Published online by Cambridge University Press: 01 July 2004
International relations scholars increasingly debate when and how
international institutions influence domestic policy. This examination
of ethnic politics in four Baltic and East European countries during
the 1990s shows how European institutions shaped domestic policy, and
why these institutions sometimes failed. Comparing traditional rational
choice mechanisms such as membership conditionality with more
socialization-based efforts, I argue that conditionality motivated most
behavior changes, but that socialization-based efforts often guided
them. Furthermore, using new case studies, statistics, and
counterfactual analysis, I find that domestic opposition posed far
greater obstacles to socialization-based methods than it did to
conditionality: when used alone, socialization-based methods rarely
changed behavior; when they did, the domestic opposition was usually
low and the effect was only moderate. In contrast, incentive-based
methods such as membership conditionality were crucial in changing
policy: As domestic opposition grew, membership conditionality was not
only increasingly necessary to change behavior, but it was also
Although several international organizations (IOs) participated actively in Eastern Europe's ethnic politics during the past decade, research on their role tends to focus on a single institution and the particular strategy it applied. The Organization for Security and Cooperation in Europe (OSCE) has been praised for easing ethnic tensions, but studies rarely focus on its concrete policy effects or they ignore the role of the European Union (EU).12
This article sorts out the institutional effects for the first time by using extensive new data to compare how the OSCE, the CE, and the EU influenced the governments of Latvia, Estonia, Slovakia, and Romania to pass certain ethnic minority legislation during the 1990s. I argue that socialization-based methods such as persuasion and social influence were not very effective when they were used alone, and I show, both through quantitative and qualitative analysis, that more rational-choice-based efforts such as membership conditionality were crucial in changing policy.3
I consider as minorities ethnic Hungarians in Romania and Slovakia, and native Russian speakers in Latvia and Estonia. I refer to the EU even when it was the EC. Similarly I refer to the OSCE, although the OSCE was the CSCE before 1995.
I make three propositions. First, membership conditionality by the EU and occasionally by the CE motivated most policy decisions, but socialization-based efforts often guided them. Case after case illustrates the link between conditionality and decisions to change policies, but the cases also show how the CE and especially the OSCE often shaped the substance of the solutions. Second, domestic opposition posed greater obstacles to socialization-based efforts than it did to membership conditionality: when European institutions used only socialization-based efforts—which they did quite frequently—governments rarely changed their behavior. Socialization-based efforts alone failed in particular when the domestic opposition was strong. Subsequently, in the rare cases when socialization-based efforts did influence ethnic policy without the added use of conditionality, the domestic opposition was usually low and the effect was only moderate. Third, as domestic opposition grew, incentive-based methods such as membership conditionality were not only increasingly necessary to change behavior, but they were also surprisingly effective. Willingness to compromise depended less on the initial position of policymakers than on how much they wanted the reward. Indeed, the vocal opposition to some of the policy compromises and the accompanying political tug-of-war underscored that most of the changes were reluctant responses to the external incentives, rather than products of socialization. While these propositions naturally rest on evidence about ethnic minorities in Europe, the varied involvement of IOs on different ethnic issues, as well as across countries and over time, provides valuable insights that deserve to be included in the larger debate about institutional effects.
I first describe the empirical puzzle and present my analytical framework drawn on international relations theory and considerations of domestic policy. I then discuss possible alternative explanations for the ethnic policy choices of states. I argue that traditional explanations of ethnic politics that locate the source of the policies domestically or with the homelands are inadequate. After describing the data and methods, I examine the statistical findings and case studies in the light of my propositions. To conclude, I assess the relative power of conditionality and socialization-based methods in changing policy and discuss the generalizability and implications of the findings.
One of the most pressing challenges for Baltic and Central European states in the 1990s was tackling ethnic minority issues such as residency rights, citizenship, language issues, and education rights. With the January 1993 dissolution of Czechoslovakia, Hungarians became a much more visible minority group, their proportion rising from 3 percent in Czechoslovakia to nearly 10 percent in the new Slovakia. Slovakia's Prime Minister Vladimir Meciar quickly realized the potential of exploiting nationalist rhetoric to maintain political office, which he managed to do with only a brief interlude until the 1998 elections. Hungarians in Romania made up about 8 percent of a population of twenty-two million. As had indeed also been the case in Slovakia, before the 1990s, Romania underwent a strong communist period that downplayed ethnic identity in favor of party identity. While ethnic Hungarians had therefore been part of the movement to overthrow communist dictator Nicolae Çeausescu, the revolutionary move toward democratization fueled—not decreased—ethnic tension. Like Meciar, Romanian President Ion Iliescu boosted his popularity through exploited nationalist rhetoric and had to cooperate with more extreme parties to maintain political power. Though of a different nature, ethnic tensions were also prevalent as the Baltic states emerged from decades of domination by the former Soviet Union. The ethnic Baltic populations had seen their share of the population in their countries decline drastically. By the time of independence, ethnic Latvians barely comprised half the population of 2.6 million, while Russian-speaking residents made up more than a million, of which 700,000 became stateless when Latvia declared independence. Similarly, about 30 percent of Estonia's 1.4 million population, or as many as 400,000, had no citizenship in Estonia or elsewhere and most spoke no Estonian. This presented a challenge to the desire of ethnic Latvians and Estonians to reassert their national identity. As this examination of the domestic politics reveals, the initial circumstances in the early 1990s did not particularly favor the accommodation of ethnic minorities. Why, then, did these countries make legislative concessions at all, and what explains the variation in the degree of accommodation and the timing of the policy decisions? Why, for example, did Latvia scrap a harsh quota system for naturalization in 1994, yet introduce language legislation in 1998 that made it difficult for non-Latvian speakers to live and work in Latvia? Or why did Slovakia forbid the issuance of bilingual school certificates in 1996, a nearly seventy-year-old practice, when it had passed laws permitting the use of ethnic place and personal names in 1994?
