DEPARTMENT OF POLITICAL SCIENCE Master’s Thesis: 30 higher education credits Programme: Master’s Programme in Political Science Date: 10 March 2021 Supervisor: Georgios Xezonakis & Jana Schwenk Words: 14900 THE RELATIONSHIP BETWEEN CORRUPTION, TRUST AND EFFICACY IN THE WESTERN BALKANS Direct and conditional effects Gentiana Gola This thesis is part of my studies at University of Gothenburg, funded by a Swedish Institute scholarship. Gentiana Gola 1 Abstract Empirical findings indicate that corruption perceptions decrease two key components of system support and legitimacy: citizens’ political trust and political efficacy. While the relationship between corruption, trust and political efficacy is well established in the literature, little attention has been paid so far to contextual factors that could moderate this relationship. This thesis tests both the direct effects of corruption on trust and political efficacy and potentially their conditional nature in a new context: The Western Balkan. To test my hypotheses, I use individual-level data from the Balkan Barometer, consisting of 6,020 observations, and OLS models. Results show that, as expected, perceptions of corruption decrease levels of political trust and efficacy in the Western Balkans. However, contrary to expectations stemming from previous research, citizens’ satisfaction with the economy appears to strengthen this relationship instead of weakening it. Gentiana Gola 2 Abbreviations CSO Civil Society Organisations FH Freedom House OECD Organisation for Economic Co-operation and Development RCC Regional Cooperation Council SELDI Southeast European Leadership for Development and Integrity TI Transparancy International WB Western Balkans WB6 Western Balkans six (Albania, Bosnia and Herzegovina, Kosovo, Montenegro, North Macedonia, Serbia) Gentiana Gola 3 Table of Contents Abstract ........................................................................................................................................................ 1 Abbreviations .............................................................................................................................................. 2 1. Introduction ......................................................................................................................................... 4 2. Literature Review ............................................................................................................................... 6 2.1. Corruption and system support ......................................................................................................... 6 2.2. Corruption’s effect on trust ............................................................................................................... 8 2.3. Corruption’s effect on political efficacy ........................................................................................... 9 2.4. Conditional effects: Economic context ........................................................................................... 11 2.5. The Western Balkans: Direct and conditional effects of corruption on political trust and efficacy 12 3. Theory and hypotheses ..................................................................................................................... 15 4. Data and methodology ...................................................................................................................... 18 4.1. Methods........................................................................................................................................... 19 4.2.1 Dependent Variable: Political Trust ............................................................................................ 19 4.2.2 Dependent Variable: Political Efficacy ....................................................................................... 20 4.3. Independent Variable: Corruption .................................................................................................. 22 4.4. Moderator Variable: Satisfaction with economy ............................................................................ 23 4.5. Controls ........................................................................................................................................... 24 5. Results .................................................................................................................................................... 25 5.1. Direct effect models: Corruption’s effect on trust and efficacy ...................................................... 26 5.2. Interaction effect models: Economic condition as a moderator ...................................................... 29 5.3. Linearity assumptions ..................................................................................................................... 33 6. Discussion and Conclusion ................................................................................................................... 33 8. References .............................................................................................................................................. 36 9. Appendix ................................................................................................................................................ 43 Gentiana Gola 4 1. Introduction From the theoretical perspective corruption is the antithesis of democratic norms and principles (Olsson, 2014), since its excluding nature impairs and discredits regime’s main mechanisms such as accountability, fairness, and equality. Due to regime transition, new democracies tend to lag far behind compared to established democracies in terms of democratic performance (Mishler and Rose, 2001), especially when it comes to fighting corruption or clientelistic linkages (Kitschelt, 2000; Linde, 2009). Considering the lack of socialization into democratic standards and the insufficient time for democracy to solidify in transitioning countries, corruption is one of the most serious threats to democratization (Bentzen, 2014). High levels of corruption have been found to influence political attitudes such as political trust (Linde, 2009) and efficacy (Bentzen, 2014), which carry important implications for system support and the overall legitimacy of the democratic regime (Stewart, et al., 1992; Putnam, 1993; Holmberg, 1997). If allowed to become entrenched in transitioning countries or young democracies, corruption can have detrimental effects on democracy. Despite these negative effects of corruption on vital attitudes and behaviours for democratic legitimacy and the high levels of corruption, research suggests that corrupt politicians are not always voted out (Rundquist, et al., 1977; Krause & Méndez, 2009). Several studies have analysed potential conditional effects, and have pointed out the role that the economic context could have in weakening corruption’s salience on voting behaviour, making citizens be more tolerant of corruption (Rundquist et al., 1977; Peters & Welch, 1980; Manzetti & Wilson, 2006; Konstantinidis & Xezonakis, 2013; Zechmeister & Zizumbo-Colunga, 2013; Carlin et al., 2015; Vera, 2018). Although still somewhat limited, these studies indicate that citizens might trade-off their preferences for corruption for other tangible benefits (such as a well-performing economy). I use established theories that tie corruption with political attitudes to explore a) the direct effect of corruption on trust and efficacy, and b) how the economic context moderates that relationship in the Western Balkans countries – Albania, Bosnia & Herzegovina, Kosovo, Montenegro, North Macedonia, and Serbia. These countries are usually referred to as the Western Balkans six (WB6). The research question is: How does corruption affect system support related political attitudes such as political trust and efficacy in the Western Balkans? I advance three main hypotheses. The direct effect of corruption on trust and efficacy are hypothesized to be negative Gentiana Gola 5 (H1 and H2), while I expect these direct effects to be attenuated in the case of voters who evaluate economic performance positively (H3). My findings for the first part – the direct effect of corruption – are in line with my expectations and show that corruption perceptions decrease political trust and efficacy. However, the findings do not support my second argument on the conditional effect, as they demonstrate that corruption’s negative effect on political trust and efficacy gets even stronger the more citizens are satisfied with the economy. Thus, according to these counterintuitive results, corruption becomes even more salient for citizens who evaluate the economic conditions positively. My research contributes to the literature by examining corruption’s effect on political attitudes such as trust and efficacy as individual-level factors in the Western Balkans. This argument follows theoretical approaches, which have been tested in other contexts, but never for cases like the young democracies of the Western Balkans. The second argument which considers the conditional nature of corruption’s relationship with political trust and efficacy has, to the best of my knowledge, never been empirically explored. As such, this research contributes to the scarce body of empirical research on the Balkans, and to the increased understanding of contexts similar to the Western Balkans, especially to new, emerging democracies that face issues such as brain drain, migration, poverty, resource displacement, inefficient institutions and the rise of extreme political actors striving to exploit such state of affairs. Thus, the findings could contribute by providing insights on the implications for system support and democratic legitimacy, and potentially help to guide anti-corruption efforts. I start this thesis by providing a literature review, which is divided into five parts – the first introduces literature on corruption’s impact on system support; the second provides an overview of existing studies on corruption’s effect on political trust, while the third and the fourth reviews present current literature on efficacy and on economic condition as a conditional effect. The fifth section of the literature review synchronizes literature that relates to these variables in the context of Western Balkans, discusses the motivation and research gap of this thesis. I continue by summarizing the theoretical framework that motivates the three hypotheses. Then, I present the employed data and the methodology, by discussing the methods, dependent variables, independent variable, moderator variable, and control variables. On the ensuing section, I present the results of my models and discuss those vis-à-vis my hypotheses. The concluding section provides a summary of my findings and proposes some avenues for future research. Gentiana Gola 6 2. Literature Review 2.1. Corruption and system support With regards to corruption – generally defined as the use of public office for private gain (McMann, et al., 2017), there is general agreement on its adverse effects on democratization or on the effects it has on system support. Corruption intrudes upon vital democratic principles such as equality, transparency, fairness and accountability, by jeopardizing democratic regime’s steadiness (Kubbe, 2013). When citizens get the perception their political representatives are engaged in corruption, trust and support for democracy decrease, and this consequently has adverse consequences for the legitimacy of the system (Anderson & Tverdova, 2003; Seligson, 2002; Norris, 1999). Warren (2004) suggests that corruption in a democracy usually signals a deficit of democracy. About thirty-five years ago, Johnston (1986) categorized various theories on corruption’s implications in three main groups. According to Johnston (1986), the first is the moralist camp, who regard corruption through a normative lens, focusing on its harmful effect to the development of societies and countries, and due to the detrimental effect it has on system support and legitimacy. The second group involves the revisionists camp, who emphasize the practical outcomes and benefits corruption has on specific issues or sectors, by increasing the efficiency of governments or institutions, minimizing gridlock, enabling the excluded to buy political access, and probably generating more efficient policy outcomes than those expected to derive from legitimate means (Ibid.). This group perceives corruption either as having no implications, or as having positive effects (Dobratz & Whitfield, 1992; Tavits, 2010; Johnston, 2006; Dych & Lascher, 2009, as cited in Bentzen, 2014). The final and third group consists of a rather neutral camp, who base their opinion on corruption’s effects on case-by-case basis, emphasizing the importance of contextual factors such as levels of economic development, governmental capacity, the relationship among main groups and elites, level of national integration, and other similar factors (Johnston, 1986). According to this