EIoP, Special Issue 2, Vol. 13 (2009), Art. 24

European Integration online Papers / ISSN 1027-5193
EIoP Logo    Special Issue 2, Vol. 13 (2009), Art. 24

How to cite?
Schwellnus, Guido, Lilla Balázs and Liudmila Mikalayeva. 2009. It ain't over when it's over: The adoption and sustainability of minority protection rules in new EU member states. In: Frank Schimmelfennig and Florian Trauner (eds): Post-accession compliance in the EU’s new member states, European Integration online Papers (EIoP), Special Issue 2, Vol. 13, Art. 24, http://eiop.or.at/eiop/texte/2009-024a.htm.

DOI: 10.1695/2009024

It ain't over when it's over: The adoption and sustainability of minority protection rules in new EU member states*

Guido Schwellnus
Center for Comparative and International Studies (CIS), ETH Zürich, Switzerland
E-Mail: schwellnus@eup.gess.ethz.ch

Lilla Balázs
Center for Comparative and International Studies (CIS), ETH Zürich, Switzerland
E-Mail: balazs@eup.gess.ethz.ch

Liudmila Mikalayeva
Center for Comparative and International Studies (CIS), ETH Zürich, Switzerland
E-Mail: mikalayeva@eup.gess.ethz.ch

Abstract: This paper conducts a multi-value Qualitative Comparative Analysis (QCA) of the formal adoption and sustainability of minority protection rules in four new EU member states (Poland, Romania, Estonia, Latvia) over a twelve-year period including pre- and post-accession phases (1997-2008) and five minority protection related issue areas (non-discrimination, language use, education, citizenship, integration of Roma) based on four conditions (external incentives, government position, veto players, size of minorities), in order to investigate under which external and domestic conditions minority protection and non-discrimination measures are adopted, maintained or revoked in new member states before and after accession to the EU.

Keywords: enlargement; minorities; educational policy; language policy; non-discrimination; Roma; citizenship; Poland; Romania; Estonia; Latvia; political science

Table of Contents:

1. Introductionto contents list

Since its inclusion in the “Copenhagen criteria”, the protection of ethnic and national minorities has been an important political condition set by the European Union (EU) for applicant states. Research so far has predominantly confirmed a strong influence of EU conditionality on the adoption of minority protection rules in Central and Eastern European candidate countries (Kelley 2004; Schimmelfennig, Engert and Knobel 2003; 2006; Schwellnus 2005). In a way, minority protection constitutes a paradigmatic case for rationalist explanations of rule adoption in reaction to external incentives: the contested and politicized character of minority rights makes it both a least likely case for alternative constructivist explanations (Checkel 2005, 10) and a crucial case for proving the effectiveness of external incentives in situations with considerable domestic opposition.

In order to establish the effectiveness of external incentives, researchers have mostly selected “hard” cases in which conditionality was applied and initial domestic resistance to adopt minority protection legislation was high. Although the possibility of purely domestically-driven rule adoption is acknowledged (albeit considered to be an “easy” case of self-socialization) and domestic adoption costs are an important explanatory factor determining whether conditionality is successful or not, some open questions remain to be answered. First, the approach does not allow any assessment about the relative empirical weight that external factors carry in comparison to domestic ones when explaining the adoption of minority protection rules in Central and Eastern Europe. Second, the effectiveness of conditionality may have been overstated by treating domestic factors as adoption costs: once initially unfavorable domestic conditions improve and rule adoption occurs subsequently, this is interpreted as an instance of successful conditionality, although the changed domestic conditions might have produced the outcome irrespective of external incentives. Third, the theoretical expectation for the sustainability of minority protection measures after accession is rather dim, because with the exception of non-discrimination on the basis of race and ethnicity, minority protection is not part of the acquis communautaire (De Witte 2000; Toggenburg 2000), so that when EU conditionality has ceased compliance is not enforced by the internal EU sanctioning mechanism. Hence, when external incentives are assumed to be the decisive factor explaining rule adoption, the revocation of these rules is to be expected once the incentives disappear (Sasse 2006; Schimmelfennig and Sedelmeier 2004, 675).

This paper builds on the insights of existing research on EU conditionality in the field of minority rights, but considers external incentives as only one possible explanation for rule adoption that can work in combination with as well as parallel to domestic factors. With its ability to account for causal complexity and equifinality, Qualitative Comparative Analysis (QCA) seems well suited to this end (see George and Bennett 2005; Ragin 2008). The paper conducts a multi-value QCA of the formal adoption and sustainability of minority protection rules in four new EU member states (Poland, Romania, Estonia, Latvia), systematically over a twelve-year period including pre- and post-accession phases (1997-2008) and five minority protection related issue areas (non-discrimination, language use, education, citizenship, integration of Roma) based on four conditions (external incentives, government position, veto players, size of minorities). The main research question is:

Under which external and domestic conditions are minority protection and non-discrimination measures adopted, maintained or revoked in new member states before and after accession to the EU?

The paper is structured as follows: part 2 presents the theoretical assumptions and hypotheses; part 3 elaborates on the case selection, the method and the operationalization of the outcome and the conditions; part 4 discusses the results of the QCA analysis based on the overall dataset as well as pre-/post-accession, country- and issue-specific subsets.

2. Theoretical background: External incentives vs. domestic factorsto contents list

The theoretical starting point of the paper is a rationalist external incentives model of externally driven rule adoption (Schimmelfennig and Sedelmeier 2004; 2005; 2007), which assumes actors to be rational utility maximizers calculating the material as well as political costs and benefits of rule adoption. From the perspective of the external incentives model, the main driving force of rule adoption is membership conditionality. “The dominant logic underpinning EU conditionality is a bargaining strategy of reinforcement by reward, under which the EU provides external incentives for a target government to comply with its conditions” (Schimmelfennig and Sedelmeier 2004, 662). The basic prediction of the external incentives model is that a candidate state “adopts EU rules if the benefits of EU rewards exceed the domestic adoption costs” (Schimmelfennig and Sedelmeier 2005, 12). External incentives alone are therefore not sufficient to induce rule adoption – they also have to surpass domestic adoption costs. If adoption costs are prohibitively high, rule adoption is not likely.

If domestic factors are conceptualized as adoption costs, which are assumed to be always larger than zero, it follows that high political costs may prevent rule adoption, while moderate costs lead the model to expect compliance as a result of effective conditionality. Governments of applicant countries are not assumed to gain benefits out of rule adoption in the absence of external incentives. The possibility of domestically driven rule adoption is acknowledged, but generally only in countries where domestic conditions were favorable before the onset of EU conditionality. In such cases, conditionality is not necessary and therefore also not applied. If conditionality is used, however, it is assumed that candidate countries incur at least moderate costs for compliance. This might lead to an over-estimation of the effectiveness of conditionality in cases where it was applied and rule adoption took place, but where domestic change led to conditions with positive net gains for rule adoption. With regard to minority protection, positive political gains might arise for governments that represent minorities, view them as an important electorate, or are ideologically leaning towards a pro-minority position.

We conceptionalize the domestic factors that might facilitate (or even in itself sufficiently cause) or inhibit rule adoption through three different conditions: first, the government position, whose policy preferences can be either in favor of (in the case of a government with a pro-minority orientation or under inclusion of the minorities themselves), indifferent or opposed to minority protection measures (the latter in case of nationalists forming or taking part in the government); second, the existence of veto players that might – depending on their policy preferences – block either positive proposals or the attempted revocation of existing rules; and third, the size of minorities, which can be interpreted as an indicator of the salience as well as possible financial costs of minority protection. These factors will be further specified below in the part dealing with the operationalization of the conditions.