International relations theory offers several mechanisms through which international actors such as IOs may influence state behavior. Two specific mechanisms are particularly relevant given the IOs' activities. The first mechanism is membership conditionality whereby institutions link admission directly to behavior. This is akin to conventional conditionality and incentive use: states respond to incentives and sanctions imposed by international actors, thereby maximizing their payoffs. This mechanism corresponds with a rationalist set of assumptions that define actors as cost-benefit-calculating, utility-maximizing actors. It fits well with Bulgarian Prime Minister Ivan Kostov's comment in April 2000: “With all my respect for the West, I am watching there only the opinion of the structures, which finance Bulgaria. All the others, whatever they say, are of no importance.”4
Reuters Wire Service, 14 April 2000.
The second mechanism—socialization-based methods—includes a broader set of socialization processes such as social influence or persuasion.5
In Kelley 2004, I refer to these as “normative pressure.”
Johnston 2001, 488.
Because membership conditionality and socialization-based efforts are not mutually exclusive, it is compelling to study both mechanisms under the same framework to sort out their effects.11
For common ground between constructivists and rationalists, see Fearon and Wendt 2003.
The IOs' activities on the ground conform well to the descriptions of the two mechanisms. The EU, the OSCE, and the CE first and foremost tried to use communication as a manipulative or persuasive tool. In a combination of ad-hoc visits, letters, reports, declarations, and other forms of interaction, institutions advocated certain legislative goals. The institutions held press conferences and followed up visits with written communication. In Latvia and Estonia, the OSCE also had in-country missions, deployed in 1993, to address citizenship, language, education, employment, and other sensitive issues. These three IOs also had monitoring mechanisms and issued reports and formal statements to guide, advise, and sometimes shame a country. These activities largely conform to what many socialization scholars study under the rubric of persuasion and social influence.13
For the EU and the CE, membership conditionality was a convenient extension of these efforts.14
Because the countries all joined the OSCE when they became independent (except Romania, which was already a member), the OSCE did not use membership conditionality.
The first direct reference was the 1993 Copenhagen Criteria, which stated the need for candidate states to demonstrate “respect for national minorities.”
Were these organizational efforts effective in changing policies relating to ethnic minorities?16
The institutions often have other goals than changing policy, against which their overall “effectiveness” should be evaluated.
Levy 1994, 289–91.
Was the use of membership conditionality effective? Again, international relations theory makes ambiguous predictions. Studies of economic conditionality are a natural body of work to examine for insights about membership conditionality. Scholars are ambivalent as to the effects of economic conditionality, however, not the least because there are multiple ways of measuring success ranging from economic performance indicators, to policy implementation and program completion measures.20
The body of other theory relevant to membership conditionality is slim. Work on political conditionality is sparse.2223
Grabbe 2001, 1026.
As with socialization-based efforts, it is important to understand why membership conditionality might fail. The theory of incentive use rests on the proposition that linking membership benefits with specific policies can entice domestic actors to change their behavior.252627
Interviews by author with Liia Hanni, member of Estonian parliament, 21 September 1999, Talinn, Estonia.
The above discussion highlights the need to understand the role of the domestic opposition. Several groups in the target states have strong stakes in the outcomes. In young democracies with ethnic minorities, parties often organize along ethnic lines and oppose international efforts.282930
Hawkins 1997, 407–8.
Figure 1 shows how policymakers' receptivity to external requirements depends on their initial policy stance on a given issue. Some policymakers will prefer policies that already align with the international requirements, and thus intervention is irrelevant. This was the case, for example, when Latvia passed a law on cultural autonomy in 1991 without any institutional prompting. On other issues, some policymakers do not initially agree, but their policy preferences do not deviate greatly from international standards. This was the case regarding efforts to ease the testing requirements for naturalization in the Baltic states. Here involvement by the United Nations Development Program (UNDP) and the OSCE proved these issues to be more a point of dialogue, training, and technical improvements. In other cases, the policy preference differs drastically from that of the institutions, as in the case of the fervent opposition of some Romanian and Slovak politicians to the conclusion of treaties with Hungary. During negotiations, Slovak Foreign Minister Juraj Schenk said that questions remained, “on which Slovakia cannot yield.”32
CTK National News Wire, 15 March 1995.