view, while corruption can have disastrous effects for some countries or societies, in other countries where corruption is more socially acceptable, it could boost the efficiency of bureaucracy and it could help to speed things up. Pursuant to this perspective, the effects corruption has on democracy depend on several structural circumstances, such as levels of Gentiana Gola 7 government, parliamentary systems, electoral laws, and proportional systems (Karp & Banducci, 2008; Bentzen, 2014). As might be expected, most of the research on corruption is in line with what Johnston termed the “moralist camp”. There is a large body of literature that employs individual-level data and highlights the negative consequences corruption has on system support and on important aspects of democracy, such as inclusiveness, quality of public services, justice, quality of government and rule of law. Drawn upon data collected in sub-Saharan African countries, Spain, Latin America, Central and East Europe, Sweden, and in cross-sectional studies on Europe or EU countries and other countries included in the Eurobarometer, this literature demonstrates how corruption negatively influences democracies by decreasing citizens’ satisfaction with institutions and democracy (Rose, et al., 1998; Seligson, 2002; Bratton & Chang, 2006; Wagner, et al., 2009; Line & Erlingsson, 2012; Villor, et al., 2012; Stockemer & Sundström, 2013). These findings are consistent with the concept that corruption weakens government’s legitimacy and undermines democratic principles. Bratton and Chang (2006) found that rule of law, together with control of corruption are the most important democratic aspects for African citizens. As Wagner et al. (2009) state, a responsive democracy is not only about constitutionally guaranteed rights and liberties, but about a set of system outputs, and about a system rested on recognized regime performance. High levels of corruption disturb the regime performance and affect system’s outputs by altering the rules, standards, and principles in a society. Persson, Rothstein, and Teorell (2013) believe that in developing countries, such as WB6, corruption is not a principal-agent problem as assumed in various scholarly articles, but rather a collective action problem. In this way, democracy as a system becomes less and less meaningful, citizens start to see it as a futile or ineffective system, and consequently cease supporting it. Carried out in different contexts, countries and cultures, these studies demonstrate that the effective and sustainable functioning of democracy requires non-corrupt governments and institutions. As already highlighted, corruption has many negative effects for new democracy, generally and on the aggregate level. One of the ways it affects democracy is on the individual level, by affecting attitudes and behaviours, such as political trust and efficacy. Gentiana Gola 8 2.2. Corruption’s effect on trust The literature suggests that understanding what shapes trust in the authority of the institutions is crucial, as societies with higher levels of trust have more robust economies (Knack & Keefer, 1997; Woolcook, 1998), better wellbeing (Helliwell, 2002), higher wealth redistribution (Uslaner, 2002), and most importantly, better working/functioning democracies (Putnam, 1993). As Holmberg (1997) states, “representative democracy is a delicate system fundamentally built on trust and fine-tuned balance between political leadership and responsiveness.” A study conducted by Dahlberg and Holmberg (2014) using Sweden as a case study, showed that institutional trust has an independent influence on system support. Research on social justice demonstrates that citizens are more willing to obey the law and show higher levels of trust in authorities when they are treated fairly (Tyler & Lind, 1992, as cited in Eek & Rothstein, 2005). This has also been found to be the case in crisis: Harring et al. (2020) conceptualize the COVID-19 pandemic as a collective action problem and find trust to be a key factor in explaining citizens’ compliance with regulations. Specifically, citizens must trust authorities to provide them with accurate recommendations that are in their best interest, and trust that they will be followed equally by everyone (Harring, et al., 2020). As has been stated, the dominant position found in literature suggests corruption has negative implications on democracy’s focal components, such as equality, free and fair elections, rule of law, protection of individual liberties, and judicial and legislative oversight over the executive. When these democratic standards are impaired, citizens’ dissatisfaction and distrust may increase. Prevalent lasting distrust in political institutions could give rise to scepticism and lack of legitimacy of the democratic regime (Linde, 2009).1 Numerous individual-level studies have shown that corruption affects governmental performance and consequently reduces trust in government’s capacity to respond to citizens’ demands. Such conclusions were drawn from Asian countries (Chang & Chu, 2006), Western democracies (Della Porta, 2000), in the EU member states (Pellegata & Memoli, 2016), in post- communist countries of Eastern Europe (Catterburg & Moreno, 2005; Mishler & Rose, 2001), in sub-Saharan Africa (Cho & Kirwin, 2007), and Latin American countries (Seligson, 2002). 1 Eek and Rothstein (2005) have distinguished this form of distrust/trust in authorities and refer to it as vertical trust, while social trust between individuals has been labelled as horizontal trust. The focus of this thesis will be on the former – usually labelled as political or vertical trust. Gentiana Gola 9 Anderson and Tverdova (2003) found that citizens coming from corrupt democracies showed less trust in public officials and less satisfaction with democracy, whereas Rothstein and Eek (2009), observed that corrupt behaviour by public authorities influences citizens’ trust in both high- trust/low corruption contexts and low trust/high corruption contexts. An extension of the above-mentioned theories and findings point to a vicious circle, bidirectional causation between political trust and corruption. According to this view, low levels of trust increase corruption experiences, since citizens are led to use alternative means to “get things done,” adversely affecting governmental institutions’ and democratic regime’s efficiency (Della Porta, 2000; Cho & Kirwin, 2007; Uslaner, 2013; Pellegata & Memoli, 2016). Thereby, as corruption and inequality increase, citizens’ trust in institutions decreases, then decreased levels of trust increase corruption. Additional relevant studies apply slightly different approaches by incorporating other important democratic standards affected by corruption. Anderson and Singer (2008) argue that macro-level variables of inequality can be predictors of individual trust in institutions and of satisfaction with democracy – such claim was verified by Schäfer (2012), who found that high income inequality affects citizens’ satisfaction with democracy. Using data from both citizen and expert perceptions of corruption to examine how transparency influences trust, political involvement, and political interest through different levels of corruption, Bauhr and Grimes (2014) conclude that in countries with high levels of corruption perceptions, political trust and political participation were both negatively affected. 2.3. Corruption’s effect on political efficacy Political efficacy as a notion alludes to what motivates political participation and refers to the sense of self-confidence and the ability to impact political processes (Milbrath, 1965). “The feeling that individual political action does have, or can have, an impact upon the political process, that is worthwhile to perform one’s civic duties,” is one of the earliest and most used definitions of political efficacy (Campbell, et. Al., 1954, p. 187). Having a voice in politics is not enough in democracy – citizens need to feel like they can actually influence decision-making (Ulbig, 2008). In research, the distinction between internal and external efficacy is prominent: the former refers to the confidence that one can comprehend and assess political processes and manifest one’s political decisions accordingly (Craig, et al., 1990), whereas the latter is defined as citizens’ Gentiana Gola 10 capability to transform their preference into action with the aim of the action being transformed into political outcome (Hayes & Bean, 1993). The universal view is that corruption affects citizens’ personal perceptions of their prospect to impact political processes and decision-making, and in cases when they determine they cannot influence the outcome they might not bother to participate at all (Olsson, 2014). Political participation has been empirically proven to be affected by political efficacy (Verba, et al., 1995). Since political participation is the cornerstone of democracy, analysing corruption’s impact on political efficacy is a matter of great significance. Corruption is contradictory to the inclusive quality of democracy, since it excludes citizens when its central instruments – openness, accountability, and responsiveness – are impaired (Olsson, 2014). Hence, political efficacy plays a significant role in democratic theory, and is therefore connected to corruption and legitimacy of the democratic regime (Stewart, et al., 1992). Iyengar (1980) argues that efficacy’s impact extends beyond current political realities or governments and that it is related to the overall responsiveness of the political regime, referring to the importance of efficacy on system support. Hypothesizing that corruption perceptions impact efficacy, this indicates that efficacy may at least indirectly affect system support, and this can be related to democracy’s legitimacy. Research that measures the relationship between corruption and efficacy requires more interpretation and analysis. Efficacy is a rather abstract concept, frequently measured through implicit determinants related to citizens’ political behaviours and attitudes, such as their civic engagement, political participation, or citizens’ interest in politics. Since political participation has been shown to be related to political efficacy (Verba, et al., 1995; Olsson, 2014), many scholars have opted for voter turnout data to measure levels of efficacy, and found that corruption decreased levels of voter turnout (Konstandinova, 2009; Daur, 2018). Dahlberg and Solevid (2016), however, found that in highly corrupt countries, voter turnout was not affected by corruption perceptions, as compared to countries with low or medium corruption levels. Research found corruption to directly affect efficacy both on the aggregate level (Stockemer, 2013), and on the individual level (Bentzen, 2014). Olsson (2014) identifies efficacy’s mediating effect on corruption and citizens’ political behaviour and argues that corruption leads citizens to think they do not have any political influence and that their political participation is unnecessary. As this literature review has shown, there is an ongoing debate on the effect of corruption on political participation and efficacy. My thesis adds Gentiana Gola 11 to this literature by testing the argument that corruption decreases efficacy and trust in an understudied context, the WB6. 2.4. Conditional effects: Economic context Previous literature underlines the potential conditional effects of corruption on individual level attitudes and behaviour. As has been discussed above, most research on both ‘developing’ and ‘developed’ countries uncovers evidence on corruption’s negative impact on political trust (Della Porta, 2000; Mishler & Rose, 2001; Seligson, 2002; Anderson & Tverdova, 2003; Catterburg & Moreno, 2005; Chang & Chu, 2006; Cho & Kirwin, 2007; Pellegata & Memoli, 2016), and political participation and efficacy (Kostadinova, 2009; Bentzen, 2014; Olsson, 2014; Daur, 2018). However, as it has been the case in the WB6, and despite the normative expectation, politicians are not always voted out when corruption becomes salient (Rundquist, et al., 1977; Krause & Méndez, 2009). This calls for more observation and deliberation on the potential conditional effects on corruption’s relationship with trust and efficacy. Although the scope of the available research is limited, several studies point to the conditional effect of the economic context, arguing that corruption effects might be masked by economic development or economic favours. The seminal work from Rundquist et al. (1977), showed that corrupt politicians can be immune from electoral punishment, because of the “implicit trading” in which voters compromise honesty and integrity in their politicians for other benefits or factors they prioritize. One group of scholars emphasizes that due to the prospect of personal or direct economic benefits, individuals may trade their votes or support for a corrupt political candidate in exchange for certain favours or increased material welfare (Peters & Welch, 1980; Manzetti & Wilson, 2006). Other studies rather focus on the collective benefits that citizens have in mind when they decide to turn a blind eye on corruption (Konstantinidis & Xezonakis, 2013; Vera, 2018). In this case, the citizens believe that the advantages of corrupt politicians – such as their promises for decreasing taxes, or their reputation for being efficient public managers – are far too beneficial for society and outweigh these politicians’ disadvantage of being corrupt. Moreover, another body of research highlights the effect of the overall economic development, suggesting that despite of corruption levels, when the economic performance is increased, citizens’ political support tends to increase as well, and politicians are not punished for scandals as severely as when the economy is not going well (Zechmeister & Zizumbo-Colunga, 2013; Carlin et al., 2015). Gentiana Gola 12 While there is little research on that, one could expect that these conditional behaviour effects could be extended to conditional effects on attitudes. If corruption does not play a significant role on citizens’ participation levels or on their support for the incumbent when they show satisfaction with economy, it would then be expected that behaviours such as trust and efficacy would be less affected as well. However, considering there are no empirical studies that investigate this in the WB6 context, this research aims to provide insight on the leverage this factor has on the failure of anti-corruption reforms in this context. 