The importance of domestic factors is even enhanced after accession. If external incentives are necessary to induce rule adoption in the candidate states, the prediction for the sustainability (let alone further development) of already adopted minority protection rules is rather pessimistic if conditionality is not replaced by other incentives, e.g. EU sanctions such as infringement procedures. To be sure, the external incentives model would not under all circumstances predict the complete formal reversal of externally induced minority protection measures (see Schimmelfennig and Trauner 2009 this issue). First, the revocation as well as the initial adoption of rules is strongly dependent on the domestic political constellation, i.e. the threat of a policy reversal would be imminent only in the case of political forces opposed to minority protection forming a post-accession government. Second, conditionality may have induced institutional changes (e.g. constitutional provisions) that cannot be reversed by simple majorities and are upheld by domestic control mechanisms (e.g. a constitutional court) acting as veto players. Third, it may be less costly to uphold formal legislation or to keep institutions in place that are costly to set up or change, but undermine implementation through cuts in funding or restrictive regulations.

Still, under certain political conditions we would expect formal rule adoption to stagnate and in extreme circumstances even formal legislation to be revoked. Because external incentives are expected to only temporarily upset the domestic equilibrium and the calculations of political actors, the external incentives model would not predict the establishment of a stable pro-minority consensus in the cases where the introduction of minority protection rules was initially contested and only achieved in reaction to conditionality. Once the incentives disappear or the reward is delivered, the domestic situation should revert to the status quo ante, because conditionality is not expected to induce preference change, socialization or internalization. Elections do of course change actor constellations and may bring more or less minority-friendly governments into power, but we should not see the emergence of a stable equilibrium in favor of minority rights, if it did not exist prior to accession and only external incentives triggered domestic change.

Three main hypotheses regarding the conditions for positive as well as negative change in formal rule adoption can be derived from these assumptions:

H#1: Positive change should occur independently of any external incentives, if domestic conditions are favorable, i.e. a pro-minority government and no nationalist veto players exist, and if minorities are small.

This “domestic path hypothesis” captures a situation with favorable domestic conditions, in which rule adoption can be expected without any external incentives. In general, it is often assumed that such cases are predominantly to be found among the early democratizers, which are often excluded from analyses of the effectiveness of EU conditionality. However, minority rights are also contested in seemingly “easy” cases, and conversely, positive domestic constellations can occur in countries facing strong EU conditionality.

H#2: Strong and determinate external incentives should induce positive change also in cases where domestic conditions are less favorable, e.g. if large minorities, indifferent or even moderately nationalist governments and/or veto players exist, as long as they are not ultra-nationalist and would incur prohibitively high political adoption costs.

This “external incentives path hypothesis” lies at the heart of the external incentives model: strong and determinate conditions should overcome domestic resistance, if the expected reward for compliance outweighs the domestic adoption costs. Since authoritarian or ultra-nationalist governments that would incur prohibitively high political adoption costs (such as the Mečiar government in Slovakia) no longer existed in any of the candidate states after accession negotiations started (in fact, this was a necessary condition for the negotiations to start), we should expect conditionality to be effective in all cases where it was applied, even with negatively oriented governments and veto players. The incentive of accession should supersede these obstacles.

H#3: Negative change should occur when external incentives are weak, indeterminate or absent and a combination of nationalist government and no pro-minority veto player exists.

This “revocation hypothesis” follows from the external incentives model, because in this case strong domestic preferences against minority protection are not countered by external incentives. Since the EU can be expected to monitor compliance and be watchful if problematic domestic conditions arise (e.g. after an election), the configuration of absent EU leverage through conditionality and negative domestic conditions should only arise after accession. Hence, we should expect revocation only to set in after accession, and then to depend on domestic conditions in the absence of external incentives.

3. Case selection, method and operationalization to contents list

As countries we selected four new member states: Poland, Romania, Estonia and Latvia. The selection follows the desire to cover countries with variation regarding different causal paths to rule adoption. Estonia and Latvia are considered to be “hard” cases, in which external incentives met strong domestic resistance, and which conversely also produced some theoretically puzzling instances of rule adoption (Schimmelfennig, Engert and Knobel 2006, 235). Romania is usually taken to be a paradigmatic case of successful conditionality, although the positive developments only set in after the minorities themselves joined the government, so that domestic conditions alone might be a sufficient explanation (see Kelley 2004, 159). Poland is mostly presented as an “easy” case of self-socialization, where conditionality was not necessary to induce rule adoption (Schimmelfennig 2000, 133), yet the development of a comprehensive protection of minorities has been slow and contested (Schwellnus 2005; Vermeersch 2007).

The analysis covers a time-period of twelve years from 1997, when the European Commission’s Opinions on the candidate countries were issued and the accession negotiations with the first countries started, to 2008, i.e. it covers both pre- and post-accession phases. Five issues are under investigation: non-discrimination, language use, education, citizenship, and the integration of Roma. The inclusion of non-discrimination introduces external incentives after accession, since the adoption of anti-discrimination legislation is part of the EU acquis and therefore backed by EU sanctions against non-compliance. Not all of the five issues are relevant for every country or at every point in time. To avoid including irrelevant cases, we excluded cases where the highest level of protection was already reached prior to the time-period in question, so that further improvement was not possible by definition.

The method applied in the paper is multi-value Qualitative Comparative Analysis (mvQCA) as developed by Lasse Cronqvist (Cronqvist 2007; see Rihoux and Ragin 2009, 69-86), which is an extension of Charles Ragin’s crisp-set or csQCA (Ragin 1987). QCA focuses on the configuration of conditions that are necessary or sufficient to bring about a specific outcome. It is specifically designed to rigorously handle and analyze a larger number of cases than usual case studies – without, however, applying statistical, regression-based techniques. QCA is capable of examining complex patterns of interactions between conditions and contains procedures to minimize these patterns in order to achieve parsimony (Ragin 1987, 121-123). The data for QCA is arranged as a “truth table” (see Appendix II). That is, each possible configuration (combination of values of the conditions) is represented as one row together with the associated value of the outcome. Configurations for which no empirical case is reported constitute “logical remainders”, which might be used as counterfactuals to achieve a more parsimonious result. The value added of mvQCA as compared to the dichotomous coding necessary for csQCA is that it can handle conditions (but not outcomes) with more than two values. Finally, the truth table is analyzed and reduced with procedures of combinatorial logic to arrive at a solution specifying a parsimonious combination of necessary and sufficient causes for the presence or the absence of the outcome to be explained. Further measures of the degree to which a QCA solution explains the outcome are its consistency, i.e. whether all cases under this constellation show the expected outcome or only a percentage of them, and coverage, i.e. how many of the cases with a positive outcome are covered by the solution (Ragin 2008; Rihoux and Ragin 2009; Schneider and Wagemann 2007).

In correspondence with the combinatorial logic of QCA, a case for the use in this paper is a specific configuration of conditions, which can vary across countries, over time, and across issues. Whenever the value of a condition changes for one country in one issue area, this new combination constitutes a new case. The main reasons for coding a domestic situation as a new case are elections and government changes. Although the Commission progress reports were delivered on a yearly basis, we aggregated external incentives over the time of a government. Overall, the analysis is based on 93 cases (62 pre- and 31 post-accession), which cover 22 of 36 possible configurations.

3.1. Operationalizing the outcome: Formal rule adoption in five minority protection related issue areas to contents list

As the outcome to be explained we selected changes in formal rule adoption – i.e. legislation or the adoption of policy programs, but not their implementation, application or enforcement – in relation to the prior level of adoption(1). For the use of this paper, we coded formal rule adoption in five minority protection related issue areas: non-discrimination, language use, education of and in the minority language, citizenship (applicable only to Estonia and Latvia) and the integration of Roma via specific policy programs. In accordance with most definitions of national and ethnic minorities we restrict the application of the term to traditionally resident minorities, thereby excluding immigrants. However, in the case of Estonia and Latvia we include the Russian-speaking population without citizenship, although their status as national minorities might be disputed. Still, international institutions generally consider them to be national minorities.

