Assessing the impact of coercive enforcement instruments of the EU on physical integrity rights in the post-soviet region

Transcription

1 Assessing the impact of coercive enforcement instruments of the EU on physical integrity rights in the post-soviet region Paper presented at the Graduate ECPR Conference, Tartu, Estonia (June 10-13, 2016) Karina Shyrokykh Graduate School for East and Southeast European Studies, Ludwig-Maximilian University, Munich, Germany Address: Maria-Theresia-Str. 21, 81675, Munich, Germany Telephone: +49(0) N.B! Please do not circulate/cite without the author s permission 1

2 Assessing the impact of coercive enforcement instruments of the EU on physical integrity rights in the post-soviet region Karina Shyrokykh 1 Abstract. The European Union (EU) is an important economic and political partner and the largest aid provider for the former Soviet republics (FSRs). All forms of cooperation between the EU and third countries include human rights clauses and in case of severe violations the EU can apply measures including aid suspension, limiting economic cooperation, individual sanctions, etc. Previous research has demonstrated the very limited success rate of the actual general sanctions, and larger success rate of individual sanctions and threat of negative measures. At the same time, conditions under which these can be effective are largely unknown. This paper analyses the effect of EU s coercive measures on physical integrity rights in the FSRs distinguishing between types of sanctions. The purpose of the present research is to understand and explain to what extent and under what conditions coercive enforcement measures can improve human rights in the region. Utilizing time-series cross sectional analysis of 12 states over 20 years, this paper demonstrates that the presence of alternative cooperation partners is a decisive condition for success of the EU s coercive enforcement measures. Additionally, states that are heavily dependent on foreign aid also improve practices as a result of the EU s coercive measures; whereas, the presence of competing interests might limit leverage of the EU and make application of coercive enforcement instrument less effective. Keywords: sanctions, coercive enforcement, European Foreign and Security Policy, European Union, post-soviet states, energy dependence Introduction The Treaty of the European Union (TEU) introduces restrictive measures as a foreign policy instrument that can be used to foster the goals of the Common Foreign and Security Policy (CFSP) one of which is human rights protection in third states. The general approach of the EU to its foreign policy is based on cooperative and integrative approach, rather than 1 Karina Shyrokykh is a doctoral candidate in Political Science at Ludwig-Maximilian University, Munich, Germany. 2

3 coercive: the EU declares its preference for positive measures over restrictive or coercive measures (European Council, 2004, p.2; Library of the European Parliament, 2013; de Vries and Hazelzet, 2005, p. 151). However, sometimes the EU applies coercive enforcement measures vis-à-vis third countries to maintain or restore international peace and security (Library of the European Parliament, 2013, p.1). Coercive enforcement measures are defined in this paper as measures that aime at coercing states to change their behaviour via imposing (or threatening) costs. Coercive enforcement measures include not only general sanctions but also individual (asset freezing, visa bans, etc.), aid and preferential trade agreements withdrawal, as well as the threat of these. In keeping with conventional usage, the sender imposes coercive measures on the target. The application of sanctions lies in the realm of foreign policy and serves objectives of the CFSP. Such measures are applied in order to further democracy, the rule of law, the universality and indivisibility of human rights and fundamental freedoms, respect for human dignity, the principles of equality and solidarity, and the respect for the principles of the United Nations Charter and international law (Art.21.2 of TEU). The EU can impose sanctions in accordance with explicit terms of agreements. Since the mid-1990s human rights clause is included in all trade and assistance agreements that allows the EU to suspend cooperation with a third country unilaterally when human rights clause is violated. The number of incidents when coercive enforcement decisions were made as a respond to human rights violations by the CFSP has increased significantly in comparison to the early 1990s when the CFSP has been launched. For example, in 2013, 17 regimes were under sanctions, while, in 1992, when the CFSP was just created, only two regimes were in this situation (Library of the European Parliament, 2013). There can be different reasons for the EU to apply these measures, such as poor democratic standards like in Belarus, violation of territorial integrity of another state, like in the case of sanctions against Russia followed by annexation of Crimea and involvement into the military conflict in the East of Ukraine. However, the present paper is focused on coercive enforcement measures introduced by a sender to tackle human rights situation in the former Soviet republics (FSRs). The present paper contributes to European and comparative studies literature by proposes a large-n quantitative study of the effects of different sanctions type and conditions under which coercive enforcement can be effective in the FSRs. Most of existing literature builds on single case studies which provide valuable insights to the specific cases on sanctions applied vis-à-vis post-soviet states (European Parliament, 2012; Giumelli and Ivan, 3