To account for domestic influences, studies commonly use macro-level factors such as the nature of party systems, the domestic institutional makeup, and economic data. These explanations, however, lack variation across issues. To increase analytical leverage, I examine the unique circumstances at the decision-making level for each policy issue in a given parliament. A country's overall receptivity to the different external strategies is a function of the receptivity of individual policymakers as well as their ability to form coalitions. Thus, I analyze how following the institutions' recommendations will influence the domestic power balance, what the views of key groups on the issue are, and what positions ethnic minorities themselves hold vis-à-vis staunch opponents.33
Before discussing the findings it is interesting to explore the non–IO-based explanations for a state's choice of ethnic minority policy. Based on theories of domestic distribution of power among social groups and political actors, scholars of ethnic issues often emphasize the role of a nation's demographics, political system of representation, and the relationship between the ethnic groups within a country.343536
Baltic News Service, 9 April 1997.
Other explanations of ethnic minority policies focus on how homelands such as Russia or Hungary may threaten or sanction states to improve the lot of their diasporas.3738
RFE Newsline, 21 July 1998.
Another possibility is that the countries simply made changes in step with their understanding and implementation of democracy. If this was the case, however, then all issues should improve over time, and efforts earlier in the 1990s should be less effective than those in the late 1990s. Contrarily, some policy compromises occurred early—as when Estonia passed amendments in 1993 to simplify naturalization for so-called “loyal citizens,”39
First, when used alone, socialization-based methods rarely change behavior; when they do, the domestic opposition is usually low.
Second, applying membership conditionality is crucial: it produces policy change much more effectively than stand-alone socialization-based efforts and it can do so even in the face of strong domestic opposition. Indeed, willingness to compromise depends less on the initial position of policymakers, than on how much they want the reward. Thus, as domestic opposition increases, it becomes increasingly necessary for institutions to use membership conditionality to change behavior.
Finally, while conditionality motivates most significant behavior changes, socialization-based efforts often play a key role in guiding the substance of the reforms.
Accurately assessing whether an IO influenced domestic policy is challenging: some policy change might instead be related to domestic political incentives; in other instances, governments may make hollow promises to the organizations. The key challenge is to isolate the effect of the institutions and to understand how that effect came about. The activities of the European institutions present a good, though not a perfect, natural experiment as Table 1 shows.
I rely on a data set and case studies that I developed based on my research and interviews with seventy-six persons at the EU in Brussels, the OSCE HCNM in The Hague, and government and other experts in the countries.40
I divided the country studies into sixty-four subcases using different issues and governments over time to define cases.41
King, Keohane, and Verba 1994, 24. Although each observation is not completely independent, the outcomes on most issues can be observed separately and they are often not correlated.
Given the focus on behavior, the most important data were the initial policy on a given issue and the final policy.42
For parsimony and consistency, I use legislation as the primary measurement, but I do consider implementation issues to the extent that they cast further light on an outcome.
The four countries provide a good analytical set, because Russian-speakers constitute the main minority group in both Latvia and Estonia, and ethnic Hungarians are the main minority group in Slovakia and Romania. At the same time, there is variation in how European institutions addressed the countries and issues. Importantly, the data allow the hypotheses to be rejected, because about half the cases were not successful cases of policy adaptation. There is also a fairly even division between a complete lack of institutional involvement and the two institutional mechanisms. Finally, although there is a selection bias in the data, this bias favors stronger conclusions, because—probably because of the increased international attention on high-profile issues—institutions were involved more when the domestic opposition was high. This makes it harder, not easier, to show institutional effects.43
The Appendix includes a table of involvement by domestic opposition.
The dependent variable is the government's legislative behavior on an issue.44
I do not use lagged variables, because the institutions directed their efforts specifically at the current governments, and the governments did consider their efforts and respond. The consequences of this choice should be considered, however. If engagement by external actors occurs in period, but not in t − 1 or t + 1, and a change occurred in period t, this would suggest that the external action was effective. If there was no change in period t, however, but a lagged change in period t + 1, then the applied method would conclude that the efforts of period t were not effective, and that change occurred in the absence of such action in period t + 1; thus it was completely uncorrelated. If socialization occurred in t − 1, however, and conditionality was added in t and change occurred, then this change could be wrongly attributed to conditionality, when it was really a lagged effect of socialization-based efforts. There are several cases, however, where actors used socialization-based efforts for multiple consecutive periods, without behavior changing until conditionality was added. There were also cases where conditionality was never applied. Most importantly, the case studies provided a way of following issues over time and considering possible lagged effects.
A list of the legal documents against which the outcomes are evaluated can be obtained from the author.
Note that by only accepting explicit linkage as “conditionality,” I group cases where the link was potentially implicit with the socialization-based efforts. If anything, the effect of this coding choice will be in the opposite direction of what I have argued. That is, socialization-based methods are likely to prove more, not less, effective.
Note that I derive the quantitative data directly from the case studies. For more on methodology see Kelley 2004.
The IOs clearly produce policy effects. Table 2 contrasts the cases of no institutional involvement with cases of involvement, which then again is broken down by type of involvement. If one compares cases of involvement with cases of no involvement, Table 2 shows that in twenty-three of the forty-four cases where the institutions were involved, or 54 percent of the cases, the policy outcome was “compatible” or “partly compatible.” This is a 34 percent improvement over the four of twenty cases when institutions were not involved.