Non-discrimination in the context of EU enlargement predominantly relates to the transposition of the EU acquis communautaire represented by the “Framework Directive on equal treatment in employment and occupation” (Council Directive 2000/78/EC, OJ 2000 L303, 16-22), and the “Directive on equal treatment between persons irrespective of racial or ethnic origin” (Council Directive 2000/43/EC, OJ 2000 L180, 22-26), the so-called “Race Equality Directive”. The coding of the level of rule adoption therefore mainly follows the degree of transposition of these Directives, although it is formulated in a way that it can also be applied to the time before the directives were adopted. The lowest level (--) is the complete absence of any non-discrimination rules. The next level (-) is characterized by general provisions such as an equality clause in the constitution, but without any specific codification in simple legislation. A medium level (=) is reached, if non-discrimination clauses are inserted in specific laws (e.g. the labor code), which amounts to a partial transposition of the EU Directives. A high level of rule adoption (+) corresponds to comprehensive anti-discrimination legislation constituting a full transposition of the EU’s anti-discrimination acquis. The highest level (++) is achieved, if non-discrimination laws exceed the minimum requirements of the Directives, especially by allowing for positive measures to support or compensate discriminated groups (“affirmative action”).

Language use refers to the use of minority languages in different aspects of private and public life (see Pan and Pfeil 2002, XXV). On the lowest level of rule adoption, the use of minority languages is only allowed in private (--). The higher levels are determined by how many of the following aspects of language use are permitted: first, whether general human rights with regard to language use can be utilized by minorities, such as the use of a translator before court (+1); second, whether minority languages may be used beyond the purely private sphere, e.g. in professional life, but not yet on the official level (+1); third, whether the minority language may be used for public signs (+1); fourth, whether it can be used in official documents and before state authorities, at least in areas with considerable minority population (+1).

Education for minorities includes two elements: education of and in the minority language (see Pan and Pfeil 2002, XXV). The lowest level of educational rights (--) corresponds to no education of or in a minority language. The most fundamental educational provision (-) is that children belonging to minorities have the right to learn their mother tongue in school. A medium level (=) of educational rights is that at least parts of the curriculum (e.g. history or religion) are taught in the minority language. A high level of protection (+) is achieved when minorities have the right to educate their children completely in the mother tongue, either in state schools with a complete syllabus taught in the minority language or in specific minority schools. The highest level (++) is reached, when minorities have their own universities.

Citizenship refers to the level of difficulty of naturalization procedures members of minorities have to go through to become citizens of the state, in case a large part of minority population does not possess its citizenship. The lowest level of rule adoption (--) corresponds to the practical impossibility for a large number of persons belonging to minorities to accede to citizenship (see for example the so-called “window system” in the Latvian case). A low level of rule adoption (-) means that the naturalization process is difficult and does not foresee exceptions for specific categories of people, such as seniors, disabled, juniors, etc. As medium level (=) are coded instances in which the naturalization process is difficult or rather difficult, but sizeable exceptions exist for specific categories of people. Granting exceptions for important groups and simplifying procedures in general is usually a clear sign of liberalization in citizenship legislation. A high level of rule adoption (+) means that not only exceptions exist, but also incentives are created in order to motivate people to naturalize. Finally, the highest level (++) corresponds to extremely easy, virtually automatic, procedures for naturalization of persons belonging to minorities.

Integration of Roma is an issue area that is not addressed via general legislation but is usually the focus of specific policy programs and action plans. The absence of any specific Roma program is coded as the lowest level of rule adoption (--). As there is no clear hierarchy of measures readily available, higher levels are represented by how many of the following areas are addressed: housing, health and social security, education and training, dissemination of information and awareness-raising. In addition, we include the temporal (only short-term solutions to specific problems, mid-term programs over at least a year, permanent programs) and territorial (local regional or national application) scope of the program. The highest level (++) is reached, if all these aspects are incorporated into a comprehensive program for the integration of Roma.

On the basis of this information about the level of protection in each issue, we firstly excluded all cases where the top level of rule adoption was already reached, so that further improvements were impossible. Secondly, we used the level coding as our main indicator for the outcome: positive change in formal rule adoption. If a policy program or piece of legislation that lifts the level of adoption in the issue area by at least one level is adopted during the period of one case, the outcome was coded as “1”(2). Additionally, we also coded as positive change the signature or ratification of important international conventions such as the Council of Europe’s Framework Convention on the Protection of National Minorities (FCNM) or the European Charter for Regional or Minority Languages (ECRML), since they are considered to be major improvements in formal minority protection despite not having direct effect(3).

3.2. Operationalizing the conditions for rule adoption: External incentives, government position, veto players, and the size of minoritiesto contents list

Based on the theoretical assumptions of the external incentives model, the three hypotheses spell out under which configuration of external and domestic conditions a positive development (H#1+2) and a revocation of already adopted rules (H#3) can be expected. In the following, these conditions shall be specified.

External incentives: Before accession, the EU gives candidate states positive incentives for compliance with conditionality in the form of a membership perspective. According to the external incentives model, rule adoption depends on the following external conditions: the size of the reward, the credibility of delivering or withholding the reward, the strength of conditionality, and the determinacy of conditions. For the countries and the pre-accession time-period covered in this analysis we assume that both the credibility of EU conditionality and the size of the most important reward – EU membership – were always high. This leaves the strength of conditionality and the determinacy of conditions as the central external factors to determine the likelihood of rule adoption.

After accession, conditionality no longer applies as the reward is paid out, but in areas that are part of the acquis the internal sanctioning mechanism of the EU sets in, so that negative incentives (sanction threat) against non-compliance replace positive incentives (membership reward) for compliance. In the domain of minority protection this only applies to non-discrimination on the basis of race and ethnicity, but not to minority rights proper. The same criteria used for conditionality apply to EU sanctions as well, although in addition to credibility also the determinacy of the rules is constant and high, since it relates to the implementation of legally defined rules such as EU Directives.

External incentives are coded dichotomously, i.e. as either present or absent. We code strong and determinate external incentives as present when in the pre-accession phase measures in the respective issue area are explicitly and clearly demanded, e.g. in the European Commission’s progress reports or as short-term priorities in the accession partnerships, or if after accession a specific sanction threat is issued, e.g. in the form of a Commission statement threatening the start of infringement procedures against a member state because of non-compliance with Community law. It is not enough that the issue falls under the general condition to transpose the acquis or is part of the vague Copenhagen criteria regarding minority protection, nor does it suffice when the issue area is mentioned without any specific requests.

Government position: According to Kelley (2004), the government participation of national minority parties on the one hand and nationalist parties on the other is an important factor in determining the state policy towards minorities and the reaction to external demands to protect minorities. Coalition governments under the inclusion of parties representing national minorities are expected to be willing to implement minority rights, whereas governments with a strong nationalist influence are likely to resist this. The straight-forward rationale behind this intuitive hypothesis is that political decisions follow directly the policy preferences of the ruling decision-makers.

We code the government position as a multi-value condition with three possible values. The GOV(+) position, i.e. a positive commitment towards minority protection, can be expected, if parties representing the minorities themselves are part of a coalition government, but also for governments that do not include minority parties but are comprised of parties that have minority protection as part of their party programs or election manifestoes. The neutral position GOV(=) applies to governments comprised of parties with an indifferent position to minority issues, or heterogeneous coalitions. In this case we would not expect the government to adopt minority-friendly policies out of their own initiative, but they are also not opposed to such policies or incur high political costs when complying with external demands to adopt or implement minority protection rules. Governments including nationalist parties are coded as GOV(-), i.e. governments with an anti-minority position, since nationalists predominantly portray minorities as a threat to national identity and unity and are likely to incur high political costs for adopting minority protection rules.

Veto players are defined as “actors whose agreement is required for a change of the status quo” (Tsebelis 2002, 17). Veto player theory predicts increasing policy stability with a higher number of veto players, because it becomes increasingly likely that a change from the status quo will be blocked. However, not only the institutional power to block a decision is important, but also the policy preferences of the veto players. A player will only veto a decision if s/he has both the institutional capability to veto and preferences that differ from the government that proposes a policy.