2.5. The Western Balkans: Direct and conditional effects of corruption on political trust and efficacy The political processes for new democracies or countries that are undergoing a transition are expected to be unlike those in well-established democracies. Established democracies with a longer history of civil liberties tend to have more trust in the system as opposed to countries that lack such background (Norris, 1999). In this regard, Mishler and Rose (2001) use the example of Eastern Europe countries to demonstrate that although they have been socialized into democratic standards, post-communist societies are still vulnerable when it comes to the democratic regime due to regime transition. Kitschelt (2000) explains that states with an absence of political experience with liberal and social democratic parties are more prone to clientelistic politics. Likewise, Linde (2009) found that Central and East European post-communist EU member states lag far behind compared to the Western European democracies in terms of democratic performance, especially when it comes to corruption. The author emphasizes that distrust in public officials and institutions is one of the main attributes of the contemporary post-communist political culture due to high levels of corruption; thus, unequivocally attributing high levels of distrust to the high levels of corruption perceptions. In view of socialization into democratic standards, threats such as corruption can have a negative impact on democracy, due to insufficient time for democracy to take shape and solidify (Bentzen, 2014). If allowed to become entrenched in transitioning countries or young democracies, corruption can have detrimental effects. Having a communist past, being developing countries, and young democracies in transition, WB6 countries are more vulnerable to corruption’s threats. While post-communist citizens tend to favour democracy as a system, the continued and prevalent discontent with the Gentiana Gola 13 democratic institutions coming mainly from corruption could result in the system’s legitimacy being questioned and to populist or extremist politicians looking to exploit such discontent from citizens (Linde, 2009). Therefore, examining corruption’s implications in the WB6 deserves more attention, due to the lack of previous empirical studies that could be used for comparative analyses with contexts discussed above. Adding the WB6 context to the literature examining corruption’s effects is not only important for these countries, but could also contribute to studies of system support and regime legitimacy for countries sharing similar backgrounds to those of WB6. A number of studies, indices and reports that study or recognize the endemic corruption in the Western Balkans have been carried out (Taylor, 2013; Bjelić, 2015; Dobranja, 2015; Jaċimoviċ-Vojinoviċ, 2015; Burazer, et al., 2017; OECD, 2017; Mujanović, 2018; Bekaj, 2020; Coppedge, et al., 2020; FH, 2020; Kmezić, 2020; OECD, 2020; RCC, 2020; TI, 2020; WB, 2020). However, empirical studies that measure the relationship between corruption, trust and efficacy in the WB6 context are hard to find. Although Golubović, et al., (2015) apply an individual-level analysis by presenting a brief research about trust in political institutions in the Western Balkan countries, authors include Croatia in their study – an EU country with different political and democratic circumstances from the rest of Western Balkans – while they do not include Kosovo and do not provide a clarification for that. So, available studies, as mentioned above, are either different on the level of analysis they apply, variables they use, or countries they include. The 1990s’ processes of economic transformation in the WB6 showcased an entanglement of economic and political power entrenched in clientelism and corruption, and as confirmed in various expert reports, corruption has even suffocated judiciary and courts, which ended up controlled by corruption, political influence, inefficiency and nepotism (Kmezić & Bieber, 2007). According to RCC’s annual report (2020), corruption is ranked as the main problem in Kosovo, and although the other WB6 countries rank corruption as the third key problem after unemployment and economic development, the majority of respondents consider most of the sectors and institutions to be corrupt, leading to an increase in the corruption perceptions in the region as compared to previous years (from 26% in 2018 to 31% in 2019). Based on a steady tendency from when the Eurobarometer started asking four of the WB6 countries, the level of citizens’ distrust in the government, parliament and political parties is higher than the levels of citizens’ trust in Albania, Montenegro, North Macedonia, and Serbia (European Commission, 2019). Balkan Barometer data of 2020 shows that institutional efforts to change this reality were Gentiana Gola 14 either missing or unsuccessful, as two-thirds of the respondents were unhappy with their governments’ efforts to reduce corruption, were less trustful of rule of law institutions compared to 2019, and grew more alienated from political participation. Throughout the Balkans, there is a prevalent perception that citizens cannot meaningfully influence policymaking or government’s actions through their political participation (Fink‐Hafner & Thomas, 2019). Yet, there is no empirical research directly examining corruption’s effect in the WB6. Without such a study, the linkage between corruption and political attitudes in the WB6 will continue to be discussed only on assumptive premises. This research is hence distinctive in terms of the relationship it measures (corruption perceptions as an independent variable; satisfaction with the economy as a moderator variable; and political trust and efficacy as dependent variables), the context it analyses (WB6); and the research level (individual-level data). It aims at filling the existing research gap by addressing the implications corruption has on citizens’ political trust, and efficacy; and by testing the potential conditionalities at play in the WB6 context, which is a representation of new democracies with a post-communist past. Considering WB6’s turbulent past and present, this research could provide substantial insight on democracy by examining behaviours that are important for system support and democratic legitimacy; and by increasing the understanding not only on the Balkans context, but also in other similar post-conflict, post- communist and developing democracies contexts. Gentiana Gola 15 3. Theory and hypotheses The theoretical perspective offered here expands upon previous scholarship on corruption’s impact on trust and efficacy, and on the conditional effect of the economic context on corruption’s effects. As mentioned above, previous research conducted in diverse settings established that corruption reduces citizens’ political trust (Della Porta, 2000; Seligson, 2002; Chang & Chu, 2006; Cho & Kirwin, 2007; Pellegata & Memoli, 2016), and that citizens of corrupt countries do not only show lower levels of political trust, but also less satisfaction with democracy (Anderson & Tverdova, 2003). High levels of political corruption expose actions and results that are looked down on by citizens, because such actions decrease the quality of government, public services, and the welfare of citizens. When bribery is required to receive public services, citizens start to doubt the integrity of the institutions and of the public servants, and realise that public servants place private gains above the constitution and laws. When cronyism and nepotism replace meritocracy in employment in the institutions, citizens question the competence of their public servants and representatives, while realising that a group of people is establishing power in unjust ways. Moreover, corruption may facilitate other criminal actions such as money laundering. Eventually, corruption affects citizens’ lives in many ways – the government spending shifts away from the public good in areas such as welfare, education, and health, toward inefficient or lavish ventures based on the corrupt elite’s preferences and interests; economic growth and investments decrease; and wealth and social inequality increase. For instance, in the context of the WB6, the fight against corruption has been an EU integration condition for the WB6 countries. Lack of independence of the institutions operating with the captured and controlled triangle of police, public prosecutor and the judiciary, made the actual implementation of anti-corruption measures almost impossible (Kmezić & Bieber, 2007), affecting EU accession of the WB6 countries. Under such circumstances, and through such mechanisms, when corruption hinders the quality of life and the prospects for development, citizens start to presume that politicians are placing personal interests and benefits above the country’s needs and goals. They perceive corrupt politicians/institutions as morally and/or professionally unfit to lead – because, they believe the corrupt elite does not have the credibility to ask for law compliance, does not have the right reputation to lead by example, does not have the interest to focus on citizens’ wellbeing, or is Gentiana Gola 16 incompetent to do so. In this manner, corruption perceptions decrease citizens’ trust in their institutions and/or politicians. Hence, I expect that: H1: As perceptions of corruption increase, political trust will decrease. As it has already been pointed out, corruption perceptions affect political efficacy in a similar fashion (Kostadinova, 2009; Stockemer, 2013; Bentzen, 2014; Olsson, 2014). Consistent with the collective action theory, scholars found that in such high corruption settings, citizens gradually start to disengage from politics (Bauhr & Grimes, 2014), not bothering about voting or other forms of political participation (Warren, 2004). Similar to political trust, corruption mechanisms such as electoral fraud, bribery, influence peddling or nepotism negate vital democratic principles and aspects such as equality, transparency, and fairness. When citizens, for example, witness a politician trying to influence a judge, they perceive their institutions as unjust and devious – they get the impression that civic engagement or fight for justice is useless, because the system is rigged and set up to produce corruption. In these circumstances, corruption becomes so widespread that it is the “rule of the game,” and the benefits of engaging in corruption become higher than the price a citizen has to pay for remaining honest. Since no “principled principals” are left to carry out anti-corruption initiatives, most citizens are more likely to embrace rather than fight corruption (Persson, et al., 2013). When citizens do not have confidence that the elections will not be rigged, that their vote matters in selecting political representatives, and when they consider that the corrupt political elites have captured the state, and everyone else if benefiting from corrupt practices, they might see no point in voting or in being politically active. This being the case, citizens become alienated from the political system, because it becomes irrational or useless for them to engage in fighting corruption, due to the fact other citizens are also either involved in corruption or have become reluctant to fight corruption (Bauhr & Grimes, 2014; Persson et al. 2013), and because they either feel the political elites being focused in their own interests and are