Table 2 also shows, however, that membership conditionality is much more effective than socialization-based methods alone. A “compatible” outcome occurred in almost three-quarters of cases when the institutions used membership conditionality, whereas it only occurred in eight percent of the cases when institutions used the socialization methods alone (or in two actual cases). This suggests that incentives are a key causal factor in policy change. Indeed, the policy outcomes of socialization-based methods alone hardly differ from those where institutions were not involved at all.
These findings hold when domestic factors are taken into account. In ordered logit analysis, shown in Table 3, regression I, institutional involvement is still significant in explaining the outcomes even when a composite measurement of domestic opposition is included. This is a rather sparse specification, though to some extent the similarities between countries such as Latvia and Estonia does provide for additional control factors, although they are not specified in the regressions. Nevertheless, it is more interesting to see whether the relative explanatory power of the different institutional mechanisms still hold also. Regression II shows that—although both efforts are significant—membership conditionality consistently carries more explanatory power than socialization-based methods. Not only is the finding more robust; the coefficient is also much greater. Further, this coefficient only captures the marginal effect of using membership conditionality. The total effect when the IOs combine socialization-based methods with membership conditionality is actually captured in the addition of the two factors, for a total coefficient of 6.0703. Thus, because the IOs always use membership conditionality in conjunction with socialization-based efforts the coefficient of the effect of this combined use exceeds the coefficient on socialization efforts alone by a factor of approximately 2.5.
It is easier to grasp the magnitudes of each factor from regression II by deriving the predicted probabilities for obtaining a compatible outcome for each type of institutional involvement, while holding the degree of domestic opposition constant.48
See Appendix for calculation of predicted scores.
Regressions III and IV show that including “minorities in government” or the presence of a “dominant leader” in the specification adds little explanatory power. Although the presence of ethnic minorities shows Chi-square significance in Table 7 below, it fails to do so in the regression analysis. Not only is “dominant leadership” statistically insignificant, but it shows a positive coefficient. This may be because the IOs were more likely to use conditionality in these cases. Overall, however, the weak showing by these factors is likely because of data limitations. What is important, however, is that even when controlling for these factors, the other explanatory variables remain robust.
The point of the regression analysis is not that socialization-based efforts always fail. Indeed, the findings on the effect of socialization-based efforts remain robust if the sample excludes cases of conditionality, showing that the effect is not somehow distorted by the fact that there are no observations of conditionality without socialization-based efforts. Indeed, in a few cases socialization-based efforts alone produced results, such as in Romania's education law improvements, or on Latvia's labor law. The significance of using persuasion or social influence in isolation, however, only appears when also considering domestic opposition. In general, it seems that both methods work better the lower the domestic opposition.49
This is a suggestion of some interactive effect, however, since the model is additive and does not include factors that allow domestic opposition to interact with institutional involvement. The sample size is too small to permit testing of interactive effects.
Although the leadership factor was not statistically significant in regression IV, Table 6 shows that authoritarian leadership, as displayed by Slovakian Prime Minister Meciar and Romanian President Iliescu, is correlated with the outcomes. Of seventeen cases with either Iliescu or Meciar in office, twelve had incompatible outcomes, and only three were clearly compatible. Table 7 shows the correlation between ethnic minorities in the government coalition and outcomes. This correlation is statistically significant, but again, the occurrence is very small; only eight of the sixty-four cases had ethnic minorities in the government. Of these eight, two had incompatible outcomes, while six had compatible outcomes. In some cases ethnic minorities clearly shaped the policies. For example, Romania's eventual passage of the education law and the law on the status of the civil service in 1999 was partly due to the negotiation position of the minorities within the government. In 1998, ethnic Hungarians in the Slovak government also helped push for the reinstatement of bilingual school certificates and the passage of a minority language law. Caution is necessary here, however, because minorities were rarely part of the government coalitions, and were never part of the government in Estonia or Latvia. It was also partly because of Western influences that the 1998 Slovak and 1996 Romanian governments included ethnic minority parities at all.50
BBC Summary of World Broadcasts, 4 July 1997.
In sum, while socialization-based efforts alone were primarily effective when the opposition was low, somewhat contrary to the expectations, domestic opposition only had a weak effect on membership conditionality. In several cases, such as citizenship and language issues in the Baltic states or the Slovak and Romanian treaties with Hungary, as well as other cases discussed later, membership conditionality worked in spite of strong domestic opposition. This shows that policymakers' willingness to compromise in exchange for benefits such as EU admission did not depend on their initial position as much as one might assume.
A close study of the policy processes in the four countries bolsters the statistical findings. To illustrate that the case studies included herein are not unique but derive from a larger set of similarly supportive cases, I overview a number of cases briefly before discussing four cases in-depth. On a general note, policymakers often did not take social influence or persuasive efforts very seriously.52
Indeed, some interviewees described how politicians would be very polite during meetings with the OSCE HCNM, only to quickly mock his suggestions after he had departed.
Many other examples similarly illustrate how socialization-based efforts alone failed to improve policy: The OSCE could not persuade the Estonian president to veto the introduction of language requirements for local and national candidates in 1998, and in 1994 the OSCE and the EU (here not yet using the membership carrot) called in vain for Estonia to issue permanent residency permits immediately to those who were residents before July 1990. The pressure by the OSCE to change the citizenship law in both Latvia and Estonia to allow stateless children to acquire citizenship went unheeded for many years, and Latvia ignored most of the OSCE recommendations in the 1998 education law. Thus, socialization-based efforts alone typically produced meager legislative results.