Instead of assuming a random distribution of policy preferences and conducting a probabilistic analysis on the basis of the number of veto players, we assume that one veto player is sufficient to block a decision and code the political positions of possible veto players in analogy to “government position”. Our coding takes three types of possible veto players into account: first, the president, who generally has veto powers; second, a parliamentary majority against the government, either in a second chamber (in bicameral systems) or the first chamber (in case of a government lacking a parliamentary majority)(4); and third, the constitutional court. Coding follows the strongest veto player (in terms of intensity of preferences) with a policy preference contrary to the government’s preference as coded under “government position”.

As in the “government position” condition we code veto players according to a multi-value coding scheme with three possible values. A VETO(=) coding indicates that veto players are absent or indifferent, so that any decision may be passed. Veto players with a VETO(+) coding are assumed to be permissive for the adoption and implementation of positive minority protection measures but veto their revocation or the adoption of restrictive rules. Veto players coded as VETO(-), by contrast, are expected to be permissive for negative measures or the revocation of positive rules, but block any improvement in terms of minority protection.

The size of minorities acts as an indicator for the political salience and potential financial costs of adopting minority protection rules: if the respective measures only apply to very small minority communities, this is not likely to be either politically very controversial or particularly costly, whereas large minorities are more likely to be considered “problematic” politically, and the provision of costly goods is also a much higher financial burden if it applies to a large number of people. Of course, large minorities can also act as a supportive factor for rule adoption, e.g. by constituting a sizeable electorate or by being able to mobilize. However, if minorities are large and organized enough to exert significant influence, this is most likely to be via their own political parties and their participation in government, which is already covered under “government position”. Hence, we assume large size to be a hindering factor for rule adoption.

The size of minorities is coded dichotomously. It is coded as large, when either the Roma community (for the Roma integration issue) or national minorities (for the other issues) constitute at least five percent of the population. Consequently, both kinds of minorities are small for Poland, Estonia and Latvia have large national minorities but small Roma communities, and in Romania both groups are coded as large (see for the official numbers Pan and Pfeil 2002).

4. Analysisto contents list

In the following, the results of the QCA analysis are discussed. We conducted four different runs: first, for the complete dataset of all cases; second, for the pre-and post-accession phases separately; third, for the cases representing each country; fourth, for each issue area. The formal QCA analysis, conducted with both the TOSMANA (Cronqvist 2007) and fs/QCA3.0 (Ragin, Drass and Davey 2006) software under the crisp-set “truth table algorithm”(5), was only used for positive change as the outcome(6). The conditions under which revocation occurred could easily be determined by hand, because only two such cases exist in the dataset (see the coding tables in Appendix I). With our 93 overall cases we cover 22 of the 36 possible configurations, with 14 logical remainders, i.e. configurations of conditions without empirical cases. However, when smaller subsets of cases are analyzed separately – especially in the country- and issue-specific analyses – the problem of limited diversity increases.

We chose a conservative strategy both for the consistency cutoff and the treatment of logical remainders. Regarding the former, only configurations that consistently show a positive outcome (i.e. a consistency of 1.00) are included in the solutions, although the comparatively high number of cases tends to produce contradictory configurations and would have lent itself to introducing some quantitative element in the form of consistency or frequency cutoffs. However, the contradictory configurations are always well below the consistency threshold of 0.75 that Ragin considers to be the minimum for an inclusion – there are no configurations in which positive outcomes almost always occur with only one or few negative cases (Ragin 2006, 293). Only in analyses in which no fully consistent solution is to be found, we consider the best contradictory configuration (represented in italics in Appendix III). The solutions produced in these cases are, however, clearly no indicator of a configuration that is sufficient for the outcome.

We also did not include counterfactual cases by allowing simplifying assumptions on the basis of logical remainders in order to come to more parsimonious results – only the “complex” solution is used, i.e. the one that is represented by actual cases. The aim is not to provide the most simple theoretically possible solution based on a minimal set of conditions (which the “parsimonious” solution does), but to identify the configurations under which the outcome to be explained consistently occurs in our cases.

In accordance with official QCA nomenclature, solution terms are represented as upper case if the condition is present and lower case if it is absent (lower case starting with a capital letter indicates only the name of the variable, not its value). Hence, “VETO(-)” means the presence of a nationalist veto player, whereas “veto(-)” indicates its absence – in this case indicating that both positive and neutral veto players are combined under this condition. “*” is a logical “and” operator joining the elements of one solution term, whereas “+” is a logical “or” that is placed between different equifinal solution terms.

4.1. All casesto contents list

The first analysis on the basis of all 93 cases produces two fully consistent solutions revealing two equifinal paths to positive change – one domestic and one including external incentives as a necessary component:

GOV(+)*VETO(+)*size + INCENT*GOV(+)*VETO(-)*SIZE

Together the two solutions cover roughly a third of all cases with a positive outcome (ten out of 29). The remaining instances of positive change are situated in contradictory configurations, none of which can be classified as almost always leading to a positive outcome, because the highest consistency achieved by such contradictory configurations is 0.5, i.e. the number of positive outcomes reported for the constellation is outweighed by an at least equal number of negative cases. Both fully consistent solutions are theoretically meaningful and confirm the theoretical expectations in principle.

The domestic solution covers a “best case” scenario, in which a pro-minority orientation of both the government and potential veto players combines with small minorities, whose accommodation is assumed to be neither costly nor politically problematic. Examples for this configuration are found in Poland, where it accounts for seven positive changes between 2003 and 2004 under the Miller (pre-accession) and Belka (post-accession) governments. Both governments were led by the post-communist Social Democrats (SLD), which had adopted a pro-minority position early after the democratization. The governments lacked a parliamentary majority after the minority-skeptical Peasant Party (PSL), which rendered the overall position of the prior government neutral, had left the coalition. Still, the government could count on support by the liberal opposition with regard to minority rights (see Schwellnus 2005), so that the minority-skeptical opposition parties did not possess a blocking majority in parliament. Since President Aleksander Kwaśniewski also belonged to the SLD, potential veto players showed a pro-minority position nonetheless. Strong conditionality prior to or sanction threats after accession were present only in the field of non-discrimination, whereas all other issues remained part of the weak conditionality of the general Copenhagen criteria. In this phase, Poland signed the ECRML, amended the Labor Code to partially transpose the EU Race Equality Directive in workplace-related areas, replaced the prior “Małopolska” pilot program for the integration of Roma (which had ended) with a long-term program for the Roma community in Poland, thus expanding the temporal and territorial scope considerably, and finally adopted a Law on National Minorities and Regional Languages.

This combination is sufficient for positive change, irrespective of external incentives being present (two cases) or absent (five cases). This is not as trivial as it might seem. First, the domestic explanation is the empirically more important of the two solutions with a coverage of 0.24 (seven cases) as compared to only 0.10 (three cases) explained by the external incentives solution. Second, comparison with other constellations reveals the importance of the constituent causal factors as necessary conditions for the solution. Both positive government position and small size of the minority appear to be necessary for a positive result in the absence of external incentives: With large minorities, none of the three cases with positively oriented governments and veto players (two without and one with external incentives; all under the centre-left Kallas government in Estonia 2002-2003) show a positive outcome. The “second best” situation with neutral veto players, which theoretically should still produce a positive result, because indifferent veto players are not expected to block positive government proposals, also fares badly: none of the three cases with conditionality present (non-discrimination under the short-lived Emsis government in Latvia 2004 and the two Ansip governments in Estonia since 2005) and only one out of eight cases in its absence (language use in Estonia under the first Ansip government as opposed to the same issue under the second Ansip government, citizenship under the two abovementioned Ansip governments as well as non-discrimination and education in Romania for the Popescu-Tãriceanu government from 2004) report a positive outcome.

In the absence of a pro-minority government the consistency with a positive outcome sharply declines as well. As the most significant example, in the well documented configuration with a neutral government, minority friendly veto players and small minorities in the absence of external incentives, five positive cases are outweighed by twelve negative outcomes, leaving the overall consistency of the configuration at 0.29, i.e. no higher as the general percentage of positive results in all cases. With neutral veto players it fares even worse (only one positive out of five cases).