not responsive to citizens’ needs, or because the power of corruption and of state capture is so eminent with all society being part of it that the citizen feels that their voice does not matter. Corruption gradually alienates citizens from the political system and directs them toward thinking they are unable to make any changes. When such events become the “new normal” in a society, elections are rigged, electoral Gentiana Gola 17 promises are broken, and the “rotten apples” are protected by the institutions they bribe. Citizens are left with no incentive to participate in politics. Thus, I argue that: H2: As perceptions of corruption increase, political efficacy will decrease. Nonetheless, this research acknowledges that the linkage between corruption, trust and efficacy is not always straightforward, and that there are circumstances that condition the effect of corruption on political trust and efficacy. Worth noting is the moderator effect the economic context has on corruption’s impact on political support (Zechmeister & Zizumbo-Colunga, 2013; Carlin et al., 2015). This argument considers the implicit trading between the citizens and the corrupt elite, in which citizens exchange their political support for benefits, favours or issues they prioritize (Rundquist et al., 1977). In such circumstances, corruption seems to be ranked lower on the hierarchy of citizens’ needs, as compared to other needs such as the economy. This appears to be the case especially in countries that struggle with the economy, such as those in the WB6. Citizens of these countries overwhelmingly agree that unemployment and economic development are the two main problems their countries face (RCC, 2020). Not only does public perception point out that economic development is an issue, but objective measurements of economic development also support this view. 90% of the economies of the WB6 consist of small and medium-sized enterprises (RCC, 2020), five out of the six WB countries experienced an increase in the hidden employment index between 2016 and 2019, and a high number of employees in the WB6 do not have healthcare insurance and receive ‘envelope salaries’ (SELDI, 2020). It is, thus, not hard to imagine the hardships the economies of these countries face to survive global challenges, such as the COVID-19 pandemic, and how desperately citizens may need any form of economic boost or support and are willing to compromise in other areas. When corrupt politicians can improve the overall economic performance, citizens may not punish them as severely. Citizens in corrupt environments are directly impacted or threatened by unemployment and poverty, are vulnerable due to bank loans, and have a history of harsh living conditions. Hence, if a corrupt politician creates jobs, brings investors in, and increases economic development, citizens may not be concerned about the corruption allegations that a politician faces. When citizens become socialized with corruption, and somehow start to witness economic benefits in corrupt settings, they may not bother as much about its consequences. If they get richer or find a job through nepotism, if taxes are decreased, or if there is economic progress, citizens Gentiana Gola 18 may not be (as) dissatisfied with corruption and their political trust and efficacy may not be as affected, in comparison to when they would not be benefiting personally or collectively. In such cases, citizens may not have an incentive to “throw the rascals out” (Rundquist, et al., 1977) and their trust and efficacy may not be (as) affected. In new democracies such as the WB6, people may be concerned with short-term benefits such as the flourishing of the economy, and that is likely to keep corruption less salient, especially when it comes to their evaluations of political trust and efficacy. The economic conditions might, hence, moderate the effect of corruption on attitudes. So, I argue that: H3a: The effect of corruption on trust is conditional upon economic perceptions. The more positive economic evaluations are, the lower the effect of corruption. H3b: The effect of corruption on efficacy is conditional upon economic perceptions. The more positive economic evaluations are, the lower the effect of corruption. 4. Data and methodology Acquiring data for the WB6 countries is challenging, considering the largest and most credible databases such as World Value Survey, European Social Survey, or Eurobarometer, do not include some of the WB6 countries, or provide limited data on the variables needed for this research. Thus, to test these hypotheses, I use public opinion survey data from the sixth edition of the Balkan Barometer, conducted over late 2019 and early 2020, and issued in 2020. This survey is commissioned by the Regional Cooperation Council (RCC), which is a regionally led cooperation framework funded by the EU, engaging RCC members from South-East Europe, the international community, and donors, to enhance the Euro-Atlantic integration of the Balkans. The initial aim of this survey was to monitor the progress of the WB countries in meeting South-East Europe 2020 Strategy – however, now the survey aims to draft a post-2020 agenda to observe the objectives and expectations on life and work, predominant socio-economic and political movements, and WB’s regional and European integration (RCC, 2021). The sample of this annual survey was stratified according to the region or county and the survey was conducted among 6,020 respondents throughout Albania, Bosnia and Herzegovina, Kosovo, Montenegro, North Macedonia, and Serbia (on average N=1000 respondents per country), aged 18+ (RCC, 2020). Respondents who reported ‘don’t know answers’ were removed from the analysis of all variables in this research, and as a Gentiana Gola 19 result, the number of observations dropped from a total of 6,020 to a minimum of 4,051 and a maximum of 5,689 depending on the model (see Table 1 in the Appendix). It should be highlighted that the survey questions use the word ‘economy’ instead of ‘countries,’ to account for Kosovo’s status without prejudice, and in line with UNSCR 1244/1999 and the ICJ Opinion on the Kosovo declaration of independence (RCC, 2020). 4.1. Methods Considering attitudes are subjective traits, and the focus of this research is attitudes, I use individual-level data as the most ideal type of data for the purpose of this research. I employ linear regression with country-fixed effects – essentially a multilevel model that accounts for unobserved differences between countries. Considering the variables I use are both discrete and categorical, I recognize that using OLS regression might not be deemed ideal; however, since the variables are coded on a linear scale, using a linear regression is still suitable. I run tests for the common OLS assumptions (see Tables 4-6, Graphs 1-12, Tests 1-6 in the Appendix). 4.2.1 Dependent Variable: Political Trust Since trust is an intrinsically subjective individual attribute, it is best measured on the individual level (Pharr & Putnam, 2000). In order to estimate the effect of corruption perceptions on an individual’s political trust, I measure trust by creating an index. The RCC survey includes five questions aiming at assessing the levels of trust in institutions: the parliament, government, courts and judiciary, ombudsman, and the supreme audit institutions. Since all factors address trust in institutions, from a theoretical perspective it seems sensible to combine them into an index. To test whether these theoretical expectations hold, correlation coefficients between the indicators were calculated (see Table 2 in the Appendix). High correlations between the indicators suggested that combining them into an index of political trust is sensible also from an empirical point of view.2 Hence, using all available survey questions measuring citizens’ political trust in separate political and public institutions, could ensure a more comprehensive and thorough capture of political trust. All five questions measuring trust were phrased as “How much trust do you have,” 2 The theoretical rationale of using an index to measure political trust relates to the fact political trust is a general concept that can depend upon different political actors and institutions. Gentiana Gola 20 followed by the institution in question. Answer categories for all questions were “totally distrust”, “tend to distrust”, “tend to trust”, “totally trust”. A summative index was created such that higher values overall represent higher levels of trust. The final variable ranges from 5 to 20. Nevertheless, acknowledging the warning of Cronbach and Shavelson (2004) that a single statistic to summarize the accuracy of a variable might not be the best option, I run a second model as a robustness check, in which political trust is measured by citizens’ trust in parliament and the results remain unchanged (see Table 3 in the Appendix, Model 1 for the main model and Model 2 for the interaction). The question wording and answer categories are as described above. As discussed, political trust can vary across institutions, but when having to choose one form of political trust, confidence in parliament appears to be the most evident alternative, as parliaments are supposed to be the main decision-making institution in all systems (O'Brien, et al., 2008). Furthermore, parliaments are not only responsible for forming and holding governments accountable, but also for making laws – thus, distrust in policy-making institutions should result in lower levels of obedience/compliance. Therefore, compared to trust in other institutions, trust in parliament seems to be the most important form of political trust if only one form of institutional trust was to be chosen (Rolef, 2006; Holmberg, 2008). 4.2.2 Dependent Variable: Political Efficacy Measuring political efficacy as an abstract concept comes with its own challenges as well. Conventionally, political efficacy is measured through both citizens’ political behaviours and attitudes such as their political participation (Konstandinova, 2009; Daur, 2018), or their interest in politics (Bentzen, 2014). Since the survey at hand did not include survey questions examining respondents’ turnout in the elections an alternative measurement had to be found. I rely on two questions that are assumed to operationalize the two types of political efficacy, internal and external efficacy, appropriately. To measure internal political efficacy, conventionally defined as “the confidence that one can comprehend political processes accordingly” (Craig, et al., 1990), I rely on the question: Information of government is easy to understand: Do you agree that written information of your Government (such as laws, decisions, web pages, forms) is overall easy to understand and uses plain language? This is appropriate as understanding government information is assumed to approximate an individual’s perceived ability to understand political Gentiana Gola 21 information. To measure external efficacy, defined as “citizens’ capability of transforming their preferences into action” (Hayes & Bean, 1993), I decided to use this question: Ability of the citizens and CSOs to scrutinize government: Do you agree that the citizens and civil society organisations can effectively scrutinize the government and make it accountable to citizens? Since external efficacy captures citizens’ feelings on their ability to affect political decision-making, this question was an appropriate fit. If citizens think they are able to understand the government’s information and that they have the correct information to take political actions such as to scrutinize the government’s work, their internal political efficacy would be affected. Whereas, if citizens think they can scrutinize their political elites, it shows that by providing them with the tools to be able to inspect their work, governments are responsive to citizens, and citizens consequently feel they can impact policymaking. Although these questions might not be ideal, they capture political efficacy by both its perspectives and focus on attitudes, as is usually the case with political efficacy. These questions complement one-another, and provide an all-inclusive measure of political efficacy, by being in line with Milbrath’s (1965) general definition of political efficacy - the sense of self-confidence (internal efficacy) and the ability to impact political processes (external efficacy). It should be highlighted that the reason these survey questions were not combined into an index is because they showed very low correlations (see Table 2 in the Appendix). Moreover, since they represent different types of political efficacy, from a theoretical point of view, more insight is to be gained by keeping the measures separate. This allows us to examine if corruption has a varying effect on different types of efficacy. The answer possibilities for both these questions were: 1) Totally disagree; 2) tend to disagree; 3) tend to agree; 4) agree. The variables are coded in such a way that higher values represent higher levels of political efficacy. A third alternative survey question, asking whether the citizens have ever done something to affect the government’s decisions, was considered to measure political efficacy. However, it was ultimately decided that instead of measuring political attitudes, this question rather measures political behaviour, since it aims to assess conventional and unconventional political participation.3 3 Have you ever done something that could affect any of the government decisions? 