In contrast, numerous cases showed the power of membership incentives. In 1994, the OSCE and CE pushed through changes in the Latvian citizenship law as a precondition to CE admission, although the Latvian population overwhelmingly favored a quota approach to naturalization that would essentially exclude hundreds of thousands of Russian-speakers from ever obtaining citizenship before their death. Likewise, once the EU threw its weight behind the OSCE and CE recommendations regarding stateless children, the legislation passed in both Latvia and Estonia. EU warnings led Slovakia to abandon a harsh penalty code, and later, to adopt a minority language law. Both Latvia and Estonia conceded on their language laws in the late 1990s after the EU linked this to admission. Conditionality drove the treaties with Hungary: Romanian President Iliescu acknowledged that the desire to join the EU and NATO “was indeed the most important aspect,” driving the cooperation on the treaty with Hungary,54
Budapest MTV Television Network in Hungarian, 1700 GMT 22 Sept. 96, FBIS-EEU-96-185.
BBC Summary of World Broadcasts, 21 March 1995.
Baltic Times, 16 May 2002.
I discuss four cases in-depth below.
The issue of a law on national minorities illustrates the insufficiency of socialization-based efforts in spite of a formal promise by the government to pass a suitable law. Even with two years of OSCE efforts to persuade the government to pass a law and numerous efforts by the CE to exert social influence on the government by shaming it through formal criticism, the efforts largely failed.
The ethnic minorities in Romania demanded a national minority law already in 1991, and the OSCE HCNM and the CE specifically raised the need in 1993. The OSCE HCNM, Max van der Stoel, addressed the issue in an August 1993 speech to the Romanian Council on National Minorities and again in a September 1993 letter to the Romanian foreign minister.5758
RFE/RL Research Report 2 (24), 11 June 1993, 38.
Council of Europe 1993, paragraph 10.
BBC Summary of World Broadcasts, 31 March 1994.
In spite of such criticism, there was no movement, except additional promises to the IOs.61
Romanian Ministry of Foreign Affairs 1994. Teodor Melescanu, Minister of State, Minister of Foreign Affairs' answer to the OSCE High Commissioner on National Minorities of 30 May 1994.
The new four-party protocol was between the Social Democracy Party of Romania, Romanian National Unity Party, the PRM, and the Socialist Workers' Party.
The CE approached Latvia's admission with demands for ex-ante policy change on Latvia's most contentious issue at the time: its citizenship law. In 1991, after Latvia regained independence from the USSR, Latvia's supreme council issued a strict resolution on citizenship636465
Author's interview with member of Latvian parliament Boris Tselevich, 19 March 1999, Riga, Latvia.
BBC Summary of World Broadcasts, 24 and 29 September 1993.
Council of Europe 1994a, Paragraph 4, and 1995.
Even so, by early 1994 a strict proposal by the Latvian National Independence Movement (LNIM) was setting the terms of the debate, including a key requirement limiting the future rate of naturalization of noncitizens to 1 percent of the total number of citizens or about 2,000 per year. This would effectively bar tens of thousands of persons from ever receiving citizenship before their death. Despite further CE criticism,69
BBC Summary of World Broadcasts, 2 March 1994.
RFE/RL, 15 June 1994. This was also confirmed in author interviews with members of the delegation, in March 1999, Riga, Latvia.
UPI, 21 June 1994.
The unrelenting CE conditionality ultimately worked. Prime Minister Valdis Birkavs actually asked the president to return the law to parliament, stressing specifically the need to meet the CE requirements: “Latvia cannot expect the Council of Europe to change the aforementioned attitude, which will practically result in delay in Latvia's admission to that organization.”72
BBC Summary of World Broadcasts, 24 and 30 June 1994.
The Farmers Union left the governing coalition on 11 July 1994 because of disagreements over agrarian issues. Latvian Way continued as a minority government.
Only persons with a special “grade I invalid status” were exempted from the language test.
Prime Minister Birkavs most pointedly confirmed that it was the membership conditionality that had been at work when he argued that, “We shall not allow this law to bar our way to Europe, the only place where Latvia can survive.”77
BBC Summary of World Broadcasts, 30 June 1994.
Author interviews with members of parliament, March 1999, Riga, Latvia.
Author's interview with Aleksandrs Kirsteins, member of parliament for the Latvian National Independence Movement and then presently with the People's Party, 19 March 1999, Riga, Latvia.
As in Latvia in the late 1990s, the European institutions were also pivotal in Estonia's decision to allow stateless children to become citizens.80
The term “stateless children” refers to children whose parents were either unknown or former citizens of the old Soviet Union with current citizenship of neither Russia nor Estonia.