In addition, the not sufficient but nonetheless noteworthy consistency of the configuration incent*GOV(+)*VETO(-)*SIZE (0.4, two out of five cases) suggests that government position is the most important of the domestic variables. One example for this configuration is given by the center-left PDSR (Romanian Party of Social Democracy, later Social Democratic Party/PSD) government in Romania, headed by Adrian Nãstase, who governed from 2000 to 2004 with the negotiated support of the Democratic Alliance of Hungarians in Romania (UDMR) on the basis of yearly renewed cooperation protocols that spelled out the exact terms and conditions of cooperation. A leading figure of the immediate post-1989 Romanian political scene dominated by salient nationalism, PDSR-nominated President Ion Iliescu constitutes a potential anti-minority veto player.

In this context, one of the main positive developments of this period was the adoption of the Law on Local Public Administration of 2001 granting the right to communication with state authorities in the mother tongue, as well as bilingual street signs in localities where minorities represent more than 20 percent of the population. Further provisions for minority language use were included into the 2002 Audiovisual Law and the 2002 Law on the Status of Policemen. The improvements brought to minority language use and education were further reinforced by the inclusion of special provisions referring to these fields in the amended Romanian Constitution of 2003. Very importantly, the National Strategy for Improving the Condition of the Roma, a “comprehensive and high quality document” (European Commission 2001b, 29), was also adopted in this period.

The role of veto players, however, is less clear, as there are no empirical cases with small minorities, a pro-minority government, and either neutral or nationalist veto players. Hence, while the theoretical assumption is that neutral veto players should not block change initiated by pro-minority oriented governments, whereas nationalist veto players should prevent any positive development, this is not supported by any empirical case.

The second solution based on the presence of external incentives (INCENT*GOV(+)*VETO(-)*SIZE) adds consistently positive results under less favorable domestic conditions, namely with large minorities and nationalist veto players. This configuration is illustrated by the case of Romania and its 1996-2000 coalition government of a predominantly right-of-center ideological orientation including the UDMR. The cooptation of the minority party into the government was contested within the governing coalition itself, had no programmatic underpinnings, and the overall coalition can be considered moderately pro-minority at best. This ambivalent government stance on minority issues was complemented by the recurring phenomenon of spontaneous loss of parliamentary majority by the government on minority issues, which has been coded as a potential anti-minority veto player.

As for external incentives, the lack of a coherent strategy to tackle the Roma issue was the main point of concern of the European Commission in its Romania Progress Report of 1998. Subsequent Accession Partnerships emphatically called for the elaboration and implementation of a “strategy to improve economic and social conditions of the Roma” as a short-term priority (European Commission 1999c, 4), which is coded as strong and explicit conditionality. In the field of education, the Commission reports in this period specifically mention the Petőfi-Schiller multicultural state university, the legal basis for which “has still to be completed” (European Commission 1999b, 18). While the establishment of the multicultural state university did not materialize, the adoption of an amended Law on Education in 1999 constituted an important development in the area of education. The new law satisfied most minority demands by removing previous limitations on education in the mother tongue. Developments in the field of Roma protection were, however, limited to legislation in the field of access to education. As a considerable development in the field of non-discrimination a highly encompassing government Ordinance on Preventing and Punishing All Forms of Discrimination was also adopted in August 2000.

However, only three cases are covered by this “external incentives” configuration, and it is puzzling from a theoretical perspective that the configuration INCENT*GOV(+)*SIZE is fully consistent with a positive outcome in combination with VETO(-), but not when combined with the more favorable veto player coding VETO(=) in three cases or even VETO(+) in one case. The role of veto players remains thus rather unclear, because in the cases examined positively inclined governments receiving explicit incentives to make formal changes to the legislation complied only if an anti-minority veto player was present but did not do so with neutral or positive veto players.

Such configurations occurred, for example, in Estonia under the government of Andrus Ansip (2005-2007). This government was moderately pro-minority, according to its program and its popularity in the minority electorate. External incentives were clearly present, as in this period the adoption of comprehensive anti-discrimination legislation became a formal requirement for Estonia as a member of the EU, and there was no anti-minority veto player. However, no positive change in the area of non-discrimination occurred. The configuration with a positive veto player present applies to the non-discrimination issue in Estonia under the centre-left coalition government of Siim Kallas (2002-2003), which was moderately in favor of minorities. Despite clear incentives for legislative action and with the President Arnold Rüütel acting as a positive veto-player, a comprehensive anti-discrimination law still was not adopted.

Moreover, the same domestic constellation as in the solution already achieves positive change without external incentives – incent*GOV(+)*VETO(-)*SIZE – in two of five cases (consistency 0.4), both reported in Romania. On the other hand, in combination with a neutral government and pro-minority veto players, the configurations with external incentives outperform those without them, both with small and large minorities. They reach a consistency of 0.5 (two out of four cases) each, as compared to 0.29 (five out of 17) and 0.17 (one out of six) for small and large minorities respectively. Still, a closer look at the cases covered under these configurations is inconclusive.

In Poland, the configurations (always with small minorities) apply to the first three governments included in the analysis. Under the conservative-liberal coalition government headed by Prime Minister Jerzy Buzek (1997-2000), which is coded “neutral” despite the inclusion of the pro-minority liberal UW because the main coalition partner AWS was a conglomerate of several parties ranging from moderately pro-minority oriented liberal conservatives to a small number of catholic nationalists, no change is reported in any of the issue areas, although the Commission’s opinion on Poland’s application for membership in 1997 noted that the “Roma (or gypsies), who are few in number in Poland, are sometimes the victims of violence or discrimination” (European Commission 1997, 18). After the UW had left the government (as did the nationalist right wing of the AWS), the Buzek minority government (2000-2001) ratified the FCNM and initiated a pilot program for the Roma community in the Małopolska region, which was welcomed by the EU (European Commission 2001, 24). Hence, positive developments occurred in areas with and without external incentives. After the change towards the leftist-rural Miller government (2001-2003), in which the minority-skeptical PSL coalition partner rendered the government “neutral” despite the positive position taken by the leading SLD, positive change was achieved in the field of language use (where no conditionality was applied) by an ordinance of the Minister of the Interior regarding the translation of Polish signs into foreign languages in March 2002, but not in the field of non-discrimination, although the necessity to transpose the Race Equality Directive was reiterated in the Commission reports (European Commission 2000, 57; 2001a, 22 and 68; 2002, 28), indicating persistent strong conditionality in this area.

In Estonia, the Laar government (1999-2002) represents the constellation with an indifferent government, pro-minority veto player and external incentives present in both the issue-areas of non-discrimination and citizenship, but a positive change is only reported in the latter case. The Parts government (2003-2005), on the other hand, shows positive change in three issue areas: non-discrimination (with external incentives), citizenship and Roma (both without external incentives). In Latvia, positive change in the configurations in question is coded only once in the field of Roma integration under the Repše government (2002-2004), where no external incentives were present, but not in the same issue area under previous governments which all represent the same configuration, not for other issues in the absence of conditionality under the Repše government (language use and citizenship), and also not for the non-discrimination issue at the same time, where external incentives applied.

In sum, although external incentives are a necessary condition in this second fully consistent solution, they seem to have a limited impact, mainly reinforcing domestic factors, most importantly a pro-minority oriented government. Given that in the cases under investigation there is no ultra-nationalist or anti-democratic government that would bear prohibitively high political costs, external incentives do not fulfill the theoretical expectation of overcoming considerable domestic opposition.