4 2013; Ditrych, 2013; Gros and Mustilli, 2015; Korosteleva, 2012; Hoch, 2011); however, there is still lack of a systematic assessment of the instrument s effectiveness in the region. Therefore, this paper aims at answering the following question: what is the effect of coercive enforcement instruments of the EU and under what conditions it is more likely to impact the human rights situation in the FSRs? The present paper contributes to the previous studies by drawing on longitudinal analysis of the sanctions and threats of their imposition introduced in order to improve human rights. The particular attention in this study is paid to the conditions under which such measures can trigger positive change. Building on rationalist theories, this paper addresses the existing debate on the issue of effectiveness of coercive enforcement measures. 2 Utilizing a time-series cross-section (TSCS) analysis of data collected over a period of 20 years ( ), the present paper demonstrates that coercive enforcement instrument can have a positive impact on a target under the condition of a) strong asymmetry in relations in favour of the EU; b) absence of an alternative regional partner, and c) absence of the EU competing interests in the target. Offering a cross-country analysis of the new data, the paper contributes to the existing literature by demonstrating the overall effect of coercive enforcement and conditions under which this instrument is more likely to have an effect. The contribution speaks to the previous research and states the importance of international and domestic factors in the study of external governance in the region. The rest of the paper proceeds as follows. The following section describes measures available at the EU level and discusses existing literature on coercive enforcement. The third section develops a theoretic framework and proposes hypotheses. The fourth section describes the research design and data. The fifth section describes the results, and the final section summarizes the paper. Throughout the paper, I focus on how coercive enforcement instruments of the EU may shape states behaviour vis-à-vis the most basic human rights: the rights to security of a person or as these are often called physical integrity rights. 3 2 The concept of effectiveness throughout the paper refers to an ability of the EU to improve the human rights situation in the recipient states. The objective of coercive enforcement measures is to promote human rights; the EU does not refer to any threshold that defines it (Source: Council, The European Consensus on Development). 3 Physical integrity rights are considered to be the most basic elements of civil and political rights of a citizen. These rights include freedom from extrajudicial killing, torture or other cruel and inhumane or degrading treatment, forced disappearance, and freedom from political imprisonment. On a hierarchy of human rights, see Meron, On a Hierarchy of International Human Rights. 4

5 Coercive enforcement instrument: conceptualization and the area of contribution This section first introduces the existing debate, describes analytical concepts employed in the paper and then summarizes the EU policy towards coercive enforcement. Existing literature does not agree on the overall effect of the instrument. Hence, general sanctions have received considerable attention of scholars while other types of sanctions have received attention in more recent studies. Some show that economic sanctions could be a detrimental foreign policy instrument, capable of destabilizing the social and economic situation in a targeted country and affect the general population while allowing the elites to avoid the cost of the sanctions (Gibbons, 1999). The suggests that general sanctions, such as trade embargoes and general financial sanctions, are more detrimental to human rights than individual sanctions (also called targeted sanctions) (Peksen, 2009). Some suggests that economic coercion is counterproductive policy tool, even when sanctions are imposed with the specific goal of promoting human rights; they also suggest that multilateral sanctions have a larger negative impact on human rights than unilateral sanctions and the longer sanctions are applied, the larger human rights cost they will inflict on societies (Ibid). Others convincingly demonstrate that sanctions sometimes can trigger positive change (Kirshner, 1997; Blanton, 1999). Despite the lack of agreement on the extent to which such measures can be effective, literature agrees that there is little analysis of condition determining success of coercive enforcement. Therefore, building on previous work suggesting some factors that might determine the extent of success of European foreign policy instruments, the present paper aims at contributing to the debate by developing the theoretical expectations and testing them empirically on longitudinal data. Coercive enforcement is an instrument that is associated with the allocation of economic cost by the sender with a variety of foreign policy goals. The concept of coercion relates to the concept of pressure, but coercion in the sense of pressure can be diplomatic, normative, social, etc. The concept of coercion per se does not reflect the presence of leverage of the sender. To stress the unequal relations between the sender and the target, the concept of coercive endorsement is introduced. The enforcement part of the term regards coercion as an activity by a powerful agent who creates and then utilizes a significant disparity in power over another in order to constrain or alter the latter s possibilities for action (Anderson, 2010, p.1). This power asymmetry may be used to put pressure on the target and may disable the target or disrupt its various possibilities for action in a systematic 5