The tune began to change, however, when the EU began to support the OSCE recommendations. Published in July 1997, the EU Commission's Agenda 2000 report, which was largely understood as a screening step for EU admission, stated frankly that “The Estonian authorities should consider means to enable stateless children born in Estonia to be naturalized more easily.” In response, the Estonian government began to consider an amendment. Andra Veidemann, then Minister of Ethnic Affairs said: “The minister of foreign affairs, I and the minister of justice discussed this and proposed an amendment [regarding stateless children]…. Right from the beginning the right wing said that they wouldn't vote for the amendment. They said it betrayed the interest of the Estonian nation…. Anyhow, different ministers defended this legislation in parliament. The EU argument was a very strong one.”81
Author's interview with Andra Veidemann, 23 September 1999, Riga, Latvia.
It was not to be so easy, however. The domestic opposition balked at the government's draft, and in early 1998 the amendments failed twice. However, illustrating the dynamics of the EU factor, when the EU negotiations were about to start, the government introduced its draft to the parliament. The bill passed in the first of the three required readings. Keeping up the pressure, the EU noted in the Estonian Accession Partnership in March 1998 the short-term objective of “measures to facilitate the naturalization process and to better integrate noncitizens including stateless children.” The EU-Estonian Joint Parliamentary Committee also urged progress. However, domestic opposition still hindered progress.
As the year passed without results, the EU used the November 1998 first regular report on Estonia to push for change again. The report noted: “it is regrettable that the Parliament has not adopted amendments to the citizenship law to allow stateless children to become citizens.”8384
Author's interview with John Packer, personal assistant to OSCE High Commissioner on National Minorities, November 1998, The Hague, Netherlands.
Author's interview with Mart Nutt, member of parliament since 1992, and author of legislation on citizenship and residency; author's interview with Andra Veidemann, member of government 1992–99, Minister of European Affairs 1996–97, and Minister of Ethnic Affairs 1997–99, 23 September 1999, Talinn, Estonia; author's interview with Kristina Mauer, UNDP in Estonia, 20 September 1999, Talinn, Estonia; author's interview with Liia Hanni, member of parliament, 21 September 1999, Talinn, Estonia.
This case illustrates how effective the additional use of conditionality can be even when there is considerable domestic opposition. It also contrasts the strong response to conditionality with the inefficacy of using socialization-based methods alone for six years. The dynamic timing of the Estonian government's decisions and EU actions highlights the power of conditionality.
Slovakia's decision not to amend its penal code illustrates that even leaders such as Slovak Prime Minister Meciar may succumb to EU pressure when pressure is linked to membership. When Meciar, after a brief interlude, returned to power in 1995, he seized control of the key areas of political power in what came to be known as the “night of the long knives.”86
Steve Kettle, “Slovakia's One-Man Band,” Transition, 23 August 1996.
CTK National News Wire, 27 April 1995.
CTK National News Wire, 24 January 1996.
The OSCE HCNM and the EU criticized the draft bill in letters and meetings and called for Slovakia to “respect ethnic minority rights and freedom of speech.”8990
Slovak National Council 1996, Article 92. In particular, the amendment said that, “Anyone who with the intention of harming the constitutional order, the integrity of the territory or the defense of the Republic or undermining its independence organizes public meetings will be liable to a term of imprisonment of between 6 months and 3 years or a fine.” Article 92b sub. 1.
In a presidency declaration a week later, the EU welcomed the ratification of the Treaty with Hungary, but in the same breath criticized the amendments and urged Slovakia to find a solution: “compatible with the conclusions of the Copenhagen Summit of the European Union and with the EU membership for which Slovakia has applied [emphasis added].”9192
Letter from the OSCE High Commissioner on National Minorities Max van der Stoel to Slovak Minister for Foreign Affairs Juraj Schenk, 13 August 1996.
CTK National News Wire, 17 December 1996.3
Referring to Article 87 of the Constitution, Kovac claimed parliament had not fulfilled its duty of discussing the law after the president returned it in April. Consequently, parliament could not discuss a further amendment before having discussed the previous version of the law.
BBC Summary of World Broadcasts, 19 February 1997.
Author's interview with Bela Bugar, 7 February 2000, Bratislava, Slovakia. Author's interview with Grigorij Meseznikov, President for the Institute for Public Affairs, 9 February 2000, Bratislava, Slovakia.
Author's interview with Peter Hunzik, founder of the Hungarian Civil Liberal Party, 9 February 2000, Bratislava, Slovakia; author's interview with Frantisek Sebej, Chairman of the Committee for European Integration, 8 February 2000, Bratislava, Slovakia.
Author's interview with Peter Weiss, Chairman of the Foreign Affairs Committee, 10 February 2000.
Thus, in this case, the European organizations helped overcome repeated attempts by nationalists to pass legislation opposed by the organizations. Most notably, the case occurred as the EU release of the Agenda 2000 opinion in the summer of 1997 was approaching, illustrating the power of the EU membership incentive.99
Other criminal defamation/insults provisions in the Penal Code were abolished in 2002 and 2003.
To this point I have shown that socialization-based efforts alone are not very effective in influencing a state's domestic policy, and that as domestic opposition to policy change grows, it becomes crucial to use membership conditionality. However, the question remains as to the relative impact of socialization efforts vis-à-vis conditionality when both efforts are used simultaneously. I have shown that conditionality is not always sufficient, given the cases where conditionality failed—particularly during Meciar's Slovakia, such as Meciar's refusal to allow issuance of bilingual language certificates, his insistence on changing the election laws, and his failure to pass a minority language law. The question is, however, when the institutions used membership conditionality effectively, would they have failed if socialization-based efforts had been absent? This is a complex question, because institutions never applied conditionality without also relying on softer efforts. There are several ways to address the question of relative impact, however.