Negative change is only reported in two cases, both under the following configuration:

INCENT*GOV(-)*VETO(+)*SIZE

Both cases occurred in the issue area of language use in 1999: first in February – in Estonia, then in December – in Latvia. To start with the Estonian case, it was the center-right “Triple Alliance” government of Mart Laar that adopted amendments to the language legislation introducing regulations of professional use of language. A list of professions and corresponding levels of proficiency in the state language, necessary for the occupation of offices was issued. It concerned not only the public, but also the private sector and covered the whole country, including the regions almost exclusively populated by minorities. It has to be noted that the overall stance of this government on the issue of minorities was rather neutral, as the nationalist part of the coalition was mostly pragmatic at the end of the 1990s, but still intransigent on the question of language use. The adopted amendments called a strong critique by European organizations including the European Commission, who dedicated a major part of its 1999 Progress Report to the negative impact of this law (European Commission 1999a, 14-15). The law was sent back to parliament by the President, where it was amended with clauses delimiting its scope. These limitations, which satisfied the European organizations, did not prevent the law from regulating the use of language in private sphere, so that the negative change compared to the situation before the adoption of the amendments was not reversed.

The Latvian case is similar, although somewhat more dramatic. The new language law, adopted under the center-right coalition government of Andris Šķēle, prohibited the use of foreign languages in dealings with authorities, except in emergencies. The previous – rather generous – version of the law had allowed for the submission of documents in Russian, German and English. While the old Civil Code permitted the use of foreign languages in some cases, the new law excluded these possibilities. In addition to this, the use of public signs in foreign languages was also forbidden. As in the Estonian case, the first version of the new law met opposition from European organizations and was sent back to parliament by the President. The final version, although still containing the abovementioned language restrictions alongside some limitations of scope, was finally judged acceptable.

Although in both cases international organizations finally accepted the outcome, we code it as a negative change, since in both cases more restrictive language regulations were passed. This disconfirms the theoretical assumption that revocation will predominantly happen in the absence of external incentives, while strengthening the observation made for positive change, namely that government position and the size of minorities are more important than veto players – pro-minority oriented veto players should have prevented revocation. The solution INCENT*GOV(-)*VETO(+)*SIZE is, however, not even close to constituting a sufficient condition for revocation. In fact, under the configuration all three possible outcomes occur: positive change (one case), status quo (five cases), and negative change (two cases). The consistency for revocation is therefore only 0.25 (two out of eight cases).

The case of a positive change under this configuration occurs in Latvia: in the issue of citizenship, under a center-right coalition government of Guntars Krasts led by the nationalist For Fatherland and Freedom Union, an important liberalization of naturalization legislation took place. The system of naturalization windows was abolished and simplified procedures were introduced for certain categories of persons (children and elderly), thus marking a change of even two levels on our scale at once. Even the opposition of some parliamentarians, who insisted on putting the issue to a referendum, could not prevent the amendments from being passed.

As an illustration for the case when the same constellation led to the preservation of the status quo, the issue of citizenship in Estonia can be cited. There, in the context of a moderately anti-minority position of the government of Mart Siimann on this issue, the presence of clear incentives and of a positive veto player could not push for an adoption of favorable pieces of legislation.

In any case, the very small number of negative cases prohibits any strong conclusions regarding the conditions for revocation.

4.2. Pre- and post-accession phases to contents list

In a second step, the 62 pre-accession cases were analyzed separately from the 31 post-accession cases in order to determine whether the solutions discovered in the overall analysis apply to one of the phases only, or whether some of the contradictory configurations are based on diverging trends before and after accession and can therefore be dissolved by treating them separately.

A look at the general percentages of positive change confirms the hypothesis that rule adoption should slow down significantly after accession. This general trend shows throughout most country- and issue-specific subsets (see second column of the table in Appendix III). All countries and issue areas taken together, positive change is reported for slightly over one third of the pre-accession cases (22 out of 62), but only about one fifth of the post-accession cases (seven out of 31). Of course, this could be due to the fact that a generally higher standard later in the process makes further improvements less likely, but since we excluded all cases in which the maximum level of protection was already reached, at least the possibility of positive change was present in all cases. On the other hand, the theoretical expectations regarding the revocation of legislation are again not corroborated, since both instances of negative change occurred prior to accession.

The QCA analysis produces the same two solutions already present in the overall analysis. This means that no contradictory configurations are resolved by splitting the dataset into pre- and post-accession cases. However, the solution including external incentives only applies to the pre-accession phase, whereas the domestic solution is present in both phases. Of course, this could be explained by the fact that there are not only considerably less post-accession cases, but EU conditionality ceased for all issue areas except non-discrimination. Still, in the remaining eight cases where external incentives were present after accession, only those report a positive outcome that are already explained by favorable domestic conditions.

The domestic solution covers three of the seven instances of positive change after accession (coverage 0.43). For the pre-accession phase this means that the coverage of the domestic solution decreases in comparison to the overall analysis, whereas the relative empirical importance of the external incentives solution slightly increases. However, the domestic solution still explains one more case than the one including external incentives (coverage 0.18 to 0.14, or four compared to three cases explained by the respective solution term) and overall solution coverage is comparatively lower for the pre-accession (0.32, i.e. seven out of 22 positive cases) than for the post-accession phase.

4.3. Country-specific analysisto contents list

The separate analysis for each country shows extreme variation in the configurations that produce positive outcomes. First, we can distinguish Poland and Romania as countries with a high percentage of positive change, especially prior to accession (general consistency of above 0.5 before accession and above 0.4 overall in both cases) from Estonia and Latvia as countries with a lower percentage (general consistency of overall 0.21 in the former and 0.25 in the latter case). The reason of the apparently low performance of Estonia and Latvia is, however, not so much the small amount of positive change in the period under study, but high political instability leading to shorter government duration. This creates more cases, while the number of positive outcomes recorded overall is roughly equal to Poland and Romania. Also, as both Estonia and Latvia guaranteed the highest level of protection in education already in 1997, this best performing issue is automatically excluded from our analysis from the very start, while no issue area had the highest level of protection in 1997 in Poland or Romania.

What is more important from the perspective of QCA, only Poland and Romania show fully consistent QCA solutions. These solutions – one for each country – correspond to the two equifinal solutions discovered in the overall analysis: Poland represents the domestic solution GOV(+)*VETO(+)*size, Romania the external incentives based solution INCENT*GOV(+)*VETO(-)*SIZE. In both cases the coverage is comparatively high (roughly 0.6 for both countries), i.e. the solutions cover a good amount of the cases within each country that show a positive outcome.

By contrast, Estonia and Latvia contribute nothing to the overall explanation, because they both show not a single fully consistent solution. If we look for the contradictory configurations with the best consistency, Estonia has one solution with a consistency of 0.67, i.e. two thirds of the cases are positive (two out of three) and a coverage of 0.40 (two out of five cases with a positive outcome):

INCENT*GOV(=)*VETO(+)*SIZE

These two positive cases were: first, the adoption of liberalizing amendments to the citizenship legislation under the center-right coalition government headed by Mart Laar (introducing simplified procedures of naturalization for young graduates and disabled persons); and second, the introduction of anti-discrimination clauses into the Law on Employment Contracts and the Penal Code under the government of Juhan Parts from the conservative-populist Res Publica party. In both cases, the center-right coalition governments are considered to be rather neutral on the issues in question – other issues, such as economy, dominated political life.

This solution seems to be consistent with the theoretical assumptions: external incentives should be able to induce an indifferent government to adopt rules even if they apply to large minorities. However, the same solution with positive governments did not result in a positive change on two occasions when it occurred. This finding together with the low consistency counter the possible conclusion that external incentives might have worked effectively under worse conditions in Estonia than in Romania.

The analysis of the Latvian cases exhibits even stranger results. This is perhaps best exemplified by the diverse result under the configuration already discussed under negative change in the overall analysis: INCENT*GOV(-)*VETO(+)*SIZE. In Latvia alone, this configuration results in a positive change under the Krasts government, in the preservation of the status quo in two other cases and in a negative change once (the above described amendments to the language legislation under the Šķēle government).

Latvia appears to be a specifically difficult case for compliance research, as no government in its modern history positioned itself or could be identified as having a positive stance to the very large Russian-speaking minorities, highly visible in all the major towns of the republic and possessing certain economic resources. The subsequent perceived vulnerability of the majority population plays for the isolation of minority parties in parliament, even when they get sizeable electoral support. With the general weakness of left-wing parties in Latvia, this leaves center-right coalitions to govern the country during all the period under study.