1) Yes, I did, I took part in public debates; 2) Yes, I did, I took part in protests; 3) Yes, I did, I gave my comments on social networks or elsewhere on the Internet; 4) I only discussed it with friends, acquaintances, I have not publicly declared myself; 5) I do not even discuss it. Gentiana Gola 22 4.3. Independent Variable: Corruption Since corruption usually exists in the shadows and is difficult to measure (Bentzen, 2014), I rely on a measurement of corruption perceptions. Corruption perceptions have been shown to be appropriate approximations of actual corruption (Charron, 2016), and the RCC survey includes measures of corruption perceptions, allowing for individual-level analysis. In order to estimate citizens’ perceived level of corruption, respondents were asked the following questions: 1) To what extent do you agree or not agree that the parliament in your economy is affected by corruption? 2) To what extent do you agree or not agree that public officials /civil servants in your economy are affected by corruption? Alternatively, a third survey question was considered to measure corruption perceptions: 3) Do you agree that in your economy the government fights corruption effectively? Upon thorough analysis, the third question on the fight against corruption appeared to account for a separate theoretical concept, since the fight against corruption and the existence of corruption perceptions are not necessarily mutually exclusive. Hence, the third survey question was discarded, and an index between the first (parliament affected by corruption) and second (public officials and civil servants affected by corruption) survey questions was created. I create an index out of these two survey questions since they examine citizens’ perceptions of corruption in parliament, public officials and civil servants. As argued above, parliaments are probably the most important political institution to examine in this research, since they are the central decision-making institution (O’Brien, et al., 2008). Examining citizens’ perceptions of whether civil servants and public officials are affected by corruption probably captures all other political institutions. These two groups include administrative or junior level civil servants with whom citizens tend to have direct contact, top executive public officials such as ministers or prime ministers, and everything in between. Moreover, they are part of all local and central political institutions such as municipalities, governments, ombudsperson and so forth. This combination is, hence, ideal to capture all potential experiences or information that contribute to citizens’ corruption perceptions. Hence, considering the high correlation (see Table 2 in the Appendix) Gentiana Gola 23 between these two survey questions, their unidimensional concept, and the fact they quite directly capture citizens’ perceptions of corruption, this index appears to be an ideal measure of citizens’ perceptions of corruption. The answer categories for this variable are: 1) Totally disagree; 2) tend to disagree; 3) tend to agree; 4) agree. I combined the questions in an additive index in which higher levels represent higher perceptions of corruption. The final variable ranges from 2 to 8. 4.4. Moderator Variable: Satisfaction with economy To assess the moderating effects of the economic context on corruption’s influence on political trust and external efficacy, an index of citizens’ satisfaction with the economy was created out of these two survey questions: 1) Economic situation: How satisfied are you with the economic situation in your economy? 2) Household finances: How satisfied are you with the financial situation of your household? Alternatively, this survey question was considered as well: 3) National economy expectations: What are your expectations for the national economy? Do you think that in 12 months the state of the economy will be: Better; Worse; The same. Since the third survey question examining citizens’ expectations of the national economy showed a low correlation with the other two variables, this variable was not included in the index. This question also appeared to account for future expectations, rather than for present economic conditions or perceptions. Whatever the answers to the third question, it would be difficult to interpret them into current perceptions regarding citizens’ satisfaction with the economy. Conversely the first and second survey questions capture citizens’ satisfaction with the economy on both the country and household levels. The questions are straightforward and directly ask about citizens’ current personal satisfaction with the economy, instead of referring to future expectations. These questions are, hence, more likely to provide answers related to citizens’ actual perceptions of the economic situation. Furthermore, these questions showed high levels of correlation (see Table 2 in the Appendix) between each-other. The answer categories for these questions are: 1) I’m completely dissatisfied; 2) I’m mostly unsatisfied; 3) Neither satisfied nor dissatisfied; 4) I’m mostly satisfied; 5) I’m completely satisfied. Higher values in this index refer to higher levels of satisfaction with the economy. The final variable ranges from 2 to 10. Gentiana Gola 24 4.5. Controls As might be expected, apart from satisfaction with the economy, there are other factors that are likely to affect political trust, efficacy and/or their linkage to corruption. Additional variables to be included in this study are age, gender, education, country, social status, working status, and the satisfaction index. Education, gender, age and social status are considered as important determinants of political participation (Verba, et al., 1995; Olsson, 2014), which as discussed is closely related to political trust and efficacy. Older people might have a more cynical and less- trustful approach toward the political system due to having lived through many years of disappointment, while younger people might be less likely to experience corruption (Seligson, 2002). Scholars have found that, for instance, higher education levels lead to higher income, and citizens with higher income tend to vote more (Oscarsson, 2007), and that although various participation forms appeal to different types of participants, citizens with higher levels of education are widespread through all participation types (Marien, et al., 2010). Political attitudes might be affected by standard demographic factors, lack of resources, psychological engagement with politics, or because they are not part of networks that bridge people to politics. Age and social characteristics such as gender and education have also been found to impact citizens’ perceptions on their capacity or ability to affect government’s decisions, with younger individuals showing more cynicism in regards to politics and less efficacy (Listhaug & Wiberg, 1998), and with women feeling that they have less influence in politics, leading them to show lower levels of efficacy as well (Abramson, 1983). It is, thus, important to disentangle these factors so as to see the real effects of corruption on political attitudes. In regards to the independent variable, cognitive resources such as education are likely to affect both political attitudes and perceptions of corruption; therefore, it is imperative they are included as controls. Charron and Rothstein (2016) found that in countries with high corruption levels, citizens tend to perceive the system as ‘rigged’ as they become more educated, and this could have an implication on the linkage between education and trust at the individual level on such contexts. Higher levels of education are not only likely to lead to higher levels of information regarding the performance of the system, but also to increased levels of a critical approach, and to an enhanced reaction toward corruption (Norris, 1999; Seligson, 2002; Agerberg, 2019). Scholars have shown that gender plays a role in how citizens experience and perceive corruption (Bauhr, et Gentiana Gola 25 al., 2018). Controlling for all country-level specific factors, such as type of political system could help to determine or explain potential significant differences, based on political systems, levels of democratization, levels of economic development, and other important country-specific factors. Moreover, the satisfaction index will be used to measure the overall sentiment of citizens regarding how things are going in their respective countries. This variable will help to control for corruption’s real effects on the general evaluation citizens make for their countries. Table 1 – Descriptive Statistics in the Appendix provides an overview of the variables used, and Graphs 13-17 shows the used variables by country. 5. Results Below I present six different models to estimate the effects of perceived corruption on citizens’ attitudes4. Models 1-3 in Table 1 refer to the direct effects of corruption perceptions on political trust and political efficacy. Hypothesis 1 (As perceptions of corruption increase, political trust will decrease) is tested in Model 1, using the political trust index. Hypothesis 2 (As perceptions of corruption increase, political efficacy will decrease) is tested in Model 2 and Model 3, with Model 2 using the variable focusing on internal efficacy, and Model 3 focusing on external efficacy. Models 4 – 6 in Table 2 mirror Models 1-3 in terms of the dependent variables, but include an interaction effect with satisfaction with economy to test H3a and H3b. Hypothesis 3a (the attenuating effect of economic perceptions on trust) is tested in Model 4; whereas, Hypothesis 3b (the attenuating effect of economic perceptions on efficacy) is tested in Model 5 and Model 6. Higher values in my independent variable show higher levels of perceived corruption. Higher values in my dependent variables indicate higher levels of trust and efficacy, while higher levels of the moderator variable indicate higher levels of satisfaction with the economy. For Hypothesis 1 and Hypothesis 2 to be supported, I expect a negative sign of the coefficient in Models 1 to 3. Meanwhile, for Hypothesis 3a and Hypothesis 3b to be supported, I expect the signs of the interaction terms to be different from the main effects, such that the corruption variable would show a negative value, and the interaction variable a positive value in the respective models (Models 4 to 6). 4 Dummy variables for each country (country-fixed effects) are not included in the tables for presentation purposes. Gentiana Gola 26 5.1. Direct effect models: Corruption’s effect on trust and efficacy Table 1: Direct Effect of Corruption on Trust and Efficacy Model 1 Model 2 Model 3 Political Trust Index Internal Political Efficacy External Political Efficacy b/se b/se b/se Corruption Index -0.960*** -0.075*** -0.115*** (0.05) (0.01) (0.01) Satisfaction with Economy Index 0.848*** 0.145*** 0.121*** (0.03) (0.01) (0.01) Age 0.095 -0.018 0.006 (0.05) (0.01) (0.01) Gender -0.141 0.016 -0.030 (0.11) (0.03) (0.03) Education: High School 0.613** 0.053 0.037 (0.19) (0.05) (0.05) Education: College / University 0.579** 0.163** 0.051 (0.22) (0.06) (0.06) Education: Master’s Degree / Doctorate 0.640 0.017 -0.062 (0.46) (0.11) (0.10) Social Status: Average -0.107 0.040 -0.022 (0.15) (0.04) (0.04) Social Status: Above Average 0.084 0.154* -0.134 (0.30) (0.07) (0.08) Work Status: Retired -0.105 0.016 -0.038 (0.23) (0.06) (0.07) Work Status: Housewife 0.317 0.111 -0.060 (0.29) (0.07) (0.07) Work Status: Student/Pupil 0.242 0.075 0.102 (0.27) (0.07) (0.07) Work Status: Self-Employed 0.103 0.052 -0.020 (0.23) (0.06) (0.06) Work Status: Employed 0.094 0.081 0.046 (0.16) (0.04) (0.04) Constant 11.316*** (0.51) 2.058*** (0.12) 2.433*** (0.13) R2 0.436 0.204 0.148 N 3239 3244 3384 Notes: * p < 0.1; ** p < 0.05; *** p < 0.01. Standard errors are adjusted for heteroscedasticity by using robust standard errors. Controlled for country dummies (not displayed). Base levels for categorical variables are: Education: No education / primary school; Social Status: Below average; Work Status: Unemployed. Gentiana Gola 27 Model 1 presents the effect of corruption perceptions on citizens’ political trust. Choosing a cut-off for the significance level of α < 0.05, the coefficient of corruption is significant. The results show that the effect of a 1 unit increase in corruption perception’s index decreases the political trust index by 0.96, on average. Hence, in line with studies that indicate toward corruption’s