6 manner. Such systematic disruption can be achieved by punishment or threat of punishment that alter broad patterns of activity, and not just specific actions. Therefore, the concept of coercive enforcement reflects not only pressure of the sender but also intrinsically builds on leverage and asymmetry of relations. Leverage in its turn is defined as vulnerability of governments to external pressure (Levitsky and Way 2010). Sanctions are an available instrument of foreign policy since 1957 and were the result of developed coordination between states (Kreutz, 2005, pp. 7-8). However, in practice, application of sanctions is often problematic since it involves costs for both the sender and the target. An additional problem with sanctions is associated with the fact that decisionmaking process is very multifaceted. For a long time decisions over sanctions had exclusively inter-governmental characteristics, with time, the decision-making process has developed more supra-national features, however, are still inflicted by the Council. In the post-soviet region, Belarusian for a long time has been under extensive sanctions, such as arms embargo, a ban on export of equipment that can be used for internal repression, a ban on provision of certain services, and restrictions on admission of persons responsible for serious violations of human rights (Council, 2012). 4 Additionally, the former Ukrainian political leadership are under the individual sanctions. 5 Likewise, Russia and Central Asian republics of Tajikistan, Uzbekistan and Turkmenistan have experienced some coercive enforcement measures that have been applied due to gross human rights violations (Table 1). In a situation of significant deterioration of the human rights situation, the EU can deliberately withdraw, or threaten to withdraw, customary trade or financial relations. These measures are usually imposed in reaction to undesirable acts that are geared towards exercising pressure normally with the ultimate aim of altering the political behaviour of the targeted party (Portela, 2014, p. 3). There are a few types of coercive enforcement measures that the EU can apply to change a certain course of action. These include individual, general and sectorial sanction, aid suspension, and threats of their application. General sanctions are applied to influence citizens of a targeted state, who in turn, will pressure their government to yield to the demands of the initiating states. Individual sanctions are applied more often and are usually seen as more effective than general sanctions (Peksen, 2009). Individual, smart or targeted sanctions aimed at putting pressure on the leaders or specific elites; and are aimed at affecting specific individuals, rather than the entire 4 Since 2012 the sanctions were prolonged every six months. 5 Valid until , accessed

7 population. Nowadays, the EU avoids applying general measures such as trade embargoes, sanctions rather take a form of individual or sectorial sanctions, such as visa bans, financial assets freezing, arms embargoes (Kreutz, 2005, p.41). They differ from general sanctions as they are aimed at non-state actors (individuals, groups, companies, etc.) and regard specific economic sectors or specific products (Giumelli, 2013). Such measures are usually seen as a part of peace-making initiatives in conflict and attempts to protect civilians from government repression (Kreutz, 2005, p.42). In addition, arms embargoes and sectoral embargoes aimed at depriving repressive regimes of access to weapons that can be used for suppression of civilians is often applied and regularly are among very first measures applied. The thinking behind these embargoes is to prevent escalation of violence by restricting trade of goods that can be used for domestic repression. Lastly, the EU practices withdrawal of assistance, however, in practice it prefers to redirect aid from government to NGOs (Ditrych, 2013; Smith, 2014, p. 53). In the post-soviet region, suspension of assistance has been applied against Belarus, Russia and Uzbekistan. Theory and hypotheses: Coercive enforcement and possible constrains The underlining logic behind coercive enforcement instrument is based on target s vulnerability and asymmetry of relationship, the concept of leverage reflects the vulnerability and one-sided dependency of the target. External actors have much less chance to influence behaviour of states over which they do not have any leverage. At the same time, the degree of leverage might be determined by a few conditions. This section aims at developing theoretical expectations on the favourable conditions building on rationalist theories within IR and employing the concept of leverage (Levitsky and Way, 2005 and 2010). Intensity of sanctions and the degree of dependency Some literature suggests that sanctions aimed at weakening of the repressive regime might deny them necessary economic, military, and other resources that are crucial for political elites to maintain stability (Kirshner, 1997). Thus, once the target regimes are denied the access to external military and other resources, sanctions should harm the coercive capacity of the repressive regime and subsequently diminish governmental repressions (Blanton 1999). Furthermore, because state resources are the major tool to assure loyalty of the military, scare resources can cause a decrease in support among domestic groups to the 7

8 repressive regimes (Bueno de Mesquita, et al., 2003; Peksen, 2009). As a result, ruling elites may lose power that opens up opportunities for the opposition to take a lead. Therefore, sanctions may induce power shifts in favour of anti-government groups. As the antigovernment groups may move towards more liberal and pro-human rights regime that eventually, any result in improvement of human rights practices. In sum, sanctions are aimed at improving human rights conditions by undermining target government s coercive power and political support from elites, as well as public support. Coercive enforcement instruments might not cause an effect by themselves, but make conditions for a change, provide a window of opportunity for other actors such as civil society, opposition groups, media, and external actors to impact the domestic balance. They can change domestic equilibrium by imposing extensive costs for human rights violations; if costs overweight the benefits of human rights suppression, the domestic elite might be more likely to change its actions. Moreover, such measures can serve as means of communication, demonstrating the sender s commitment to human rights. Therefore, sanctions might rigger positive change not due to the direct cost they inflict on the target, but rather since they also send a signal of disapproval. Therefore, assuming that states behave rationally, states will change their practices if faced with coercive enforcement measures. Since the instrument is aimed at raising the cost of human rights abuse, the cost-benefit calculations might result in change of behaviour. In other words, the higher the cost that the target faces, the more likely it is to adjust its practices. Therefore, I expect that the effect of coercive enforcement increases with its intensity (Hypothesis 1a). The first hypothesis aims at accessing the baseline model against which the rest of the hypotheses will be tested. At the same time, leverage of the sender might strongly depend on the extent of dependency of the target. The degree to which the target relies on the sender s financial assistance and trade relations might constitute a degree of leverage. The following hypothesis further develops this argument. As human rights clause is an integral part of economic and assistance agreements between the EU and third countries, the EU can unilaterally freeze cooperation as the human rights situation deteriorates. Since the early 1990s cooperation with the EU has been linked to the issues of human rights, democracy and good governance (Shyrokykh, forthcoming). The stated policy objectives were the promotion of civil, political liberties, democratic government and an accountable and effective administration (Ibid). Considering the concept of leverage, states that are more dependent on export to the EU and aid from the EU might suffer more economic losses if they are subjected to coercive measures; with increased 8

