fragmentation. that significant impact on policy volatility.71 Most importantly, political polarization is associated with increased in cases are policy volatility the under study. These results consistent with the argument that policy volatility is amechanism by which polit conven ical polarization affects economic growth. They also fit the on tional wisdom the dynamics of economic liberalization in the region. In nonpolarized countries dominated by anticommunists, such as an Estonia, swings in policy have been minimal after initial jump in turnover in early the transformation. Government has been frequent as are to Estonia, but all the major political parties committed creating some a form of market economy with considerable scope for private in been minimal. property, subsequent changes policy have In as nonpolarized countries dominated by the ex-communists, such in min Turkmenistan, annual changes theWorld Bank Index have been imal throughout the 1990s, as liberalization has proceeded fairly slowly. to Similarly, according theWorld Bank Index, nonpolarized Kazakhstan a has made slow but steady progress toward liberal economy. in as By contrast, polarized countries, such Ukraine, Bulgaria, and in starts. In Russia, liberalization has proceeded fits and Ukraine the World Bank Liberalization Index fell sharply in 1993 after initial halt ing steps toward amore liberal economy. In Bulgaria the index fell in 1994 and 1995, after the election of the Bulgarian Socialist Party. In Russia the economic environment became far less liberal in 1998. In deed, the only cases inwhich theWorld Bank Liberalization Index fell a in significantly in given year have occurred polarized political settings.72
69 are errors on Again the results stronger using robust standard with clustering country rather than the HCCM correction suggested by Long and Ervin (fn. 42). 70 The between oil and volatility is sensitive to the coding of 71 relationship policy polarization. and are correlated at .45. either from the individu Democracyit Fragmentation. Dropping analysis to ally does not change the results. These results are, however, sensitive the coding of polarization. 72 cases The Liberalization Index experiences relatively few reversals. Nonetheless, these of back seem to sliding be important. Using the average year-on-year increase in the Liberalization Index? rather than the absolute value year-on-year change?as the dependent variable in the regression occur ten produces different results. For individual elements of the index, reversals times and take place in Russia, Ukraine, Romania, Bulgaria, Belarus, and Azerbaijan in years inwhich the political system is polarized. 332 WORLD POLITICS
environment in The uncertainty of the policy polarized countries has also been reflected in high levels of market volatility. In highly polar ized Russia, for example, the market price for privatization vouchers fell more in by than 25 percent following parliamentary elections Decem ber 1993, inwhich reformist parties fared poorly.73 In addition, the main index for the Russian equities market, the Russian Trading Sys more tem, increased by than one-third in the month following Yeltsin's first-round victory in the 1996 presidential elections.74 In contrast, par liamentary elections in the Czech Republic in 1996 and 1998 caused on only small ripples the main equities market index, the PX-50. Thus, seem to countries with polarized political systems have especially volatile economic environments.
IV. Implications and Conclusion
casts on two to My evidence doubt the dominant approaches the poli tics of economic reform. In contrast to the J-curve argument, I find that and elite partisanship, the strength of the executive, relations with the IMF are not to European Union and the significandy related economic contrast to growth. In the prescriptions motivated by the partial reform are to view, regime type and broad governing coalitions also unrelated to growth. In addition, economic growth is unrelated geography.75 Observers of the postcommunist world have debated at length the roots economic in Poor economic of performance the region. perfor mance is typically attributed either to rent seekers supported politically or to im by large ex-communist parties liberal politicians who have economic posed overly ambitious policies.76 My work suggests that are such analyses useful but incomplete. It is the combination of both relatively strong ex-communist and anticommunist factions, each with to sufficient power prevent the other from implementing its version of the economic rules of the game that has undermined economic growth. or Conversely, countries with dominant anticommunist ex-communist political factions have produced better economic performance than more their polarized counterparts. Indeed, this argument suggests that