negative effect on political trust (Della Porta, 2000; Seligson, 2002; Chang & Chu, 2006; Cho & Kirwin, 2007; Pellegata & Memoli, 2016), I find support for Hypothesis 1. Apart from satisfaction with the economy and education (high school and college/university), the other control variables are not statistically significant. Political trust seems to be positively affected by citizens’ satisfaction with the economy, as expected. Higher levels of education have a positive effect on trust compared to the base level of primary education, but the effect is not significant for all levels. The R2 of 0.436 indicates that 43.6% of the variance in political trust is explained by the model. Model 2 measures the effect of corruption perceptions on citizens’ political efficacy, by using the internal efficacy variable as a dependent variable. A p-value of p < 0.01 demonstrates that the coefficient of corruption perceptions is significant and that the null hypothesis can be rejected. These findings indicate that the effect of a 1 unit increase in corruption perception’s index decreases internal efficacy by 0.075 on average. Moreover, as expected by previous body of scholarship (Kostadinova, 2009; Stockemer, 2013; Bentzen, 2014; Olsson, 2014), the negative coefficient confirms that Hypothesis 2 is supported, and that corruption perceptions have a negative effect on internal efficacy. Statistically significant control variables in this model are satisfaction with the economy, education, work status, and social status. Satisfaction with the economy increases internal efficacy. Compared to those with no education, individuals with a college or university degree are significantly more likely to report that they perceive government information easy to understand. Similarly, individuals with above average social status have significantly higher internal efficacy than those with a below average social status. The R2 value of 0.204 reveals that 20.4% of the variance of internal efficacy is explained by the model. Nevertheless, the other part (79.6%) that is not explained by this model, is explained by other variables that could have a potential effect on internal efficacy, but that are not included in this model. Model 3 also examines the effect corruption perceptions have on political efficacy, but it employs an external efficacy variable. Similar to the internal efficacy variable in Model 2, the p- Gentiana Gola 28 value of 0.01 in this model indicates that the coefficient of corruption perceptions is significant and that the null hypothesis is, thus, rejected. This outcome suggests that the effect of a 1 unit increase in corruption perception’s index decreases an individual’s external efficacy by 0.115, on average. The coefficient’s negative sign in this model confirms Hypothesis 2, and is in accordance with the claim that external efficacy is negatively affected by corruption perceptions. Satisfaction with the economy is statistically significant and it positively affects external political efficacy. The R2 value of 0.148 indicates that 14.8% of the variance of external efficacy is explained by the model. Nevertheless, the other part (85.2%) that is not explained by the model, is explained by other variables that could have a potential effect on external efficacy, but that are not included in the model. Gentiana Gola 29 5.2. Interaction effect models: Economic condition as a moderator Table 2: Interaction effect models Model 4 Model 5 Model 6 Political Trust Index Internal Political Efficacy External Political Efficacy b/se b/se b/se Corruption Index -0.707*** -0.006 -0.078* (0.13) (0.03) (0.03) Satisfaction with Economy Index 1.100*** 0.210*** 0.147*** (0.13) (0.03) (0.03) Corruption Index*Satisfaction with Economy Index -0.045* -0.012** -0.007 (0.02) (0.00) (0.01) Satisfaction Index 0.078 0.029** 0.040*** (0.04) (0.01) (0.01) Age 0.086 -0.019 0.005 (0.05) (0.01) (0.01) Gender -0.131 0.020 -0.024 (0.11) (0.03) (0.03) Education: High School 0.631*** 0.057 0.039 (0.19) (0.05) (0.05) Education: College / University 0.600** 0.171** 0.060 (0.22) (0.06) (0.06) Education: Master’s Degree / Doctorate 0.659 0.027 -0.046 (0.46) (0.11) (0.11) Social status: Average -0.099 0.042 -0.021 (0.15) (0.04) (0.04) Social status: Above Average 0.089 0.158* -0.123 (0.30) (0.07) (0.08) Work status: Retired -0.073 0.020 -0.032 (0.23) (0.06) (0.07) Work status: Housewife 0.358 0.115 -0.056 (0.29) (0.07) (0.07) Work status: Student/Pupil 0.275 0.085 0.114 (0.28) (0.07) (0.07) Work status: Self-Employed 0.115 0.052 -0.020 (0.23) (0.06) (0.06) Work status: Employed 0.104 0.081 0.045 (0.16) (0.04) (0.04) Constant 9.515*** 1.550*** 2.099*** (0.98) (0.22) (0.23) R2 0.437 0.208 0.152 N 3239 3244 3384 Notes: * p < 0.1; ** p < 0.05; *** p < 0.01. Standard errors are adjusted for heteroscedasticity by using robust standard errors. Controlled for country dummies (not displayed). Base levels for categorical variables are: Education: No education / primary school; Social Status: Below average; Work Status: Unemployed. Gentiana Gola 30 In order for Hypothesis 3a and 3b to be supported, the corruption variable would have to show a negative value, and the interaction variables positive values in the models in Table 2. The table shows negative signs in both the corruption and the interaction variables, suggesting to the opposite expectations of the hypothesis. Considering that these models (4-6) are mirroring the direct effect of models 1-3, the results of the control variables are almost identical, and will not be discussed. In Model 4, while corruption affects political trust negatively, higher levels of satisfaction with economy have a positive effect on political trust. Both effects are statistically significant. The interaction effect, contrary to my expectation, has a sign that goes in the same negative direction as the main effect of corruption, indicating that as satisfaction with economy increases, the negative effect of corruption becomes stronger. Thus, there is no support for Hypothesis 3 which states a moderating effect of satisfaction with economy on the effect of corruption on political trust. Based on the significance value of 0.1, the null hypothesis is rejected. The R2 of 0.437 indicates that 43.7% of the variance in political trust is explained by the model, and that when the interaction effect was added the R2 was marginally improved by 0.01% compared to Model 1(from 43.6% to 43.7%). The Model 4 margins plot also illustrates that the effect moves in the opposite direction. Figure 1: Margins plot for Model 4 Gentiana Gola 31 In Model 5, although corruption has a negative effect on internal political efficacy, the effect is not statistically significant. Satisfaction with economy, however, has a statistically significant positive effect on internal political efficacy, showing that as satisfaction with economy increase, so do the levels of internal efficacy. Regarding Hypothesis 3b, stating a moderating effect of satisfaction with economy on the effect of corruption on internal efficacy, Model 5 shows similar result as Model 4 – negative signs in both the main and the interaction variables demonstrate that as satisfaction with economy increases, the negative effect of corruption on internal efficacy becomes stronger. Hypothesis 3b is also rejected. These results are interesting, since they go against the expectation that the economic conditions weaken corruption’s negative influence on efficacy. The significance value is 0.05, and hence the result is statistically significant. Compared to Model 2, the R2 value has been increased by 0.4% (from 20.4% to 20.8%). The margins plot for Model 5 also shows that the effect moves in the opposite direction. Figure 2: Margins plot for Model 5 Gentiana Gola 32 Similarly to Models 4 and 5, corruption has a negative effect, and satisfaction with the economy has a positive effect on the dependent variable – external political efficacy. Both corruption and satisfaction with the economy have statistically significant effects. This model is also similar in regards to the shown results of the moderator effect. Negative signs in both the main and interaction variables indicate that as satisfaction with the economy increases, corruption’s negative effect of corruption on external political efficacy becomes stronger. When satisfaction with the economy increases, the effect of corruption perceptions on citizens’ belief that they are able to scrutinize the government is still strengthened. However, as compared to Models 4 and 5, this negative effect is less strong for Model 6, since the interaction effect is not statistically significant, and the null hypothesis cannot be rejected. Although Hypothesis 3b is also rejected in this model, as compared to the significant effect for internal efficacy, there is no significant effect for external efficacy. The R2 value has been increased by 0.4% as compared to Model 3 (from 14.8% to 15.2%). The margins plot for Model 6 shows that the effect moves in the opposite direction as well. Figure 3: Margins plot for Model 6 Gentiana Gola 33 5.3. Linearity assumptions Linearity assumptions were tested (see Tables 4-6, Graphs 1-12, Tests 1-6 in the Appendix). Multicollinearity was not found in any of the tested models (direct effect models). When conducting an outlier analysis, several outliers were detected. However, since the number of observations in all the models is considerably large, outliers should not have a too large influence on the results. To adjust for the heteroskedasticity that was detected in several models, robust standard errors were introduced in all six models (c.f. heteroscedasticity analysis in the appendix, including the test for normal distribution of residuals). 6. Discussion and Conclusion This research aimed to examine the implications corruption has on political trust and efficacy. To do so, I established a dual theoretical rationale to explain for corruption’s effects. First, I hypothesized that corruption perceptions decrease political trust and efficacy. Second, I hypothesized that the economic conditions moderate the effect of corruption on political trust and efficacy. Utilizing Balkan Barometer individual-level data with 6,020 observations from Albania, Bosnia and Herzegovina, Kosovo, Montenegro, North Macedonia, and Serbia, I tested these expectations on six models and found evidence that supports the first assumption, but not the second. My results support the hypotheses that corruption decreases political trust and efficacy. However, the results do not support the hypotheses stating a moderating effect of satisfaction with the economy. Instead, the results from the models show that the satisfaction with the economy does not weaken corruption’s negative influence on trust and efficacy, but does the exactly opposite – the higher the levels of satisfaction with the economy, the stronger the effect of corruption becomes on political trust and political efficacy. Contrary to internal efficacy, the effect of external political efficacy is, however, not statistically significant. The finding that satisfaction with the economy increases the negative effect of corruption on trust and internal efficacy could be due to citizens’ perceptions that corruption is holding the country back. As citizens become more satisfied with the economy and the existing economic progress, they could expect such progress to be even larger and more persistent if corruption levels were lower. In other words, in Gentiana Gola 34 this case citizens might think that the country could be doing even better if it was not for corruption, and that corruption is not allowing for more progress. Given the lack of previous literature that could empirically explain these results, one could only speculate about their outcomes and implications. I would argue that these findings suggest that the WB6 citizens are not (as) socialized with corruption as is sometimes expected (Tavits, 2010). Ethnic conflicts, wars, communism, and the breakup of Yugoslavia contributed to the slow development of these countries compared to Western European democracies. Their political elite engaged in corruption and installed a system that lacked transparency, making it difficult for civil society and citizens to hold them accountable. A lot of civic voices and protests were silenced, and many elections were rigged due to this deeply embedded corrupt system that only favoured a few (Kmezić & Bieber, 2007). Considering 90% of the economies of the WB6 consist of small and medium-sized enterprises (RCC, 2020), and that shadow economy and high levels of informality in employment are prevalent in the WB6 (Laderchi & Savastano, 2013; Kelmanson, et al., 2019), these countries face enormous challenges to survive global challenges, such as the COVID-19 pandemic. Aware of corruption’s effect on the economic development in their region (Kadia, 2020), WB6 citizens’ satisfaction with the economy is apparently making them even more aware that they could have been doing better if corruption would not be as high. Being part of Europe and not far away from Western democracies, or from countries such as Croatia and Slovenia that shared a similar faith and are now EU-member states with consolidated democracies, WB6 citizens seem to be frequently reminded that they should be raising their standards and that the progress in their countries is slow. These could be reasons as to why despite their satisfaction with the economy, their political trust and efficacy are decreased due to corruption. The public’s disappointment with most political parties and the lack of different political party alternatives throughout the region could also explain the (not as) decreased political efficacy in the conditional effect models. These findings show the importance of applying previous research in new settings and contexts. Yet, in regards to external validity, it might be difficult to draw similar conclusions or generalizations for other new post-conflict democracies considering the unique and distinctive historical setup of the WB6 countries. There is also a possibility of reverse causality or of a vicious circle, leading to a bidirectional causation between political trust and corruption (Della Porta, Gentiana Gola 35 2000; Cho & Kirwin, 2007; Uslaner, 2013; Pellegata & Memoli, 2016). As discussed above, another shortcoming could also be the less-than-ideal measurement of political efficacy. Nonetheless, although theories that already exist in the literature were initially used to build the expectations of this research, this is the first study that investigates the effect of corruption on trust and political efficacy in the context of the WB6 countries. The study of this setting is important, because these countries represent new democracies that are aiming at integration in the EU and other transnational institutions. The novelty of this research is not only limited to the fact that this dataset is tested in this context for the first time, but also by its surprising finding on the interaction effect of the economic condition and its effect on corruption. To summarize once again, while this research confirmed the theoretical expectation that corruption negatively affects trust and political efficacy, the surprising finding that satisfaction with the economy does not weaken this relationship begs further testing. While above I speculate about potential reasons, a thorough theoretical and empirical investigation is needed. In doing so, the understanding of corruption’s effect on political attitudes across various contexts could be enhanced. This understanding is crucial for the implications corruption has in terms of democracy and system support. Gentiana Gola 36 8. References Abramson, P. R., 1983. Political Attitudes in America: Formation and Change. 1st ed. San Francisco: Freeman. Agerberg, M., 2019. The Curse of Knowledge? Education, Corruption, and Politics. Political Behavior, 41(1), p. 369–399. Anderson, C. J. & Tverdova, Y. V., 2003. Corruption, Political Allegiances, and Attitudes Toward Government in Contemporary Democracies. American Journal of Political Science, 47(1), pp. 91-109. Bauhr, M., Charron, N. & Wängnerud, L., 2018. Exclusion or interests? Why females in elected office reduce petty and grand corruption: Exclusion or Interests. European Journal of Political Research, 58(3), pp. 1043-1065. Bauhr, M. & Grimes, M., 2014. Indignation or Resignation: The Implications of Transparency for Societal Accountability. Governance: An International Journal of Policy, Administration, and Institutions, 27(2), pp. 291-320. Bekaj, A., 2020. Escaping the slide towards authoritarianism in new democracies alerts from the Western Balkans, Stockholm: International IDEA. Bentzen, S. S., 2014. Perceptions of Corruption & External Political Efficacy: Regime Stability & Democratic Legitimacy, Oxford, Mississippi: Department of Political Science, The University of Mississippi. Bjelić, P., 2015. Building Competitiveness and Increasing Trade Potential in the Western Balkans: Policy Making in Preparing for European Integration . In: M. Thomas & V. Bojicic-Dzelilovic, eds. Public Policy Making in the Western Balkans: Case Studies of Selected Economic and Social Policy Reforms. Springer. Bratton, M. & Chang, E. C. C., 2006. State Building and Democratization in Sub-Saharan Africa: Forwards, Backwards, or Together?. Comparative Political Studies, 39(9), pp. 1059-1083. Burazer, N., Hackaj, K., Shehaj, A. & Stefanovski, I., 2017. Democracy in Progress: Shadow Report on Political Copenhagen criteria in Western Balkans EU Candidate States, Belgrade: Centar Savremene Politike. Campbell, A., Gurin, G. & Miller, W. E., 1954. The Voter Decides. Carlin, R. E., Love, G. J. & Martı´nez-Gallardo, C., 2015. Cushioning the Fall: Scandals, Economic Conditions, and Executive Approval. Political Behavior, 37(1), pp. 109-133. Catterburg, G. & Moreno, A., 2005. The Individual Bases of Political Trust: Trends in New and Established Democracies. International Journal of Public Opinion Research, 18(1), pp. 408-43. Chang, E. C. C. & Chu, Y., 2006. Corruption and Trust: Exceptionalism in Asian Democracies?. The Journal of Politics, 68(2), pp. 259-271. Gentiana Gola 37 Charron, N., 2016. Do corruption measures have a perception problem? Assessing the relationship between experiences and perceptions of corruption among citizens and experts. European Political Science Review, 8(1), pp. 147-171. Charron, N. & Rothstein, B., 2016. Does education lead to higher generalized trust? The importance of quality of government. International Journal of Educational Development, 50(1), p. 59–73. Cho, W. & Kirwin, M. F., 2007. A vicious circle of corruption and mistrust in institutions in sub- Saharan Africa: A micro-level analysis, Working Paper No. 71: Afrobarometer. Craig, S. C., Richard, N. G. & Glenn, S. E., 1990. Political efficacy and trust: A report on the NES Pilot Study items. Political Behavior, 12(3), pp. 289-314. Cronbach, L. J. & Shavelson, R. J., 2004. My current thoughts on coefficient alpha and successor procedures. Educational and Psychological Measurement, 64(3), pp. 391-418. Dahlberg, S. & Holmberg, S., 2014. The Importance of Institutional Trust for Regime Support, Gothenburg: The Quality of Government Institute, Department of Political Science, University of Gothenburg. Dahlberg, S. & Solevid, M., 2016. Does corruption suppress voter turnout?. Journal of Elections, Public Opinion and Parties, 26(4), pp. 489-510. Dahl, R. A., 1971. Polyarchy: Participation and Opposition. 1st ed. New Haven: Yale University Press. Daur, V., 2018. Individual Heterogeneity in Corruption Voting: How External Political Efficacy Shapes Electoral Responses to Corruption, Gothenburg: Department of Political Science, Gothenburg University. Della Porta, D., 2000. Social Capital, Beliefs in Government, and Political Corruption. In: S. J. Pharr & R. D. Putnam, eds. Disaffected Democracies: What's Troubling the Trilateral Countries?. Princeton: Princeton University Press, pp. 202-228. Dobranja, D., 2015. Investment Policy Design, Process and Implementation in Kosovo. In: M. Thomas & V. Bojicic-Dzelilovic, eds. Public Policy Making in the Western Balkans: Case Studies of Selected Economic and Social Policy Reforms. Springer. Eek, D. & Rothstein, B., 2005. Political Corruption and Social Trust: An Experimental Approach, Gothenburg: Quality of Government Institute. European Commission, 2019. Eurobarometer 11/2019. [Online] Available at: https://ec.europa.eu/commfrontoffice/publicopinion/index.cfm/Chart/index [Accessed 5 January 2021]. Ezrow, L. & Xezonakis, G., 2011. Citizen Satisfaction With Democracy and Parties’ Policy Offerings. Party Politics, 44(9), pp. 1152-1178. Gentiana Gola 38 Fink‐Hafner, D. & Thomas, C. S., 2019. The Balkan experience: Explanations of public affairs and interest group activity. Journal of Public Affairs, 19(2), p. e1919. Freedom House [FH], 2020. Freedom in the World, Freedom House. Golubović, N., Džunić, M. & Golubović, S., 2015. Trust in Political Institutions in Western Balkan Countries. Law and Politics, 13(1), pp. 13-21. Harring, N., Jagers, S. C. & Löfgren, Å., 2020. COVID-19: Large-scale collective action, government intervention, and the importance of trust. World Development, 138(1), p. 105-236. Hayes, B. C. & Bean, C. S., 1993. Political efficacy: A comparative study of the United States, West Germany, Great Britain and Australia. European Journal of Political Research, 23(3), pp. 261-280. Helliwell, J. F., 2002. How's Life? Combining Individual and National Variables to Explain Subjective Well-Being. Economic Modelling, 20(2), pp. 331-360. Holmberg, S., 1997. Dynamic opinion representation. Scandinavian Political Studies, 20(1), pp. 265- 283. Holmberg, S., 2008. Effects of Trust in Parliament, University of Gothenburg: QoG Working Paper Series 2008:13, The Quality of Government Institute, Department of Political Science. Iyengar, S., 1980. Subjective Political Efficacy as a Measure of Diffuse Support. The Public Opinion Quarterly, 44(2), pp. 249-256. Jaċimoviċ-Vojinoviċ, D., 2015. Montenegro: Enhancing Competitiveness and Accelerating European Union Integration – The Role of Higher Education. In: M. Thomas & V. Bojicic-Dzelilovic, eds. Public Policy Making in the Western Balkans: Case Studies of Selected Economic and Social Policy Reforms. Springer. Johnston, M., 1986. The Political Consequences of Corruption: A Reassessment. Comparative Politics, 18(4), pp. 459-477. Kadia, A., 2020. Corruption and Economic Growth in the Balkan Countries. European Journal of Economics and Business Studies, 3(3), pp. 88-97. Karp, J. A. & Banducci, S. A., 2008. Political Efficacy and Participation in Twenty-Seven Democracies: How Electoral Systems Shape Political Behaviour. British Journal of Political Science, 38(2), pp. 311-334. Keefer, P., 2007. Clientelism, Credibility, and the Policy Choices of Young Democracies. American Journal of Political Science, 51(4), pp. 804-821. Kelmanson, B. et al., 2019. Explaining the Shadow Economy in Europe: Size, Causes and Policy Options , IMF Working paper WP/19/278: International Monetary Fund . Kitschelt, H., 2000. Linkages between citizens and politicians in democratic polities. Comparative Political Studies, 33(6/7), pp. 845-879. Gentiana Gola 39 Kmezić, M., 2020. Rule of law and democracy in the Western Balkans: addressing the gap between policies and practice. Southeast European and Black Sea Studies, 20(1), pp. 183-198. Kmezić, M. & Bieber, F., 2007. The Crisis of Democracy in the Western Balkans: An Anatomy of Stabilitocracy and the Limits of EU Democracy Promotion. 1st ed. Balkans in Europe Policy Advisory Group. Knack, S. & Keefer, P., 1997. Does Social Capital Have an Economic Payoff? A Cross-Country Investigation. The Quarterly Journal of Economics, 112(3), pp. 1251-1288. Konstantinidis, I. & Xezonakis, G., 2013. Sources of tolerance towards corrupted politicians in Greece: the role of trade offs and individual benefits. Crime Law Soc Change, 60(1), p. 549–563. Kostadinova, T., 2009. Abstain or Rebel: Corruption Perceptions and Voting in East European Elections. Politics & Policy, 37(4), pp. 1555-5623. Krause, S. & Méndez, F., 2009. Corruption and elections: An empirical study for a cross-section of countries. Economics and Politics, 21(2), pp. 179-200. Kubbe, I., 2013. Corruption and trust: a model design. In: T. Debiel, A. Gawrich & H. Neimann, eds. (Dys-)Functionalities of Corruption: Comparative Perspectives and Methodological Pluralism. Special Issue 3 ed. Würzburg: Comparative Governance and Politics, Springer, p. 117–135. Laderchi, C. R. & Savastano, S., 2013. Poverty and Exclusion in the Western Balkans: New Directions in Measurement and Policy. 1st ed. New York: Springer. Linde, J., 2009. Why Feed the Hand that Bites You? Quality of Government and System Support in the Post-Communist EU Member States, Gothenburg: The Quality of Government Institute, Department of Political Science, University of Gothenburg. Linde, J. & Erlingsson, G. Ó., 2012. The Eroding Effect of Corruption on System Support in Sweden. Governance: An International Journal of Policy, Administration, and Institutions, 26(4), p. 585–603. Listhaug, O. & Wiberg, M., 1998. Confidence in Political and Private Institutions. In: H. Klingemann & D. Fuchs, eds. Citizens and the State. Oxford: Oxford Scholarship , p. Online: November 2003. Marien, S., Hooghe, M. & Quintelier, E., 2010. Inequalities in Non-institutionalised Forms of Political Participation: A Multi-level Analysis of 25 countries. Political Studies, Volume 58, pp. 187-213. McMann, K. M., Seim, B., Teorell, J. & Lindberg, S. I., 2017. Democracy and Corruption: A Global Time-Series Analysis with V-Dem Data, Gothenburg: The Varities of Democracy Institute. Milbrath, L. W., 1965. Political Participation: How and Why Do People Get Involved in Politics?. 