9 dependency, targets become more vulnerable and therefore might be more likely to adjust their practices. Ina situation when a target relies on economic relations with the EU, the target will be more likely to change its behaviour as a result of coercive enforcement. Dependent targets are more vulnerable and one can expect there coercive enforcement to have more effect. In other words, the larger the asymmetry of relations (in favour of the EU) is, the more likely coercive enforcement is to have a positive effect on the human rights situation (hypothesis 1b). Competing interests and presence of alternatives Hypotheses discussed in this section account for factors that might limit this dependency and flatten the situation of asymmetry in relations between the sender and the target. For most of the FSRs the EU is the largest trading partner, the largest importer, and the largest aid supplier, which implies that it can exercise certain leverage over the states of the region. 6 However, there are major structural limitations to the EU leverage such as presence of alternative regional powers and EU s own interests in the region. Earlier works suggest that interests of the sender in a relationship with the target might undermine the leverage of the former (Levitsky and Way, 2010). Under conditions when a sender has economic or security interest, measures like sanctions might damage not only the target but also the sender. For example, The EU has been often criticised for minor sanctions imposed on Russia for gross human rights violations during both Chechen wars. Both wars were accompanied with allegations of war crimes; however, EU sanctions were rather limited. Likewise, existing literature suggests that the inconsistences between the EU s statements on democracy promotion and its actions on the ground in oil- and gas-rich states, which literature explains by the presence of competing interests (see e.g., Hyde-Price, 2006; Youngs, 2004). For instance, Schimmelfennig, points out that that despite the pervasive political and legal rhetoric of democracy and human rights promotion, [the EU s] actual policy seems to match rhetoric only when consistency is cheap ; otherwise, it is driven by a host of other geopolitical, economic or security interests (2009, p. 15). Similarly, much research suggests that the EU in pursuing its policy objectives abroad often disregards its normative goals when overriding security or commercial interests are at stake (Warkotsch, 2008; Knodt and Jünemann, 2008). 6 According to the European Commission s reports, since the mid-2000s, most of the post-soviet states have the EU as a major trading partner ( 9

10 A competing interest is anything that interferes with the declared objective of the EU. Competing interests can arise from the presence of strong secondary interests in the target state such as related to economy, security or other first-order issues (Mearsheimer, 2001). A competing interest exists when foreign policy decisions concerning a declared objective (such as human rights, democracy and good governance promotion) may be influenced by other undeclared and officially not stated interest. When having competitive interests in target states, sanctions may affect not only the target but also the sender, and the larger the cost for the sender is, the less likely the sender will be stick to the originally declared human rights promotion policy. Furthermore, to be effective coercive enforcement measures have to be credible that is to say believable. And the presence of competing interests could limit credibility of such measures. Hence, with the presence of competing interests undermines effectiveness of coercive enforcement (hypothesis 2a). Building on the rational cost-benefit calculation of both the sender and the target, I expect that exceeding costs for the sender will limit credibility of coercive measures and, therefore, their effectiveness. Furthermore, the extent of leverage of the EU might depend on the presence of other viable alternative providers of benefits in the region. Previous studies demonstrate that interdependence between Ukraine and Russia shapes the context which determines the policy development in Ukraine, showing how this interdependence limits effectiveness of external governance (Dimitrova and Dragneva, 2009, p. 853). Similarly, Levitsky and Way suggest that presence of alternative to the West normative and economic powers might undermine the regional importance of democracy promoting actors (2010, p. 41). Literature points on Russia as a major illiberal regional power, whose economic, military and other interferences might blunt the impact of liberal external actors; studies suggest that Russia presents a viable alternative of the EU in the region. A powerful illiberal allay can absorb some of the economic costs imposed by liberal actors. For instance, it can provide an alternative market for exports, provide financial and technical assistance, and cover the costs of sanctions (Haufbauer et al., 1990: 12; Tolstrup, 2015). For states in the post-soviet area Russia is the major alternative to the West. By providing alternative normative and economic axes of integration, cooperation with Russia might imply less cost than developing cooperation with the EU. Since the presence of alternatives diminishes leverage of the EU, their presence might undermine effectiveness of coercive enforcement. Effectiveness of coercive enforcement instrument might decrease when an alternative regional power gains more leverage over the 10