73 Timothy Frye, "Russian Privatization and the Limits of Credible Commitment," in David Weimer, ed., The Political Economy of Property Rights: Institutional Change and Credibility in the Reform Planned Economies (New York: Press, 1997). of Centrally74 Cambridge University Timothy Frye, Brokers and Bureaucrats: Building Market Institutions in Russia (Ann Arbor: Uni versity of Michigan Press, 2000); Roderick Kiewiet and Mikhail Myagkov, "The Emergence of the Private Sector in Russia: A Financial Market Perspective," Post-Soviet 14 (January-April 1998). 75 Affairs Fish (fn. 5); and (fh. 5). 76 Kopstein Reilly Aslund (fh. 7); Stiglitz (fn. 2); Mau (fn. 22). PERILS OF POLARIZATION 333
there is a need for more research into how the countries dominated by to ex-communists have managed avoid the prolonged economic de more clines that have marked polarized countries. This article contributes to the ongoing analysis of the relationship between democratization and economic performance in the postcom munist It eco world. calls into question the commonly held view that not nomic and political reform go hand in hand in these cases.77 It is or se democracy autocracy per that drives economic growth, but rather it is the balance of political power between the largest anticommunist and ex-communist economic parties parties that shapes performance. It two re also complements influential arguments about economic form in the region. First, it builds on thework of Hellman by identify ing the political conditions that may underpin the "partial reform In countries equilibrium."78 polarized the short-term winners from incentives to the economic reform have particularly strong lobby state, if only tominimize the possibility of a large swing in policy. Indeed, the evidence presented here finds that countries with polarized political systems have been likely to fall into the "slow growth equilibrium" de or scribed by Hellman, while countries in which either ex-communists anticommunists dominate have largely avoided this path.79 In addition, it identifies the political conditions that lead reform to oriented politicians adopt the strategies and tactics advocated by Shleifer and Treisman.80 Reformist politicians facing strong opposition as from ex-communist parties may favor co-optation, Shleifer and Treisman argue occurred in Russia. In contrast, their counterparts in nonpolarized settings will eschew these tactical compromises because can costs on they impose the of transformation their opponents with little resistance. to on Finally, this article also contributes long-standing debates the treatment political business cycle.81 In the standard of the political to rates business cycle, incumbent politicians seek inflate growth prior to to elections improve their chances of retaining office. Despite the 77 Nelson (fn. 30); Aslund, Boone, and Johnson (fn. 6). 78 Hellman (fn. 3). 79 The polarized countries also experience higher increases in income inequality than do nonpolar ized countries. Data from seventeen countries collected by Branko Milanovich reveal that the average was increase in income inequality between 1988 and 1994 46 percent in polarized countries and 31 in percent the nonpolarized countries. See Milanovich, Income, Inequality, and Poverty during the Tran sition (Washington, D.C.: World Bank, 1998). 80 Shleifer andTreisman (fh. 7). 81 E. B. Tufte, Political Control of the Economy (Princeton: Princeton University Press, 1978); Doug lass Hibbs, Jr., The American Political Economy: Macroeconomics and Electoral Politics (Cambridge: Har on vard University Press, 1987); Alesina and Rosenthal (fn. 16). The literature the political business over cycle is vast and has evolved significantly the past twenty years. The discussion here is abridged. 334 WORLD POLITICS
theoretical in this advances literature, the empirical record of the polit ical business cycle has been mixed.82 Indeed, I find that in the post rates as communist world growth in polarized systems fall elections in are approach, whereas nonpolarized settings these electoral effects on absent. By focusing the incentives of firms, the argument advanced can account outcome: sur here for this the heightened uncertainty that in a to as rounds elections polarized setting weakens incentives invest to elections approach, leading lower growth in polarized settings. Political polarization has slowed economic growth in the postcom in rates munist world. In addition, polarized countries growth have are most an been highest in years that distant from election and decline as a elections approach. Thus, political polarization is key mechanism in that shapes economic behavior the postcommunist world. Whether cases an these insights travel beyond the postcommunist is open ques new tion. Have political struggles between old and regime factions gen erated similar levels of polarization in democratizing countries in Latin or eco America Africa? Has political polarization generated similar nomic are outcomes? These questions for future inquiry.