Only two of the positive changes in Latvia resulted from the same configuration:

incent*GOV(-)*VETO(+)*SIZE

First, in the absence of clear incentives and under a center-right coalition government of Vilis Krištopāns, in March 1999 the Law “On the Unrestricted Development and Rights to Cultural Autonomy of Latvia’s Nationalities and Ethnic Groups” was adopted. It contained anti-discrimination provisions designed specifically for Latvian residents who were not Latvian citizens and guarantees, for example, to all permanent residents equal rights to work and remuneration. Second, in Andris Bērziņš from the Latvia’s Way government (2000-2002), a major information campaign aimed at “advertising” naturalization was initiated, bringing the official engagement in the sphere of naturalization one level up, with no incentives from the EU being present nor the government being pro-minority. However, in four other cases this same configuration did not bring about a positive change: for example, this applies to the issue of language for the Krištopāns government – the same government that initiated a positive change in the area of non-discrimination. Hence, the consistency of this solution is only 0.33 (two out of six cases).

The best consistency (0.50) of all the configurations in the Latvian subset analysis is achieved by two solutions:

INCENT*GOV(=)*VETO(=)*SIZE + incent*GOV(-)*VETO(=)*SIZE

However, both solutions are based on single cases with a positive outcome (i.e. one out of two cases each), which results in a low coverage (0.14 for each term and 0.29 for the overall solution), and they are also only due to the ratification of the FCNM by Latvia in 2005, which can arguably be considered a weak case of improvement, as it not only requires implementation but transposition into domestic law to take effect. In sum, the QCA analysis of the Latvian cases gives no clear and meaningful account for the instances in which positive change took place.

4.4. Issue-specific analysisto contents list

As a last analysis of case subsets, the five different issue-areas are analyzed separately. Perhaps not surprisingly, citizenship, which applies only to the two Baltic countries, is the “odd one out” with regard to the solutions produced by the QCA analysis. What does surprise, however, is that it shows fully consistent solutions at all (which was not the case in either of the country-specific analyses for Estonia and Latvia), and that its general percentage of positive change is comparatively good, especially in the pre-accession phase (0.44), sharply dropping only after accession with not a single positive development reported.

Of the two solution terms, the first (INCENT*GOV(=)*VETO(+)*SIZE) corresponds to the contradictory configuration showing the highest consistency (0.67) in Estonia, the second (incent*GOV(-)*VETO(+)*SIZE) to the contradictory configuration covering two positive cases (but also four negative ones, hence a consistency of only 0.33 in the country-specific analysis) in Latvia. This means that treating citizenship separately from the other issues at least partly clarifies some of the contradictions in both countries (without completely resolving them, because the non-citizenship issues under these configurations still produce one positive outcome each), leading to fully consistent – albeit highly individualized – solutions, as both solutions are based on a single case.

The solutions of the other four issue areas largely correspond to the two solution terms of the overall analysis. Still, they show some interesting variation. Education of and in the minority language exhibits both the external incentives based and the domestic solution of the overall analysis. However, the configuration resembling the “domestic path” includes the absence of conditionality (instead of its irrelevance): incent*GOV(+)*VETO(+)*size. This is due to the fact that the configuration INCENT*GOV(+)*VETO(+)*size is simply missing. The same applies to Roma integration policies. In this issue area, also the constellation usually found with external incentives as a necessary component is shifted away from external influence, as one case of rule adoption in the absence of conditionality (found in Romania) renders external incentives irrelevant: GOV(+)*VETO(-)*SIZE. The adoption of Roma integration programs can therefore be explained without reference to external incentives.

The use of minority languages and non-discrimination result in only one solution term each, namely the domestic path. Language use is the only issue where consistent positive outcomes are exclusively reported in a configuration without external incentives (incent*GOV(+)*VETO(+)*size). In combination with the finding that the two instances of revocation also concern this issue and took place in the presence of external incentives, we can conclude that for the use of minority languages EU conditionality had no decisive positive impact. Although external incentives are rarely present in this issue area (three out of 23), this is also true for other issues in which positive outcomes in the presence of external incentives are reported (Roma: three out of 20; education: one out of 11).

Non-discrimination, by contrast, is the issue area in which external incentives are most often present (17 out of 25 cases) due to the fact that it is part of the acquis communautaire and therefore compliance is routinely demanded in the progress reports before accession starting from 2000 and also beyond accession. This puts into perspective that the solution for non-discrimination includes the presence of external incentives together with highly favorable domestic conditions (INCENT*GOV(+)*VETO(+)*size), as there are no cases with this domestic configuration combined with the absence of external incentives. Also, the fact that this issue area is the one with the highest number of cases involving conditionality, but shows no better percentage of positive outcomes (0.28) than the other issue areas questions the effectiveness of conditionality.

5. Conclusions to contents list

This paper conducted a multi-value QCA of the formal adoption and sustainability of minority protection rules in four new EU member states (Poland, Romania, Estonia, Latvia) over a twelve-year period including pre- and post-accession phases (1997-2008) and five minority protection related issue areas (non-discrimination, language use, education, citizenship, integration of Roma) based on four conditions (external incentives, government position, veto players, size of minorities), in order to investigate under which external and domestic conditions minority protection and non-discrimination measures are adopted, maintained or revoked in new member states before and after accession to the EU.

The overall analysis of all 93 cases shows two equifinal paths to positive change – one domestic, in which pro-minority oriented governments and veto players in conjunction with small minorities always lead to positive change irrespective of external incentives, and one including external incentives as a necessary component, which consistently produces a positive outcome in less favorable domestic conditions, i.e. with large minorities and in presence of nationalist veto players. The first solution supports the “domestic path hypothesis” (H#1), while the second largely corroborates the “external incentives path hypothesis” (H#2). However, although both solutions are generally consistent with the theoretical assumptions, they reveal that domestic conditions need to be more favorable than expected in order to be sufficient conditions for positive change. Specifically, external incentives turn out to be less effective than assumed in overcoming domestic opposition, as they depend on pro-minority oriented governments to be consistently successful(7). Also, the external incentives solution is empirically less important than the domestic path, as it covers fewer cases (see Ragin 2006, 299).

The separate analysis of pre- and post-accession cases reveals a marked decline in positive developments after accession, but no revocation of minority protection rules. After accession only the domestic explanation remains, despite external incentives still being present in one issue area, namely non-discrimination. The two instances of revocation do not lend support to the “revocation hypothesis” (H#3) and also put the effectiveness of conditionality slightly into question, as both happened prior to accession and despite strong conditionality, so that again domestic factors seem to be more important.

Cross-country analysis shows a clear distinction between Poland and Romania on the one hand, and Estonia and Latvia on the other. The latter produce no consistent QCA solution, i.e. show no constellation under which positive change always occurs. Hence, the overall solutions are entirely based on Polish and Romanian cases. Moreover, each of the countries accounts for one of the two equifinal solutions discovered in the overall analysis, which indicates that country-specific variation is the most important factor in producing different paths to positive change. Across issues, the absence of a superior performance of non-discrimination despite more frequently used conditionality is noteworthy, as is the role of language use as a problematic issue with both instances of revocation belonging to this issue area.

Theoretically, the results suggest that the theoretical model used to specify the conditions for rule adoption seems to capture some country cases – namely Poland and Romania – rather well (with the abovementioned qualifications), whereas the Estonian and Latvian cases as well as the (absence of) revocation are not sufficiently explained by the conditions considered in the model. From a QCA perspective, this is only an intermediate result: one next step could be the inclusion of further factors on the basis of theoretical assumptions and empirical case knowledge in order to resolve the contradictory configurations as far as possible. This will be a task for future research.


Referencesto contents list

Checkel, Jeffrey T. 2005. It’s the process stupid! Process tracing in the study of European and international politics. ARENA Working Paper No. 26/2005, http://www.arena.uio.no/publications/working-papers2005/papers/wp05_26.pdf (accessed December 7, 2009).