11 target. Coercive enforcement is more likely to be followed by improvement of domestic practices when the target does not have alternatives to the West. The more dependent the target is, the more likely it is to be impacted by coercive enforcement. Therefore, the presence of alternatives to the EU limits effectiveness of coercive enforcement measures (hypothesis 2b). Research design: Data, method, variables, and models The data collected for this study is based on annual observations of human rights practices of the post-soviet states from 1992 to The number of units (states) analysed is 12 that are Armenia, Azerbaijan, Belarus, Georgia, Kazakhstan, Kyrgyzstan, Moldova, Russia, Tajikistan, Turkmenistan, Ukraine and Uzbekistan. 7 The data have a time-series cross-section (TSCS) structure, consisting of repeated observations over 20 time periods which satisfy methodological requirements. 8 The results presented in this paper are estimated using fixedeffects linear dynamic model with Driscoll-Kraay standard errors with lagged dependent variable. Driscoll-Kraay standard errors are robust to cross-sectional and temporal dependence, as well as heteroskedasticity and autocorrelation (Hoechle, 2007). This covariance estimator works for both, balanced and unbalanced panels and it is capable to handle missing values. Fixed-effects estimates are robust to unobserved heterogeneity between groups; therefore this estimate is a good fix for unobserved time-invariant characteristics of states (Back, 2001; Plümper et al., 2005). Cultural, national identity, culture and political legacies are time-invariant. Since I am not focussed on detecting their effect, I control for such variables by including the fixed effects. To exclude possibility of bias associated with simultaneity and endogeniety, the results have been estimated on a dynamic model by lagging the independent and control variables by at least one year. Dynamic TSCS model also minimizes possible simultaneity or reverse causation problem (Fortin-Rittberger, 2014; Arellano and Bover, 1991). For the analysis, both the episodes of application sanctions and non-application episodes are included to overcome a selection bias detected in previous studies (Peksen, 2009; Morgan and Miers, 1999). Moreover, literature suggests that not only instances of 7 The Baltic States are excluded from the analysis since their transition processes were significantly impacted by the membership conditionality; whereas the rest of the post-soviet states were not exposed to this conditionality. 8 The minimum number of units should be 10 and the minimum number of time periods should be 10 (Source: Beck, Time-Series Cross-Section Data ). 11

12 sanctions and their types can provide useful insides, but also data on threats of sanctions (Peksen, 2009). Therefore, the data on coercive enforcement cover period of between 1992 and 2011 (the time-range is determined by the data availability on the dependent variable) and includes 51 instance of coercive enforcement episodes (various sanctions and threat) and 189 episodes of non-application. For data on coercive enforcement application, I use the Threat and Implementation of Economic Sanctions (TIES) dataset (Morgan et al., 2006), Hufbauer, Schott and Elliott (HSE) (Hufbauer et al., 2007) sanctions datasets. Table 1 shortly summarizes the history of coercive enforcement instruments applied vis-à-vis countries in this region. Table 1. Negative measures imposed by the EU against post-soviet states for violations of human rights Year Country Coercive enforcement measures taken sanctions initiated Belarus Aid programmes halted and ratification of the PCA postponed Belarus Restrictive measures against certain officials Belarus Travel restrictions, ban on high-level contact (prolonged: 17/10/2004) Belarus Travel restrictions and freezing of assets (prolonged: 10/04/2006; 06/04/2009; 15/12/2009; ) 2007-till now Belarus GSP suspended (22/06/2007) Belarus Persons and entities admission restrictions, asset freezes and travel bans (introduced: 20/05/2006; amended: 09/07/2008) 2008 Measures against certain officials undone; release of frozen funds, arms embargo remained (Introduced: 09/07/2008) Belarus Travel restrictions against certain official and entities, arms embargo, equipment for repressions embargo, assistance withdrawal (introduced: 25/10/2010; prolonged: 31/12/2011; 20/06/2011; 10/10/2011; 23/01/2012; 23/04/2012) Belarus Asset freezes and travel bans (persons and entities), arms embargo, embargo on equipment for internal repression; ban on provision of certain services (introduced: 17/10/2012; prolonged: 30/10/2013; 02/08/2014; 30/10/2015) 1995 Russia Cooperation agreement postponed Russia Aid suspension, postponement of agreement, sectoral embargoes Russia Arms embargo Tajikistan TACIS aid suspended for human rights violations Tajikistan Postponement of the ratification of the PCA Uzbekistan Arms embargo, aid withdrawal, restrictions on admission of certain persons (2007 visa restrictions suspended, 2009 arms embargo and other restrictions lifted) Turkmenistan Postponement of the ratification of the PCA, withdrawal of aid Turkmenistan Postponement of ratification of the PCA 2014-present Ukraine Individual sanctions (introduced: 06/03/2014) Source: Smith, K. (2014), Appendix 1; European Council, Council Decision (CFSP) 2015/1957 of 29 October 2015 amending Decision 2012/642/CFSP concerning restrictive measures against Belarus; European Commission, Restrictive measures (sanctions) in force, updated on 15/01/ For the summary statistics of the main variables used and the transition probability table, see Appendix 1. 12