82 James A. Alt and Alec Chrystal, Political Economics (Berkeley: University of California Press, 1983). APPENDIX 1
Average Years Years Annual of of Average Polarization Polarization Average Abs. Value Annual (Traditional Ex- (AllEx- Polarization Change in GDP Com Parties: Com Parties: Continuous Liberalization Growth,
Country 20% Threshold) 20% Threshold) Measure Index 1989-98
Albania 1991-95 1991-95 20.3 .42 .29 1997-98 1997-98 Armenia 4.6 .59 -4.93 Azerbaijan 1991-92 1991-92 8.1 .72 -7.68 Belarus 1990-96 1990-96 18.0 .47 -1.49 Bulgaria 1990-98 1990-98 30.9 .76 -3.95 Croatia 11.8 .13 -2.16 4.4 .57 -.22 Czech Rep. Estonia 1990-92 1990-92 12 .48 -2.94 Georgia 1990-92 1990-92 7.8 .73 -9.26 Hungary 1994^98 2.4 .17 -.31 11.4a Kazakhstan 0 .59 -4.67 Kyrgyzstan 1990-95 1990-95 35 .94 -3.91 Latvia 1990-93 1990-93 14.6 .58 -3.72 Lithuania 1993-96 3.1 .59 -3.8 11.9a Macedonia 1990-93 1990-93 15.7 -5.11 Moldova 1990-94,1998 1990-94, 20.9 .60 -9.12 1998 Poland 1989 1989,1993 0 .29 1.8 98 31.8a Romania 1990-98 1990-98 20.8 .69 -3.04 Russia 1990-98 1990-98 25,5 .70 -6.26 Slovakia 4.9 .58 .34 Slovenia 1992-98 0 .15 .24 19.3a Tajikistan 1995-98 1995-98 0 .44 -8.17 Turkmenistan 0 .43 -8.17 Ukraine 1990-98 1990-98 19.8 .69 -8.9 Uzbekistan 0 .75 -.77
Mean 10.8 .54 -3.80 (13.0)a
a This includes all ex-communist parties 336 WORLD POLITICS Appendix 2: Variables from Robustness Checks Not Defined in Text
1. Presidential power. Based on Shugart and Carey (fh. 49), as adapted by Frye, Hellman, and Tucker (fn. 49). 2. 0 if / has a of Fragmentation. country noncompetitive system govern ment, allowing elites to make policy with few institutional or partisan con straints (Azerbaijan, Kazakhstan, Turkmenistan, Uzbekistan, and post-1996 a Belarus); 1 if i has single-party parliamentary government or a presidential government with majority support in the assembly (Lithuania in the mid-1990s; Ukraine under President Kravchuk; and Moldova under the Agrarian Demo or cratic Party); 2 if/ has a two-party a divided presidential government (Russia, Kyrgyzstan, and Poland during the coalition government of the Democratic Left Alliance and the Peasant Party); 3 if / has a three-party government; and 4 if/ has a of four or more government composed parties. on 3. Elite partisanship. It ismeasured based the occupational history and partisan base of the executive?the president in a presidential system and the prime minister in a parliamentary system. It takes a value of 1 for leaders who held a top position in the communist party prior to 1989 and retained that po sition after 1989 (for example, Karminov inUzbekistan); 2 for a leaderwho held a lower-level party post in the communist era or held high office in the commu nist but as a party became the executive running head of reformed communist party (for example, Kwasniewski in Poland) or against an unreformed commu nist party (for example, Yeltsin in Russia); 3 is for executives who did not hold a party post in the communist era (for example, Klaus in the Czech Republic). GDP 4. % from agriculture. EBRD (fn. 9). 5. Life expectancy. EBRD (fn. 9). 6. Size of the economy. EBRD (fn. 9). 7. Initial GDP circa 1989. EBRD (fn. 9). 8. Geographic proximity. Miles from Vienna 9. IMF. 1 for each under an IMF reduction or structural year poverty adjust ment agreement, and 0 otherwise. 10. EU. 0 if no formal relationship with the EU; 1 if applied for member ship in the EU; 2 if it has signed an interim agreement; 3 if it has signed an as in a sociation agreement with the EU given year. Appendix 3 Descriptive Statistics
Mean (SD) N Minimum Maximum
Government spending 38.9 188 10 83 (11.4) .47 1 Democracy 206 0 (.50) .56 Openness 250 0 100 (.38) War .13 250 0 1 (.33) Elite partisanship 2.1 250 1 3 (.92) GDP per capita 4267 249 1099 12906 (2246) Communist Party 43.45 246 0 100 (33.7) .84 European Union 250 0 3 (.99) FSU .60 250 0 1 (.49) Oil .14 250 0 1 (35) % GNP 18.8 3.7 67.7 from agriculture 182 (13.7) Presidential power 8.9 180 0 21 (5.46) GDP 35204 249 1152 495263 (74159) 1.86 Fragmentation 205 0 4 (1.17) IMF .38 250 0 1 (.48) GDPGrow -3.82 249 -52.6 14.2 (9.89)