1st ed. Chicago: Rand McNally College Publishing Company. Mishler, W. & Rose, R., 2001. What are the origins of political trust? Testing institutional and cultural theories in post-communist societies. Comparative Political Studies, 34(1), pp. 30-62. Gentiana Gola 40 Moran, J., 1999. Patterns of Corruption and Development in East Asia. Third World Quarterly, 20(3), pp. 569-587. Mujanović, J., 2018. The EU and the Crisis of Democracy in the Balkans. Political Insight, 9(1). Norris, P., 1999. Critical Citizens: Global Support for Democratic Governance. 1st ed. New York: Oxford University Press. Nunnally, J. C., 1978. Psychometric theory. 2nd ed. New York: McGraw-Hill. O'Brien, M., Stapenhurst, R. & Johnston, N., 2008. Parliaments as Peacebuilders in Conflict-Affected Countries. New York, World Bank institute. Olsson, S. A., 2014. Corruption and Political Participation: A Multilevel Analysis, Gothenburg: The Quality of Government Institute, Department of Political Science, University of Gothenburg. Organisation for Economic Co-operation and Development [OECD], 2017. Trust and Public Policy: How Better Governance Can Help Rebuild Public Trust, Paris: OECD Publishing. Organisation for Economic Co-operation and Development [OECD], 2020. Government at a Glance: Western Balkans, Paris: OECD. Oscarsson, H., 2007. A Matter of Fact? Knowledge Effects on the Vote in Swedish General Elections, 1985–2002. Nordic Political Science Association, 30(3), pp. 301-322. Pellegata, A. & Memoli, V., 2016. Can Corruption Erode Confidence in Political Institutions Among European Countries? Comparing the Effects of Different Measures of Perceived Corruption. Social Indicators Research, 128(1), p. 391–412. Persson, A., Rothstein, B. & Teorell, J., 2013. Why Anticorruption Reforms Fail—Systemic Corruption as a Collective Action Problem. Governance, 26(3), pp. 449-471. Peters, J. & Welch, S., 1980. The effects of charges of corruption on voting behavior in congressional elections. American Political Science Review, 74(3), pp. 697-708. Pharr, S. & Putnam, R., 2000. Disaffected Democracies. 1st ed. Princeton: Princeton University Press. Putnam, R. D., 1993. Making Democracy Work: Civic Traditions in Modern Italy. 1st ed. Princeton: Princeton University Press. Regional Cooperation Council [RCC], 2021. About. [Online] Available at: https://www.rcc.int/balkanbarometer/about [Accessed 10 February 2021]. Regional Cooperation Council [RCC], 2020. Balkan Barometer 2020: Public Opinion Analytical Report, Sarajevo: Regional Cooperation Council. Rolef, S. H., 2006. Public Trust in Parliament – A Comparative Study. Jerusalem, Knesset. Gentiana Gola 41 Rose, C. R., Mishler, W. & Haerpfer, C. W., 1998. Democracy and its Alternatives: Understanding Post-Communist Societies. 1st ed. Baltimore, MD, USA: Johns Hopkins University Press. Rothstein, B. & Eek, D., 2009. Political Corruption and Social Trust: An Experimental Approach. Rationality and Society, 21(1), p. 81–112. Rundquist, B. S., Strom, G. S. & Peters, J. G., 1977. Corrupt Politicians and Their Electoral Support: Some Experimental Observations. American Political Science Review, 71(3), pp. 954-963. Schäfer, A., 2012. Consequences of social inequality for democracy in Western Europe. Zeitschrift für Vergleichende Politikwissenschaft, 6(2), p. 23–45. Seligson, M. A., 2002. The Impact of Corruption on Regime Legitimacy: A Comparative Study of Four Latin American Countries. The Journal of Politics, 64(2), pp. 408-433. Southeast European Leadership for Development and Integrity [SELDI], 2020. Hidden economy in the Western Balkans 2020: Trends and policy options, Policy brief no. 11: SELDI. Stewart, M. C., Kornberg, A., Clarke, H. D. & Acock, A., 1992. Arenas and Attitudes: A Note on Political Efficacy in a Federal System. The Journal of Politics, 54(1), pp. 179-196. Stockemer, D., 2013. Corruption and Turnout in Presidential Elections: A Macro‐Level Quantitative Analysis. Politics & Policy, 41(2), pp. 189-212. Stockemer, D. & Sundström, A., 2013. Corruption and citizens’ satisfaction with democracy in Europe: what is the empirical linkage?. In: T. Debiel & A. Gawrich, eds. (Dys-)Functionalities of Corruption: Comparative Perspectives and Methodological Pluralism. Wiesbaden: Springer VS, pp. 137-158. Tavits, M., 2010. Why Do People Engage in Corruption? The Case of Estonia. Social Forces, 88(3), pp. 1257-1280. Taylor, A. J., 2013. The European Union and State Building in the Western Balkans. Politics and Governance, 1(2), p. 183–195. The Economist, 2010. Calm now, a storm to come? Early indications from Kosovo's election. [Online] Available at: https://www.economist.com/eastern-approaches/2010/12/13/calm-now-a-storm-to-come [Accessed 5 January 2021]. Transparency International [TI], 2020. Corruption Perceptions Index (2015-2020). Transparency International. Ulbig, S. G., 2008. Voice Is Not Enough: The Importance of Influence in Political Trust and Policy Assessments. The Public Opinion Quarterly, 72(3), pp. 523-539. Uslaner, E., 2002. The Moral Foundation of Trust. 1st ed. University of Maryland, College Park; Institute for Corruption Studies: Cambridge University Press. Gentiana Gola 42 Vera, S. B., 2018. Accepting or Resisting Corruption? Responses to Corruption and Competence in Highly Corrupt Environments, Gothenburg: Working Paper Series 2018:10, The Quality of Government Insitute, Department of Political Science, University of Gothenburg. Verba, S., Schlozman, K. L. & Brady, H. E., 1995. Voice and equality. Civic Voluntarism in American Politics. 1st ed. Cambridge: Harvard University Press. Villoria, M., Ryzin, G. G. V. & Lavena, C. F., 2012. Social and Political Consequences of Administrative Corruption: A Study of Public Perceptions in Spain. Public Administration Review, 73(1), pp. 85-94. Wagner, A. F., Schneider, F. & Halla, M., 2009. The quality of institutions and satisfaction with democracy in Western Europe: A panel analysis. European Journal of Political Economy, 25(1), p. 30–41. Warren, M. E., 2004. What Does Corruption Mean in a Democracy?. American Journal of Political Science, 48(2), pp. 328-343. World Bank [WB], 2020. Worldwide Governance Indicators (WGI) Project, Washington DC, USA: The World Bank. Xezonakis, G., 2012. Corruption Voting and Political Context: Testing the Micro Mechanisms, Gothenburg: The Quality of Government Institute, Department of Political Science, University of Gothenburg. Zechmeister, E. J. & Zizumbo-Colunga, D., 2013. The Varying Political Toll of Concerns About Corruption in Good Versus Bad Economic Times. Comparative Political Studies, 46(10), p. 1190– 1218. Gentiana Gola 43 9. Appendix Table 1 – Descriptive Statistics Variable Obs Mean Std.Dev Min Max Corruption Index 5,415 6.211819 1.488166 2 8 Political Trust index 5,160 10.48953 3.977243 5 20 Information of Government 5,263 2.523276 0.8568035 1 4 Ability to Scrutinize Government 5,492 2.399126 0.9368239 1 4 Satisfaction with Economy Index 4,051 5.570723 2.16313 2 10 Age 6,020 3.381063 1.640463 1 6 Education 6,020 2.178073 0.6775425 1 4 Social status 5,941 1.823767 0.4941551 1 3 Work Status 6,020 3.940033 2.042225 1 6 Satisfaction Index 6,020 4.350997 1.486012 1 6 Gentiana Gola 44 Table 2 – Pairwise Correlations Corruption Index Parliament Corruption Political Trust Index Government Information Scrutinize Government Satisfaction with Economy Index Corruption Index 1.0000 Political Trust Index -0.4827 -0.4723 1.0000 Government Information -0.2336 -0.2025 0.4501 1.0000 Scrutinize Government -0.2686 -0.2380 0.5343 0.2919 1.0000 Satisfaction with Economy Index -0.3115 -0.3044 0.5629 0.4294 0.3527 1.0000 Gentiana Gola 45 Table 3 – Robustness Check: Trust in Parliament as a Dependent Variable Model 1 Model 2 Trust in Parliament Trust in Parliament b/se b/se Corruption Index -0.254*** -0.132*** (0.01) (0.03) Satisfaction with Economy Index 0.242*** (0.03) Corruption Index*Satisfaction with Economy Index -0.011** (0.00) Satisfaction Index 0.007 (0.01) Age 0.026** 0.020 (0.01) (0.01) Gender -0.001 -0.004 (0.02) (0.02) Education: High School 0.052 0.076* (0.03) (0.04) Education: College / University 0.095* 0.091* (0.04) (0.04) Education: Master’s Degree / Doctorate 0.074 0.118 (0.08) (0.09) Social status: Average 0.209*** -0.015 (0.03) (0.03) Social status: Above Average 0.339*** 0.002 (0.06) (0.07) Work status: Retired 0.045 -0.014 (0.04) (0.05) Work status: Housewife 0.082 0.042 (0.05) (0.06) Work status: Student/Pupil 0.132** 0.075 (0.05) (0.06) Work status: Self-Employed 0.051 -0.046 (0.04) (0.05) 6.WorkStatus: Employed 0.067* 0.005 _cons 3.183*** 1.748*** (0.08) (0.20) r2 0.259 0.433 N 5269 3543 Notes: * p < 0.1; ** p < 0.05; *** p < 0.01. Standard errors are adjusted for heteroscedasticity by using robust standard errors. Controlled for country dummies (not displayed). Base levels for categorical variables are: Education: No education / primary school; Social Status: Below average; Work Status: Unemployed. Gentiana Gola 46 Table 4 - Multicollinearity Test for Model 1 Variable VIF 1/VIF Corruption Index 1.12 0.892663 Satisfaction with Economy Index 1.30 0.770303 Age 2.24 0.445617 Gender 1.07 0.936377 Education: High School 2.73 0.365790 Education: College / University 2.89 0.346049 Education: Master’s Degree / Doctorate 1.26 0.793633 Social Status: Average 1.42 0.703706 Social Status: Above Average 1.33 0.752517 Work Status: Retired 2.58 0.387111 Work Status: Housewife 1.39 0.717031 Work Status: Student/Pupil 1.46 0.682710 Work Status: Self-Employed 1.41 0.711411 Work Status: Employed 2.04 0.489407 Mean VIF 1.73 Test 1 – Results of a Heteroskedasticity Test for Model 1 Breusch-Pagan / Cook-Weisberg test for heteroskedasticity Ho: Constant variance Variables: fitted values of Political Trust Index chi2(1) = 0.07 Prob > chi2 = 0.7889 MODEL 1 Gentiana Gola 47 Graph 1 – Normality of residuals / Kernel density estimate for Model 1 Graph 2 – Residual-versus-fitted plot for Model 1 MODEL 1 Gentiana Gola 48 Table 5 - Multicollinearity Test for Model 2 Variable VIF 1/VIF Corruption Index 1.12 0.896105 Satisfaction with Economy Index 1.26 0.794336 Age 2.18 0.458567 Gender 1.07 0.937636 Education: High School 2.96 0.337560 Education: College / University 3.13 0.319928 Education: Master’s Degree / Doctorate 1.29 0.777342 Social Status: Average 1.40 0.713108 Social Status: Above Average 1.34 0.744155 Work Status: Retired 2.53 0.394961 Work Status: Housewife 1.42 0.706665 Work Status: Student/Pupil 1.50 0.665086 Work Status: Self-Employed 1.43 0.699223 Work Status: Employed 2.11 0.474167 Mean VIF 1.77 Test 2 - Heteroskedasticity for Model 2 Breusch-Pagan / Cook-Weisberg test for heteroskedasticity Ho: Constant variance Variables: fitted values of Internal Efficacy chi2(1) = 68.97 Prob > chi2 = 0.0000 MODEL 2 Gentiana Gola 49 Graph 3 – Normality of residuals / Kernel density estimate for Model 2 Graph 4 – Residual-versus-fitted plot for Model 2 MODEL 2 Gentiana Gola 50 Table 6 - Multicollinearity Test for Model 3 Variable VIF 1/VIF Corruption Index 1.12 0.889614 Satisfaction with Economy Index 1.30 0.767267 Age 2.25 0.445223 Gender 1.07 0.935953 Education: High School 2.74 0.364751 Education: College / University 2.90 0.345052 Education: Master’s Degree / Doctorate 1.26 0.796252 Social Status: Average 1.42 0.704051 Social Status: Above Average 1.33 0.749446 Work Status: Retired 2.61 0.382809 Work Status: Housewife 1.41 0.709262 Work Status: Student/Pupil 1.49 0.669115 Work Status: Self-Employed 1.42 0.701862 Work Status: Employed 2.09 0.478243 Mean VIF 1.74 Test 3 - Heteroskedasticity for Model 3 Breusch-Pagan / Cook-Weisberg test for heteroskedasticity Ho: Constant variance Variables: fitted values of External Efficacy chi2(1) = 38.81 Prob > chi2 = 0.0000 MODEL 3 Gentiana Gola 51 Graph 5 – Normality of residuals / Kernel density estimate for Model 3 Graph 6 – Residual-versus-fitted plot for Model 3 MODEL 3 Gentiana Gola 52 Graph 7 – Normality of residuals / Kernel density estimate for Model 4 Graph 8 – Residual-versus-fitted plot for Model 4 Test 4 - Heteroskedasticity for Model 4 Breusch-Pagan / Cook-Weisberg test for heteroskedasticity Ho: Constant variance Variables: fitted values of Political Trust Index chi2(1) = 0.14 Prob > chi2 = 0.7061 MODEL 4 Gentiana Gola 53 Graph 9 – Normality of residuals / Kernel density estimate for Model 5 Graph 10 – Residual-versus-fitted plot for Model 5 Test 5 - Heteroskedasticity for Model 5 Breusch-Pagan / Cook-Weisberg test for heteroskedasticity Ho: Constant variance Variables: fitted values of Internal Efficacy chi2(1) = 71.91 Prob > chi2 = 0.0000 c MODEL 5 Gentiana Gola 54 Graph 11 – Normality of residuals / Kernel density estimate for Model 6 Graph 12 – Residual-versus-fitted plot for Model 6 Test 6 - Heteroskedasticity for Model 6 Breusch-Pagan / Cook-Weisberg test for heteroskedasticity Ho: Constant variance Variables: fitted values of External Efficacy chi2(1) = 40.83 Prob > chi2 = 0.0000 MODEL 6 Gentiana Gola 55 Graph 13 – Political Trust Index by country Political Trust Index Gentiana Gola 56 Graph 14 – Internal Efficacy by country Graph 15 – External Efficacy by country Information of Government Ability to Scrutinize the Government Gentiana Gola 57 Graph 16 – Corruption Index by country Graph 17 – Satisfaction with Economy Index by country Corruption Index Satisfaction with the Economy