13 The data is complemented from the websites of the Commissions accounts in Restrictive measures (sanctions) in force. 10 In the HSE dataset, the variable Issues is examined to identify reasons for sanction (and threat) imposition. In particular, for this research sanctions introduced with the purpose of human rights improvement are considered. 11 Furthermore, measures taken vis-à-vis individuals or groups irrelevant to the government are not considered. For instance, sanctions introduced against elites of de facto state of Transnistria are not coded as measures against Moldova, since the state does not have full control over such territories. The same logic applies to the breakaway regions of Abkhazia and South Ossetia in Georgia. Variables The dependent variable is human rights practices; it is measured by the Cingranelli and Richards s (CIRI) index (2014). It is a nine-level index ranging from 0 (no respect) to 8 (full respect). Data are available from 1992 to 2011 for all countries of interest. The index is constructed based on the Amnesty International and State Department s reports and measures the number of violations occurred in each state. The CIRI index includes four human rights indicators: freedom from torture, freedom from extrajudicial killings, freedom from forced disappearance and political imprisonment. Zero violations would dictate a score of 2 (full respect), 1 to 49 violations would dictate 1 (moderate respect), and 50 and more violations would dictate a score of 0 (no respect). The analysis accounts for both the general effect of coercive measures and effects of disaggregated elements. Therefore, the paper discusses two sets of independent variables. The first independent variable is coded to reflect the intensity of sanctions. The intensity variable is an additive index accounting for the number of sanctions type in force each year. The variable has five levels. It accounts for the following elements: threat of sanctions, individual sanctions, arms embargoes, postponement of agreements, and aid withdrawals. The variable refers to the total number of instruments applied vis-à-vis each state. The purpose of using an additive variable is to account for the severity of coercive enforcement. For instance, if a country received a threat of sanctions and later the same year individual sanctions have been introduced, the level of intensity is 2. If a country received neither threats nor actual sanctions, the level of intensity of sanctions is European Commission, Restrictive measures (sanctions) in force, 2016 ( 11 Other sanctions are introduced as one of the control variables. 13

14 Next, I create five binary variables accounting for each specific elements of coercive enforcement and accounting for whether it has or has not been in force each year. The data on every type of coercive enforcement instrument (threats, individual sanctions, arms embargos, aid suspensions, postponement of agreements) is coded as a binary variable, where 0 denotes no measures this year, and 1 measures in place. These two sets of variables are introduced to account for the effect of disaggregated elements and the general effect of coercive instruments. The model also includes a number of controlled variables to rule out alternative explanations. Although being the largest, the EU is not the only external actor that can exercise coercive enforcement. To separate the effects of coercive enforcement measures of the EU from other external actors, I account for other coercive enforcement measures. The variable is coded as a binary variable and accounts for whether or not other coercive measures were in force. In addition, the model controls for the effects of developmental differences across countries accounting for gross domestic product (log of GDP per capita) measured in Million USD in current prices. The data are obtained from the World Bank (2016). The presence of international and civil wars might also impact the level of respect for human rights. The Correlates of War dataset is used to control for these (Sarkees and Wayman, 2011). These data are complemented by using World s Major Episodes of Political Violence dataset (Marshall and Cole, 2014). Moreover, since CIRI is an index rather than a ratio, a variable accounting for the size of population is introduced (log of population) as a control variable. A democracy variable is included to control for the impact of the regime type on human rights abuse. The democracy variable is taken from Polity IV dataset (Marshall and Jaggers, 2016) and accounts for a level of democracy of domestic institutions. Each country s democracy score ranges from -10 to 10, where 10 represents the highest level of democracy. Knowing that sanctions are the result of a deteriorating human rights situation, in order to avoid the problem of reversed causality, I have to account for the past practices of states using a lag of the dependent variable. To account for domestic dynamics taking place unrelated to external interference, the model also accounts for instances of regime change. The regime change is coded based on conceptualization of Stykow (2014). The variable is coded as a binary variable where 0 entails no ruling elite change in a given country in a given year and 1 entails ruling elite group has been changed. To test the conditional hypotheses (Hypotheses 1b, 2a and 2b), I introduce variables used for interaction effects. The concept of dependency of the target on the EU is 14