One way is to examine whether the policy changes endure, because this could be read as evidence in favor of internalization, a result more likely brought about through socialization. In this study, there are actually no examples of reversal in the policies. However, there are several competing explanations. The durability of these policy changes may, for example, arise because of a certain path-dependency of policymaking in the countries that makes it difficult to actually reverse policies in the short term. Also, although the CE membership incentives no longer exist, the incentive of EU membership has remained powerful long after some of the legislative changes occurred. Although all states, except for Romania, were confident that they would join the EU in May 2004 (even Romania in spring 2004 appeared quite likely to join the EU in 2007), some incentive remained not to “rock the boat” as a state entered the EU harbor. Further, continued routine monitoring and attention from the international community decreases the chance of backsliding by imposing reputational costs: it could be damaging to a country's credibility to have the issues resurface. Thus it is possible that reputational concerns act as reinforcement. The observed durability is therefore not strong evidence of internalization. That strong domestic opposition often remains when conditionality is effective also suggests that broad internalization is not driving the change.
Given the data constraints, there are three other ways to sort out the effects. First, variation in the level of socialization-based efforts may make it possible to say whether socialization-based efforts condition the effect of membership conditionality. In all but one of the nineteen conditionality cases, however, the OSCE and the CE were active participants and it is difficult to compare their levels of engagement. This brings up another important point, which is that the relationship between the OSCE, the CE, and the EU often became intertwined because the EU relied on the OSCE and the CE for evaluation and information. Thus the very character of the socialization-based efforts changed once behavior was linked to admission in an organization: even the normative actors indirectly gained instrumental leverage through their relationship with the admitting organization.
Counterfactual analysis is a more effective way to assess how pivotal socialization efforts were. Does systematic examination of each case suggest that the socialization efforts created links in the process of change that would not otherwise have been made? In the three cases of successful institutional influence discussed above, this does not seem to be the case. Although the socialization efforts framed the problems, facilitated dialogue, and helped formulate solutions, the conditionality appeared to be such a strong motivating factor that it is plausible that the results would have come about eventually even if these supportive actions had been absent.
In the case of stateless children in Estonia, for example, the OSCE clearly was highly active—yet the timing of compromises corresponded highly with the timing of EU actions regarding membership: the compromises were reached prior to EU meetings, the drafting of EU reports, and interaction with EU officials. This suggests that joining the EU was by far the most important motivation for the legislative reforms. It is quite possible, however, that the EU would not have framed the issues the way it did without the OSCE involvement, or—more generally—that the softer actors influence the content of norms that the more instrumental actors apply. However, given the strong EU membership ambitions and the open EU support for the OSCE, the high level of OSCE involvement itself does not necessarily mean that the socialization-based OSCE efforts had any independent effect. The efforts were no doubt helpful. For example, as the Estonia citizenship issue was drawing to a climax in 1998, OSCE staff visited Talinn urging specific swing policymakers to change their position. The reasoning of the OSCE staff was both instrumental (by arguing that the benefits of conversion outweighed the cost) and moral (by arguing that the reforms were morally right).100
Author's interview with John Packer, Personal Assistant to the OSCE High Commissioner on National Minorities, October 1998, The Hague, Netherlands.
In the case of Slovakia's rejection of changes to its penal code, the OSCE was again highly active and probably flagged the attention of the EU on the matter. While the OSCE and EU were mutually supportive, the fact that Slovakia rejected amending its penal code as the EU was preparing the Agenda 2000 report and after the EU officially linked the law to admission suggests that the relative causal impact of conditionality was larger than that of persuasion or social influence. Indeed, the Slovakia parliament rejected the law because a few members of the governing coalition finally defected from the party line on the third attempt to pass the law. However, several factors make it unlikely that these politicians had changed their beliefs drastically since the two previous votes: only a few months had passed since those votes and there were little to no socialization efforts during that time. Furthermore, the law had been slightly softened since the two previous attempts; this should have made the law more acceptable to any swing voters, not less. Thus, it makes most sense to interpret the rejection of the law as a response to the EU pressure in the light of the upcoming evaluation for admission.