Cronqvist, Lasse. 2007. Tosmana: Tool for small-N analysis [version 1.3]. Marburg, http://www.tosmana.net (accessed December 7, 2009).

De Witte, Bruno. 2000. Politics versus law in the EU’s approach to ethnic minorities. EUI Working Paper No. RSC 2000/4. San Domenico: European University Institute.

European Commission. 1997. Agenda 2000 – Commission opinion on Poland’s application for membership of the European Union, DOC/97/16, Brussels, 15 July 1997.

European Commission. 1999a. 1999 regular report from the Commission on Estonia’s progress towards accession, Brussels, 28 September 1999.

European Commission. 1999b. 1999 regular report from the Commission on Romania’s progress towards accession. Brussels, 28 September 1999.

European Commission. 1999c. Romania 1999 accession partnership.

European Commission. 2000. 2000 regular report from the Commission on Poland’s progress towards accession. Brussels, 8 November 2000.

European Commission. 2001a. 2001 regular report on Poland’s progress towards accession. SEC(2001) 1752, Brussels, 13 November 2001.

European Commission. 2001b. 2001 regular report on Romania’s progress towards accession. SEC(2001) 1753, Brussels, 13 November 2001.

European Commission. 2002. 2002 regular report on Poland’s progress towards accession. SEC(2002) 1408, Brussels, 9 October 2002.

George, Alexander and Andrew Bennett. 2005. Case studies and theory development. Cambridge: MIT Press.

Kelley, Judith. 2004. Ethnic politics in Europe: The power of norms and incentives. Princeton: Princeton University Press.

Pan, Christoph and Beate Sibylle Pfeil. 2002. Minderheitenrechte in Europa. Handbuch der europäischen Volksgruppen Band 2. Wien: Braumüller.

Ragin, Charles C. 1987. The comparative method: Moving beyond qualitative and quantitative strategies. Berkeley: University of California Press.

Ragin, Charles C. 2000. Fuzzy-set social science. Chicago: University of Chicago Press.

Ragin, Charles C. 2006. Set relations in social research: Evaluating their consistency and coverage. Political Analysis 14: 291-310.

Ragin, Charles C. 2008. Redesigning social inquiry: Fuzzy sets and beyond. Chicago: University of Chicago Press.

Ragin, Charles C., Kriss A. Drass and Sean Davey. 2006. Fuzzy-Set/Qualitative Comparative Analysis 2.0. Tucson, Arizona: Department of Sociology, University of Arizona.

Rihoux, Benoit and Charles C. Ragin, eds. 2009. Configurational comparative methods: Qualitative Comparative Analysis (QCA) and related techniques. Thousand Oaks: SAGE.

Sasse, Gwendolyn. 2006. Gone with the wind? Minority rights in Central and Eastern Europe before and after EU enlargement. Paper presented at the workshop “Ethnic mobilization in the New Europe”, April 21-22, in Brussels, Belgium.

Schimmelfennig, Frank. 2000. International socialization in the New Europe: Rational action in an institutional environment. European Journal of International Relations 6 (1): 109-139.

Schimmelfennig, Frank. 2003. The EU, NATO and the integration of Europe: Rules and rhetoric. Cambridge: Cambridge University Press.

Schimmelfennig, Frank, Stefan Engert and Heiko Knobel. 2003. Costs, commitment and compliance: The impact of EU democratic conditionality on Latvia, Slovakia and Turkey. Journal of Common Market Studies 41 (3): 495-518.

Schimmelfennig, Frank, Stefan Engert and Heiko Knobel. 2006. International socialization in Europe: European organizations, political conditionality, and democratic change. Basingstoke: Palgrave Macmillan.

Schimmelfennig, Frank and Ulrich Sedelmeier. 2004. Governance by conditionality: EU rule transfer to the candidate countries of Central and Eastern Europe. Journal of European Public Policy 11 (4): 661-679.

Schimmelfennig, Frank and Ulrich Sedelmeier, eds. 2005. The Europeanization of Central and Eastern Europe. Ithaca/London: Cornell University Press.

Schimmelfennig, Frank and Ulrich Sedelmeier. 2007. Candidate countries and conditionality. In Europeanization: New research agendas, ed. Paolo Graziano and Maarten Vink, 88-101. Basingstoke: Palgrave Macmillan.

Schimmelfennig, Frank and Florian Trauner. 2009. Introduction: Post-accession compliance in the EU’s new member states. In Frank Schimmelfennig and Florian Trauner (eds): Post-accession compliance in the EU’s new member states, European Integration online Papers, Special Issue 2, 13, http://www.eiop.or.at/eiop/pdf/2009-SpecIssue-2_Introduction.pdf.

Schneider, Carsten Q. and Claudius Wagemann. 2007. Qualitative Comparative Analysis (QCA) und fuzzy sets: Ein Lehrbuch für Anwender und jene, die es werden wollen. Opladen: Barbara Budrich.

Schwellnus, Guido. 2005. The adoption of non-discrimination and minority protection rules in Romania, Hungary and Poland. In The Europeanization of Central and Eastern Europe, ed. Frank Schimmelfennig and Ulrich Sedelmeier, 51-70. Ithaca/NY: Cornell University Press.

Toggenburg, Gabriel. 2000. A rough orientation through a delicate relationship: The European Union’s endeavours for (its) minorities. European Integration online Papers 4 (16), http://eiop.or.at/eiop/texte/2000-016a.htm (accessed December 7, 2009).

Tsebelis, George. 2002. Veto players: How political institutions work. Princeton: Princeton University Press.

Vermeersch, Peter. 2007. Ethnic minority protection and anti-discrimination in Central Europe before and after EU accession: The case of Poland. Journal of Ethnic and Minority Issues in Europe 1/2003, http://www.ecmi.de/jemie/download/Focus1-2003_Vermeersch.pdf (accessed December 7, 2009).


Endnotesto contents list

(1) Formal rule adoption thus corresponds to the transposition stage of implementation as outlined in the introductory chapter (see Schimmelfennig and Trauner 2009 this issue). However, since among our issue areas only non-discrimination actually includes the transposition of EU law, we have opted for rule adoption as a more general term.

(2) To account for the revocation of minority protection rules, negative change was coded as “-1”, which was treated as a “0” for the QCA with positive change as the outcome. However, as we coded only two instances where a negative change was recorded, a separate formal QCA run for negative change was not feasible.

(3) Positive change in form of the signature or ratification of international conventions as opposed to changes in the domestic level of protection is represented in brackets in the coding tables (see Appendix I).

(4) Not every government lacking a parliamentary majority constitutes a case of the first chamber acting as veto player, as the government may be able to rely on ad hoc support by opposition parties. Only when there is a majority against the government regarding the respective issue we code it as a veto player.

(5) The fs/QCA software was used in addition to TOSMANA, because it allows the calculation of consistency and coverage. In order to accommodate the multi-value coding of the government as well as the veto player condition within the dichotomous crisp-set analysis of the fs/QCA program, we split both conditions into two separate conditions: Gov(+)/Gov(-) and Veto(+)/Veto(-), with the neutral value expressed by the combination of both being coded as “0”. The main disadvantage of this procedure is the creation of logically impossible remainders, since the combination of both conditions being present cannot occur. However, since we do not use logical remainders as simplifying assumptions, the results of both calculations are the same.

(6) As part of the standard QCA procedure we also checked for the absence of positive change, but do not report the results here, because conditions for no positive change are very diverse (up to seven distinct solution terms for some subsets).

(7) The fact that GOV(+) appears in both solution terms does not mean that a pro-minority oriented government is a necessary condition for positive change – there are instances of positive change in the absence of such a government, they are just not part of the solution because they are found in contradictory configurations (see Schneider and Wagemann 2007, 112-114).




European Integration online Papers | ISSN 1027-5193
© 1997-2014 ECSA Austria | All Rights Reserved. Webmaster Editorial Assistant