15 operationalized via the size of financial assistance (in relation to GDP). The larger the size of the assistance in comparison to GDP is, the more the target is dependent on the sender. 12 For this purpose, I divide the value of financial assistance by the value of GDP. These data are collected from the official Eurostat webpage and are complemented by the data from World Bank (2004) and EU trade statistics (2015). To account for the presence of competing interests, I construct a proxy variable that accounts of energy cooperation. Energy security is the major competing interest of the EU in the post-soviet region (e.g., Baran, 2007; Schmidt-Felzmann 2008 and 2011; Tsertsvadze, and Axyonova, 2013). It is an ordinal variable that measures a type of energy cooperation. It is coded as 0 if there is no cooperation in the energy sector, 1 if a state is a transit state, 2 if a state exports energy resources to the EU, and 3 if there is a special energy cooperation designed to secure export of energy resources in the future, construction of new pipelines, permanent working bodies facilitating this cooperation. 13 Alternative regional power might cover the costs imposed by the EU and, therefore, limit leverage of the EU. Hence, the presence of an alternative partner might negatively affect the effectiveness of coercive enforcement instruments. The presence of alternative partner is approximated via extent of export to Russia. If the extent of export is larger than average in the region, then Russia is a viable alternative. Variables used for the analysis are shortly summarized in Table 2 below. Table 2. Summary statistics of the main variables used in the models Variable Obs. Mean Std. Dev. Min Max Physical integrity Index Intensity of measures Aid suspension Individual measures Sectoral embargo Postponement of econ.coop Threat of measures International war Civil war Democracy Regime change Other sanctions Population GDP per capita Source: Smith, K. (2014), Appendix 1; European Council, Council Decision (CFSP) 2015/1957 of 29 October 2015 amending Decision 2012/642/CFSP concerning restrictive measures against Belarus; European Commission, Restrictive measures (sanctions) in force, updated on 15/01/ In the Appendix, additional proxies for dependency are also discussed. 13 The data is coded based on Annual Reports accounting for cooperation with the FSRs such as the TACIS, PCA, individual Action Plans, bilateral agreements of energy cooperation retrieved from the EEAS available at and European Commission, available at accessed May 15,

16 Model specification In order to choose the model specification, I run series of statistical tests checking for the presence of autocorrelation, heterogeneity of units, spatial correlation, time and individual effects, heteroskedastcity, and unit roots. 14 Issues detected (heteroskedastcity, autocorrelation and special correlation) are considered when choosing model (Hoechle, 2007). The analysis is conducted with fixed-effects dynamic model with Driscrol-Kraay standard errors and lagged dependent variable (LDV). Dynamic models allow studying how past values of the dependent and independent variables affect current outcome. Dynamic panel model also helps to rule out endogeneity. In other words, the model acknowledges that coercive enforcement instruments are the result of the human rights situation. Therefore, a lag of minimum two years is used (Yi,t-2). This technique separates the effects of last practices that triggered sanctions from the effect of coercive enforcement on practices. Both, the original model and models used for robustness check allow me to accounting for the past practices. Moreover, the structure of lags fits the empirical knowledge that EU is usually responsive to human rights violations. Results obtained by the original model with LDV are checked against the Arellano and Bond (1991) dynamic model based on a generalised method of moments (GMM). Both models are robust to heterogeneity, crosssectional dependence, and serial correlation, and, therefore, appropriate (results are reported in Appendix). Therefore, the original dynamic TSCS model with fixed effects and LDV has the following form: y i,t = α + y i,t 2 + x i,t 1 β + v i + e i,t In the equation, y i,t is the value of the dependent variable for each cross-sectional unit (state) 1,2,3, N, denoted by the subscript i and t refers to time period 1,2,3, T. x i,t 1 is a repressor (independent variable) observed for this unit lagged at one year, e i,t represents the within-country error, and the term v i is a dummy variable for each of the units; regressor y i,t 2 introduces past values of the outcome variable as an explanatory variable. 14 Serial correlation was detected using Wooldrige test (p =.0136). To detect the structure of serial correlation, Arellano - Bond test was used (first order: p =.0025; second order: p =.3852). Wald test was used to detect heteroskedasticity (p =.0004). Pesaran's test was used for checking for cross sectional dependence (p =.0008). 16

17 Results and discussion This section summarizes the results of the analysis and discusses their implications. It begins with the description of the results of the baseline hypothesis that suggests that effects of coercive enforcement increase with its size. Analysis of the data reveals that the first hypothesis holds. The more intense measures are, the more likely they are to have a positive impact (Model 1 in Table 3). Figure 1 presents the predictive margins plot with 95 per cent confidence interval based on Model 1 (in Table 3). These results also suggest that coercive enforcement effect increases with time (lag of three years: , p =.003). Table 3. The effects of coercive enforcement on the human rights situation, Model 1. Model 2. Hypothesis 1a Hypothesis 1a: disaggregated Past human rights practices ** ** (0.07) (0.07) Intensity of measures * (0.11) Aid suspension (0.41) Individual sanctions (0.40) Arms embargo (0.53) Postponement of agreement ** (0.18) Threat (0.43) Log population * *** (1.66) (1.41) Log GDP per capita *** *** (0.26) (0.25) International war ** ** (0.51) (0.55) Civil war ** ** (0.29) (0.26) Democracy * ** (0.03) (0.03) Other sanctions (0.15) (0.14) Regime change (0.24) (0.22) Constant ** *** (27.66) (23.04) Observations R All independent variables are lagged by two years. Level of statistical significance: * p < 0.10, ** p < 0.05, *** p <

18 Human rights practices When disaggregating coercive enforcement measures and paying attention to what particular effect different types of measures have, positive effects are detected for arms embargo (significant at p =.198) and postponement of agreements. Thus, the results demonstrate that effects of disintegrated types of measures are also more likely to be detected with time; no statistically significant effects were detected at smaller lags. Sectoral embargoes (arms embargoes) are likely to have a positive effect suggesting that depriving suppressive regimes of means of suppression can impact the practices. However, the observed effect can be explained with an alternative explanation. Usually, arms embargoes are introduced among the most rapid measures to stop massive violence. Therefore, it might be that this particular type of measure serves as a mean of communication to the suppressive government. Introduction of these, therefore, might symbolically communicate disapproval Extent of coercive enforcement measures Figure 1. Effects of coercive enforcement measures on the human rights situation. The second theoretical expectation assumes that dependency on the EU might increase the effectiveness of coercive measures hold (Table 4 and Figure 2). Figure 2 suggests that effects of measures differed in states that are mode dependent on the EU aid. The figure compares states that are more and less dependent on EU aid. The results suggest that the more dependent state is, the more vulnerable it becomes and in such states coercive enforcement instrument might trigger positive change. 18

19 Table 4. The effects of coercive enforcement on the human rights situation, Model 3. Model 4. Model 5. Hypothesis 1b Hypothesis 2a Hypothesis 2b Past human rights practices *** *** *** (0.08) (0.06) (0.07) Intensity * Dependency ** (0.04) Intensity * Competing interests ** (0.09) Intensity * Alternative partners (0.59) Intensity of measures *** * (0.09) (0.12) (0.53) Dependency * (0.02) Log population ** ** * (1.52) (2.02) (1.60) Log GDP per capita ** *** ** (0.28) (0.30) (0.28) International war ** * ** (0.52) (0.54) (0.51) Civil war *** ** (0.38) (0.25) (0.25) Democracy ** ** ** (0.03) (0.03) (0.03) Regime change (0.27) (0.24) (0.24) Other sanctions (0.15) (0.14) (0.14) Competing interests (0.54) Alternative partners (0.21) Constant *** ** * (25.14) (32.84) (26.94) Observations R All independent variables are lagged by two years. Level of statistical significance: * p < 0.10, ** p < 0.05, *** p <

20 Less assistance More assistance Past human rights practices Intensity of measures Log GDP per capita Intrawar Interwar Democracy Regime change Other sanctions Figure 2. Coefficient plot for further robustness check (Model 3). Hypothesis 2a draws on the relationship between competing interests of the EU suggesting that these might undermine the positive effect of international pressure. Results of Model 4 suggests that when there is no energy cooperation, coercive enforcement might impact positively on human rights and the size of the effect is larger than in Model 1 (coefficient for intensity without interaction: *** ). Figure 3 below compares the coefficient of intensity of sanctions when there are competing interests and where there are no competing interests. Findings suggest that energy cooperation is negatively associated with the effect of coercive measures on human rights. EU being concerned about its energy security loses part of its leverage over the target. The presence of competing interests might undermine the EU credibility in the eyes of the target and, therefore, undermine human rights promotion effectiveness. This finding confirms previous qualitative studies that observe some deteriorating effects of energy cooperation (Baran, 2007). The quantitative analysis of empirical data support previous qualitative research and yields policy implications. The results are robust to alternative model specifications: controlling for the level of democracy and introduction of interaction term of energy cooperation and the level of democracy. This suggests that it is not the regime or the interplay between the regime type and energy cooperation, but rather energy cooperation per se that negatively affects human rights promotion effort. 20

21 No energy cooperation Energy cooperation Past human rights practices Intensity of measures GDP per capita Intrawar Interwar Democracy Regime change Other sanctions Figure 3. Coefficient plot with further robustness check. Energy cooperation not only limits leverage of the EU introducing more symmetry into the relationship, but also might undermine the credibility of the sender s commitment. Coercive enforcement can be effective when the target believes that the sender will stick to its earlier stated objectives (Eaton and Engers, 1992, p. 901). Since sanctions impose costs not only on the target but also on the sender, a credible sanction is one that imposes few costs on the sender while causing significant harm to the target. Basically, it implies that the higher the cost for the target and the lower the cost for the sender, the instrument is more likely to be credible and, therefore, effective. Moreover, not only energy exporters are important for European energy security, but also energy transit countries. For instance, energy cooperation is one of the priority areas of ENP since the beginning (e.g., Dimitrova and Dragneva, 2009, p. 865). The present findings support earlier case studies and suggest that presence of competing interests might, in fact, undermine EU human rights promotion efforts. The findings suggest that both export and transit states experience the same effect of competing interests; foreign policy instruments aimed at human rights improvement might lose their power as a result of the presence of competing interests. 21