Counterfactual analysis is more challenging in Latvia's 1994 change to the citizenship law, because the membership incentive came from the CE, which is also a heavily norm-based institution. Thus one cannot simply separate the actors and ask: Would the CE have succeeded without the OSCE? Both the CE and the OSCE clearly helped define a solution to the problem of naturalization at the time. The question, however, is whether the willingness to compromise resulted solely from the CE's conditionality or whether it indeed depended also on the CE norm-based efforts. Understanding this requires a more in-depth examination of the CE efforts themselves. While the CE had made seven visits between 1991–93, it made only two in 1994 before the passage of the citizenship law: first in January, but then not again until August 1994, just a few days before the final draft was approved. The CE did not issue formal documents on Latvia during spring 1994. The absence of visits in the spring of 1994 suggests that the main activity of the CE was not intensive persuasion efforts. One must also recall that the Latvian parliament actually passed an unsatisfactory version of the law in June 1994 and then reversed itself only two months later. However, during these two months the relationship with the CE seemed very much to be one of negotiation—not about what was “right”—but about what was sufficient to gain CE approval: As the opinion on membership stated later, adoption of the law had been a “major pre-condition for accession.”101
The last way to view the question of the relative causal power of conditionality vis-à-vis socialization-based efforts is to examine what rhetoric accompanied the policy changes: if the socialization-based efforts were important, one might expect some rhetoric about the moral imperative of the changes, rather than just comments linking the changes to EU membership. Indeed, politicians may prefer to present the policy positions as their own to avoid being seen as puppets. Of course, this effect could be distorted by the fact that policymakers might prefer to blame the IOs for unpopular moves, even if they actually agreed with them. Given these contradictory expectations, it might be fair to predict a mix of public rhetoric on the causes for the changes. However, the comments by policymakers rarely ever reflected any moral support for the policies at all. Indeed, when Slovak Prime Minister Mikulas Dzurinda drew fire for making concessions to Hungarians on a minority language law, he almost excused himself from his action: “I am not responsible for the fact that someone has suggested that the law on the usage of ethnic minority languages in public administration be drafted. I repeat, it was not me who made sure this law was embodied in the constitution and it was not me who concluded the basic treaty with Hungary in France.”102
BBC Summary of World Broadcasts, 9 March 1999.
Baltic Times, 23 April 1998, 5.
These inquiries do not prove that conditionality would have worked in isolation, but they do suggest that incentive-based methods such as membership conditionality are the primary factors in changing behavior. Before concluding that socialization-based efforts may as well be discarded, however, it is useful to return to the findings of economic conditionality. Economic conditionality has faced the most difficult of challenges and only met with partial success. However, recent efforts to improve economic conditionality stress the necessity of engaging the domestic actors through “ownership” of the reforms, the notion that domestic actors are involved in the designing of policy solutions.104
The ethnic politics of the Baltic and Central European countries offer a clear example of how and when IOs can influence domestic policy. The European organizations effectively used membership conditionality and socialization-based efforts to influence domestic legislation—even when they faced quite strong opposition to their policy recommendations. Further, by examining the IOs' efforts side by side and considering how domestic opposition conditioned the efforts, the study addresses a long-standing debate among international relations scholars about the relative explanatory power of socialization-based efforts such as persuasion and social influences vis-à-vis more rational-choice-based efforts such as membership conditionality. Both the statistical and the qualitative analysis showed that membership conditionality was much less sensitive to domestic opposition than were socialization-based efforts. That is, the spectrum of policymakers who responded to conditionality pressure was larger than expected, while the spectrum of policymakers who responded to socialization alone was narrower. In most cases, IOs could change policy with membership conditionality, whereas socialization-based efforts only really worked when the domestic opposition was quite low or if ethnic minorities themselves had bargaining power in the government. Examining issues over time, socialization-based efforts were typically fruitless for several years, only for policies to change dramatically once the linkage to membership was made. Case studies, counterfactual analysis, and examination of public rhetoric further support that conditionality was the factor motivating change. It is worth noting, however, that the IOs always applied membership conditionality to an existing mix of diplomatic efforts. These efforts often helped frame the problems, facilitate dialogue, and formulate solutions. Thus, conditionality motivated the actors while socialization-based efforts guided them.
Two pertinent questions remain: Will the legal changes change deeper social behavior, and are the findings generalizable to other issues and other regions? I address these briefly in turn. Legislation is one of the behaviors that external actors can influence most concretely, but it is naturally only one dimension of reform. Given the continued monitoring by the European institutions, however, there is good reason to believe the legislative concessions will have some beneficial, though hardly a cure-all, effect on ethnic relations. At least incidences of nationalist rhetoric have decreased, particularly in Slovakia and Romania, where they were commonplace in the early 1990s. Further, ethnic minorities are reaping benefits from the rights regarding education and language use. Naturalization in the Baltic states still suffers from implementation problems, primarily because of the magnitude of the problem, but progress is nevertheless occurring.
It is most natural to limit the conclusions of this study to political conditionality. As discussed, it would be unfair to conclude that this research proves that studies berating the ineffectiveness of economic conditionality are wrong. It is important to note, however, that this study does highlight that the converse is also true: disillusions with economic conditionality may not apply to political conditionality.
To some extent the findings are clearly specific to the EU and Europe. The insights will most readily apply to candidates such as Turkey and the Balkan states queuing for EU membership. The implications may be broader, however. NATO and the World Trade Organization (WTO) also rely on conditionality in entrance negotiations, and the Organization of American States and Mercosur have also had human rights related requirements. The Organization of Economic Cooperation and Development (OECD) has used political criteria in its Development Assistance Committee and is also debating more overt membership criteria for the swelling applicant pool. Thus, “clubs” will continue to play a large part in global cooperation. Insight about leverage from inclusion and exclusion may—or could if one so uses it—apply to them in the future.
The predicted scores for Table 4 are calculated as follows. If
is the predicted score, and k1 and k2 are the cut points, and uj is the error term, which is assumed to be logistically distributed in ordered logit, then the probability that Sj + uj lies between the two cut points is: