MSc Political Science International Relations
Master Thesis
The Effects of Corruption on Support for an Authoritarian Leader
by
Martin Winroither ID: 13884077June 2022
Supervisor: Research Project:
Prof. Dr. Tom van der Meer A global crisis of democracies?
Change and continuity in 21st- century politics
Department of Political Science
Abstract
In the last couple of years, the term “democratic backsliding” received much scholarly attraction. It has been rather controversial, and scholars are debating whether this phenomenon is happening on a global basis and if widespread support for alternatives to democracies exists also in consolidated democracies. Over the last decades, in many consolidated, liberal democracies, political parties and candidates, who try to concentrate power in the executive and blame a corrupt political class for most problems, have had strong electoral success.
Surprisingly, despite widespread knowledge about the detrimental implications of corruption, there is hardly any research regarding the link between corruption and support for an authoritarian leader. In this paper, I set to shed light on this topic with an empirical quantitative analysis, to see if corruption leads citizens to support an authoritarian leader in consolidated democracies, namely members of the European Union. I conduct a regression analysis with individual and country-level variables, setting out to determine if levels of corruption are influencing citizens’ desire for an authoritarian leader and further show what other country- and individual-level factors are relevant. This is accompanied by a literature review on the possible mechanisms behind. The findings show that corruption influences a population to be more likely to support an authoritarian leader. Higher levels of corruption lead to stronger support.
Furthermore, levels of inequality matter as well. Citizens are more likely to desire a strong leader if levels of inequality are high. On an individual level, education, income, and age are all significant.
Keywords: Corruption, Authoritarianism, Strong Leader, Support for Authoritarianism, Support for Strong Leader, European Union, EU, Europe, Democratic Backsliding, Democratic Deconsolidation,
Table of contents
1. Introduction ... 1
2. Literature review and theoretical framework ... 5
2.1. Defining corruption ... 5
2.2. Theoretical framework and hypothesis: Why corruption leads to support for authoritarianism ... 6
2.2.1. Effects of Corruption ... 7
2.2.2. Support for authoritarian leadership ... 11
3. Data and Methods ... 16
3.1. Dependent variable ... 16
3.2. Independent variables ... 17
3.3. Methods ... 19
4. Results and analysis ... 20
5. Conclusion ... 27
References ... 29
Declaration of Authorship ... 35
Appendix ... 36
Overview of tables
Table 1: Descriptives for Support for an Authoritarian Leader ... 21Table 2: Control of Corruption Indicator (Worldwide Governance Indicators) ... 22
Table 3: Main results of the regression models ... 24
Table 4: Split up of country-level variables ... 26
1
1. Introduction
On July 25th, 2021, Kais Saied, Tunisia’s elected president “invoked emergency powers under the 2014 Tunisian Constitution to oust Prime Minister Hicham Mechichi, assume control over the government, shutter Parliament, and begin his rule of the country by decree” (Global Anticorruption Blog, 2021). Despite his actions being unconstitutional and described by many as a coup, only two months after his power grab, thousands of his supporters have rallied in the capital to back up his suspension of parliament and his promises to change the political system.
“We ask the president to dissolve parliament and hold accountable those who made the people suffer for a decade”, one of the demonstrators said (The Guardian, 2021). Despite his authoritarian and antidemocratic actions, he still had a large number of supporters.
One of the reasons why he claimed his extraordinary consolidation of power was necessary, was to end rampant corruption. He stated that the country is ruled “by two regimes, an apparent regime, that of the institutions, and a real regime, that of the mafia,” and he has vowed not to
“engage in dialogue with ‘thieves’” (Global Anticorruption Blog, 2021). While many outsiders fear that Tunisia might backslide into an authoritarian regime again, many of his supporters believe in his ideals and were mostly concerned with the country’s corrupt political class. They either did not share skeptical opinions, or they saw him as a necessary evil to cleanse the country from corrupt politicians.
Interestingly enough, looking at recent history this move is far from unique. In their book “How democracies die” Steven Lewitsky and Daniel Ziblatt (2018) give illustrative, historic examples of how democracies can slowly erode in small steps. Taking a look at Venezuela for example, Hugo Chavez, was a political outsider “who railed against what he cast as a corrupt governing elite, promising to build a more ‘authentic’ democracy that used the country’s vast oil wealth to improve the lives of the poor” (Lewitsky & Ziblatt, 2018, p. 10). While first democratically elected in 1998, he started acting authoritarian in 2003 holding off a referendum, which would have forced him to leave office. He further increased his repressive regime in 2006, closing down a major television station, and exiling and arresting opposition politicians, media figures, and judges. Furthermore, he eliminated the presidential term limits. In 2017 his successor Nicolas Maduro usurped Congress and Venezuela was finally recognized as an autocracy (Lewitsky & Ziblatt, 2018, p. 11).
2
Another poignant example can be seen in Peru. In 1990 Alberto Fujimori, a university rector and complete political outsider at that time created his own party and ran as a presidential candidate. Despite being “the underdog”, he unexpectedly won against his opponent Vargas Llosa, who was backed by the media, business leaders, and the establishment. However,
“Peruvians were disgusted with the established parties. In protest, many of them turned to the political nobody whose campaign slogan was ‘A President Like You’” (Lewitsky & Ziblatt, 2018, p. 64) Once in office, instead of negotiating with members of congress he started calling them “unproductive charlatans” and lashed out against judges as well. He called the political elite corrupt and described them as an oligarchy that was ruining the country. Soon he began to rule by executive decree to bypass congress, despite many decrees being declared unconstitutional by the courts. The conflict escalated and on the 25th of April 1992, Fujimori dissolved Congress and the constitution (Lewitsky & Ziblatt, 2018, p. 65). Lastly, in Hungary, the populist party Fidesz was elected on explicitly anti-corruption platforms. However, since the party has been in government, the country has been more corrupt than ever, and liberal democracy has steadily declined (Rohac et al., 2017, p. 393).
All these examples above have some points in common. On the one hand, in all cases, the countries already faced existing problems when a political outsider stepped into office. On the other hand and more importantly, in all three countries the public opinion about the ruling political class was overwhelmingly negative. It was seen as rigid, unproductive, and more importantly as corrupt. As one woman in Chávez’s home state of Barinas put it on election night, “Democracy is infected. And Chávez is the only antibiotic we have” (Lewitsky & Ziblatt, 2018, p. 10).
They began to see the political outsider as the only solution and supported him despite authoritarian tendencies until democracy was slowly completely toppled. Despite democratic backsliding and authoritarian actions, the population of a country seems to be more concerned about corruption and a corrupt political class. The idea that someone had to step into office to get rid of the corrupt politicians became stronger than the fear of democratic backsliding and the population began to excuse undemocratic actions. Of course, it is important to mention that most of these democracies mentioned above were not being considered consolidated democracies. However, while consolidated democracies are usually more resilient, it is possible that the same effect might occur in them as well.
3
In the last couple of years, the question, whether there is a crisis of democracy has received much scholarly attraction. While scholars remain divided on the nature and the depth of this topic, quantitative indices show declines in levels of global democracies and in the number of countries, which make democratic gains (Cianetti & Hanley, 2021). The terms “democratic backsliding” or “democratic deconsolidation”, meaning the gradual erosion of democratic institutions, as well as constitutional safeguards, are much discussed in academic literature.
While numerous studies have been conducted, whether this phenomenon is happening on a global basis, or solely in specific counties, this paper will focus on citizens’ views toward democracy and authoritarianism in Europe. Lewitsky & Ziblatt (2018) show that norms, such as mutual toleration are crucial for a well-functioning democracy. They call them “soft guardrails”, which place checks and balances on leaders who show authoritarian tendencies (Levitsky & Ziblatt, 2018). However, if citizens are open to nondemocratic approaches,
“would-be autocrats may find opportunities to transgress the unwritten rules that help to hold democracies together” (Wike & Fetterolf, 2018, p. 146).
In their influential yet controversial article, Foa & Mounk (2016) argue that citizens in consolidated democracies are turning away from democracy and are willing to express support for authoritarian systems. Analyzing the World Value Survey, they claim that even in consolidated democracies citizens express increased support for authoritarian alternatives compared to the past and that millennials are withdrawing from democracy (Foa & Mounk, 2016). While their work sparked a lot of controversy and critiques regarding their empirical analysis and their claim toward millennials (see Voeten 2017; Zilinsky, 2019; Norris, 2017), it is still interesting that even in consolidated democracies, relatively high rates of support for anti-democratic alternatives do exist. Especially support for authoritarian leadership with “a strong leader”, that is not constrained by parliament and elections, seems to be particularly interesting.
However, while other scholars have conducted further research since then, most academic literature focuses on individual-level attributes on support for a strong leader and leaves out country-level indicators. Literature, which analyzes country-level indicators and its support for nondemocratic alternatives, rarely exists. Even less literature exists regarding corruption and its effect on support for a strong leader without democratic constraints. To my knowledge, virtually no existing academic work analyzes the relationship between corruption and support for authoritarianism in democracies. This is particularly surprising as over the last decades, in many consolidated, liberal democracies, political parties and candidates, who try to concentrate
4
power in the executive and blame an alleged corrupt political class for most problems, have had strong electoral success (Foa & Mounk, 2017, p. 8). It is often the common rhetoric of these populist leaders, to discredit opponents by accusing them of corruption and position themselves against the “corrupt elite” (Rummens, 2017). It would be expected that their rhetoric is more successful in countries where levels of corruption are higher and therefore that in countries where there are higher levels of corruption, citizens tend to favor one strong, authoritarian leader, who can “cure” the corrupt system with strong actions. The rise of some of these populist parties, who position themselves against a corrupt elite and typically lash out against the media, shows that even in consolidated democracies authoritarian support exists. And not coincidentally, it is populistic parties which are perceived as a threat to liberal democracy.
While the examples might be just illustrative, they might provide some insight that is worth researching.
This is the research gap that this paper is trying to fill. In the next section, I will systematically analyze and define the term “corruption”, followed by the theoretical framework, of why corruption would lead to support for authoritarianism? We will hereby highlight the numerous pieces of literature about corruption and the effects it has on society and why this could lead to the desire for a strong leader and subsequently to support for authoritarianism.
5
2. Literature review and theoretical framework
This section contains a literature review about the phenomenon of corruption, the numerous effects and implications it has, and the theoretical framework for the following analysis. First, we will define corruption and show literature on its implications. We, then move on to the theoretical framework and why corruption might lead to support for authoritarianism. Finally, I deduct the hypothesis which I will test in chapter three, followed by a conclusion.
2.1. Defining corruption
The term “corruption” itself is highly value-laden and the reason for much discussion in the academic field. To begin with, despite the fact that there exists a great amount of research on the phenomenon of corruption there is no consensus in academia for one clear definition. What citizens label as corruption differs across space, time, and social groups (Wickberg, 2021, p.
82). In other words, there is no universal definition of corruption. “There remains a striking lack of scholarly agreement over even the most basic questions about corruption, [such as] the very definition of ‘corruption’ as a concept” (Heywood, 2015, p. 1-2 as cited in Wickberg, 2021, p. 86). It is a category label and the term itself is highly emotional and causes a lot of anxiety in societies, due to the fact that it undermines the values and rules of democracy.
Historically, the term corruption comes from the Latin word “corrumpere” which means to destroy or ruin something and was used to describe the decay of an original pure physical, moral, or social state. Over time, and with the upcoming of modern belief systems, which started to differentiate between public and private life, the meaning started to fluctuate, and the term became generally redefined as the misuse of public power for private gain (Wickberg, 2021, p. 82-85). Currently, by the most prominent definition, corruption is usually defined as the “abuse of public office (or power, or entrusted authority) for private gain” (You, 2018, p.
477).
In addition, citizens have different conceptions of corruption as well as different levels of tolerance for corruption. Certain forms of corruption, such as nepotism, or favoritism are considered intolerable by some people, while others only view embezzlement or bribery as a corrupt practice. These different understandings vary within the culture. Another problem is
6
that, even if there would be one single understanding of corruption in a society, it is impossible for citizens to know the actual levels of corruption. Corrupt acts are often being carried out in the shadows and presumably, only a small amount is ever revealed. People therefore either under- or overestimate the actual levels of corruption (You, 2018, p. 475).
As a matter of fact, measuring corruption proves to be rather difficult, and every study which tries to measure corruption faces challenges. Objective measures, for example, the conviction rate or the news coverage about corruption, might not show actual levels of corruption but rather the freedom of the press or the effectiveness of the judicial system of a country. Measurements of perceived corruption are more reliable, yet face the problem that, as discussed above, perceptions of what is being considered corrupt might differ across countries and cultures (You, 2018, p. 479-480). Despite the fact that corruption has a rather fluid definition there are methods to measure corruption. The most prominent are the Corruption Perception Index by Transparency International and the World Bank’s Control of Corruption indicator. Both are composite indexes of corruption perceived by experts and businesspeople, aggregated from multiple sources (You, 2018, p. 480). Scholars agree that both indicators show the perceptions of corruption of the economic elite and the international private sector. Within the European Union, both indicators correlated almost perfectly in 2009 (Pellegata & Memoli, 2016, p. 396).
Academic critique regarding both indicators exists, as it only showcases experts’ opinions of corruption, who are more likely to overreport and reflect corruption on a higher level than ordinary people. However, there is currently no better way to measure and study it on a large scale. In addition, studies have shown that there is a strong correlation between elite and mass perceptions of corruption, for example measured by Transparency International’s Global Corruption Barometer. In other words, there is a high correlation between petty and grand corruption. As Uslaner (2017) argues: “We may not be able to specify exactly what corruption is, but people (and elites) do seem to know where their country stands” (p. 303).
2.2. Theoretical framework and hypothesis: Why corruption leads to support for authoritarianism
In this section, we discuss possible mechanisms why corruption might affect support for a strong, authoritarian leader, based on existing academic literature. As I will now show in detail it has an impact on the economy, as well as societal developments, and has been researched in great lengths (Dimant & Tosato, 2018).
7 2.2.1. Effects of Corruption
Generally speaking, there are two types of arguments on the effects of corruption. The more dominant approach to corruption is the so-called “sand in the wheels” hypothesis. In this theory, corruption has an overall negative effect on a country specifically on the economy and political trust. It puts “sand” in the wheel, which makes economic development and political transitions difficult (Aidt, 2009, p. 271).
The second strand of thought is the so-called “grease the wheel hypothesis”. This idea was coined by Leys (1965) and Leff (1964). They argued that corruption can have beneficial trades which would have not taken place otherwise. This means that corruption can actually promote efficiency as it allows actors of the private sector, to correct government failures. (Aidt, 2009, p. 272). In other words,
“[…] corruption may be beneficial in a second-best world by alleviating the distortions caused by ill-functioning institutions. The grease the wheels argument postulates that an inefficient bureaucracy constitutes a major impediment to economic activity that some ‘‘speed” or ‘‘grease” money may help to circumvent.” (Meon & Weil, 2010, p. 244).
This does not mean that corruption has general good effects, only in countries where other aspects of governance are failing, it can be beneficial. (Meon & Weil, 2010, p. 244). However, numerous findings reject the “grease in the wheels hypothesis” and find little to no evidence for positive effects, as I will show now.
Support for the positive effects of the “grease the wheels hypothesis” have been found by Andres & Ramlogan-Dobson (2011) who show that the relationship between inequality and corruption in Latin America is inversely related. Corruption actually reduces levels of inequality and might be perceived as a means of “pro-poor redistribution”. However, these findings are specific to the context of Latin America where a strong informal sector as a source of income for poorer people exists (Andres & Ramlogan-Dobson, 2011, p.972).
Meon & Weill (2010) show that in countries with weak institutions, the effects of corruption are not as destructive. “However, if some estimations imply a statistically significant positive marginal effect of an increase in corruption on efficiency in poorly governed countries, others simply imply that that effect becomes insignificant in these countries” (Meon & Weill, 2010, p. 253-254). Therefore, they conclude to find evidence that corruption can act as a grease.
8
Literature, which rejects the “grease the wheel” and supports the “sand in the wheel” hypothesis is more widespread. Kaufmann (2004) conducted a cross-national study about the relationship between human rights and corruption. His results show that more corrupt countries have lower levels of civil and political liberties (Kaufmann, 2004 as cited in Dimant & Tosato, 2018, p.
346). Cardona et al. (2018) could further empirically prove the connection between corruption and human rights. Comparing countries cross-nationally, they showed that if levels of corruption are reduced human rights are increased. However, the link itself is not constant and some “non-corruption-related factors enable a country to free itself from the gravest human rights violations. Once this boundary has been crossed, less corruption will help the human rights situation to improve from a medium to a high level” (Cardona et al., 2018, p. 341).
Some scholars argue that corruption has also an effect on the inequality of a country and increases poverty. In their work, Gupta et al. (2002) prove that high corruption on the rise leads to higher levels of poverty and income inequality, and show that policies which reduce corruption will likely reduce the levels of poverty and income inequality as well. However, the negative effects of corruption can be reduced by equal levels of access to education, good management of natural resources, effective social programs, and growth. (Gupta et al., 2002, p.
40). Levels of satisfaction with the government of a country are also affected by corruption. It has been shown that in post-communist countries, satisfaction with government is negatively affected by corruption and that the effects are more damaging if citizens feel like that there are no positive changes, regarding the political and economic sphere in their countries. (Habibov et al., 2019, p. 748).
Looking at the economic sphere, negative implications can be seen as well. In their study of American states, Gleaser & Saks (2006) prove that corruption negatively affects the economic developments in states, as well as income inequality but is uncorrelated to the size of government. States with higher levels of education had furthermore, less corruption. (Glaeser
& Saks, 2006, p. 1067). Aidt (2009) however, could not find quantitative evidence that corruption affects the growth rate of the GDP per capita negatively on the macro level, yet argues that there is evidence from surveys and field studies, that show substantial costs of corruption and shows that corruption is a hindrance for development which is sustainable. (Aidt, 2009, p. 288)
It seems to be well established, that corruption, social and political trust affect each other.
Looking at social trust, in 1993 Robert Putnam studied regions in Italy and found out that those
9
with lower levels of social trust have higher levels of political corruption. Following his idea La Porte et al. (1997), Bjornskov (2010), Graeff and Svendsen (2013), and Kubbe (2014) confirm the findings through cross-national studies. Further extending the model, Uslaner (2008) presents the thesis of a so-called “inequality trap”, where a causal link, starting from high inequality leads to low social trust, which leads to high levels of corruption which in turn amplifies high levels of inequality. “Corruption rests upon a foundation of unequal resources and it leads to greater inequality in turn” (Uslaner, 2017, p. 302). Testing his hypothesis, he compares countries cross-nationally and shows that those countries with lower levels of corruption are those which have higher levels of social trust (Uslaner, 2008 as cited in You, 2018, p. 481). In summary, numerous studies are confirming the causal effect that social trust affects corruption. However, despite these findings, the causal direction is debated and there is some academic disagreement (You, 2018, p. 482). Studies which try to show the opposite direction, that corruption affects social trust “are generally based on the theory that institutional fairness, of which corruption is an important element, affects individuals’ institutional trust as well as social trust” (You, 2018, p. 483).
When it comes to political trust, researchers found out that corruption and institutional quality have an effect on institutional or political trust. Political trust is defined as the faith that citizens have in institutions and political actors, that they will act according to their views and not in a harmful manner (Hakhverdian and Mayne, 2012, p. 740). It has been empirically proven that people in Europe are less likely to trust government when corruption is high (Hakhverdian &
Mayne, 2012). Corruption, therefore, reduces institutional as well as interpersonal trust (Kubbe, 2014, p. 16). These findings can be observed for post-soviet countries as well. Studies show that corruption erodes trust in political as well as social institutions, such as financial institutions, trade unions, nonprofit organizations, and international investors (Habibov et al., 2017, p. 178).
Furthermore, education matters as well. Comparing 21 European democracies Hakhverdian &
Mayne (2012) show that citizens with higher levels of education are better at identifying which factors are undermining democratic institutions and are therefore more worried by these actions.
“In countries with comparatively high levels of corruption, we find that education dampens political trust; in countries with low levels of corruption, education actually boosts political trust” (Hakhverdian & Mayne, 2012, p. 739).
10
Going a step further, Uslaner (2017) argues that corruption and inequality are intertwined in a vicious cycle which leads to the enrichment of the wealthy and makes citizens believe that corruption is the only way to get rich which leads to the loss of political trust. Analyzing different regions across the world with different political systems he confirms his findings and shows that when inequality is high people believe that leaders are more receptive to the interests of the wealthy compared to the rest (Uslaner, 2017, p. 313). In his comparative study, You and Khagram (2005) argue that “income inequality increases the level of corruption through material and normative mechanisms”, fortifying the argument of the inequality trap (You &
Khagram, 2005, p. 136). Other studies confirm that corruption, social and political trust influence each other and have negative consequences on citizens’ opinion of democracy, weakening the support for democratic norms and values (Choi, 2014, p.1).
Corruption further influences the political legitimacy and attitudes about the government of a country. Citizens in democracies are more negative towards their evaluation of the performance of the political system and have less trust in civil servants when corruption is high. This effect can however be mitigated by strong economic development and growth (Anderson & Tverdova, 2003, p. 101). This means that corruption can have a negative implication for the legitimacy of a political system. This could be further proven by Seligson (2006), in the Latin American context. An increase in corruption is followed by an erosion of support for democracy.
Furthermore, his findings show that corruption reduces interpersonal trust (Seligson, 2006, p.
402). Pellegata & Memoli (2016) analyze the effects of corruption on institutional confidence in EU member states with a multivariate analysis and show that countries with higher levels of corruption have less confidence in government and parliament (Pellegata & Memoli, 2016, p.
391). Furthermore, trust in parliament is consistently reduced by corruption (Van der Meer, 2010, p. 530). Chang & Chu (2006) get the same results for Asian democracies as well. Despite political corruption in Asia being described as systemic, their empirical analysis shows that higher levels of perceived corruption, reduces institutional trust (Chang & Chu, 2006, p. 269).
All in all, despite broad research, scholars remain divided about the effects of corruption. There still exists academic debate and contradictions on the implications, corruption actually has on a country. Academic research has so far not been able to truly understand corruption and its interactions and implications on the macro-, meso-, and microlevel. However, most empirical work has found evidence which supports the “sand in the wheels” hypothesis and rejects the
“grease the wheels” hypothesis”, and show that corruption has many different negative effects on a country.
11 2.2.2. Support for authoritarian leadership
After highlighting the extensive literature regarding corruption and the effects and implications it has on society, we have to ask ourselves, why would these effects also increase the desire for a strong leader? In order to answer this question, we have to dive into the existing literature for support for authoritarianism. What drives people into the hands of authoritarian parties and leaders? Ingelhart (2018) argues that historically in the 1930s, democracy’s biggest setback came when fascism became dominant in Europe. He states that the reason this was possible was due to sharp economic decline. He argues that the Nazi Party under secure conditions had only three percent in national elections in 1928 but grew drastically in 1932 parallel with the Great Depression (Inglehart, 2018, p. 21). Furthermore, “to a large degree, the shifts between democracy and authoritarianism can be explained by the extent to which people feel that their existence is secure” (Inglehart, 2018, p. 21). Under circumstances of high scarcity, xenophobia was a successful strategy. Under these conditions, “people tend to close ranks behind strong leaders, a reflex that in modern times leads to support for authoritarian, xenophobic parties”
(Inglehart, 2018, p. 21). Another point Inglehart raises is that extreme levels of inequality are not compatible with democracy. Parallels can be seen between the rise of authoritarian parties and the rise of inequality over the last thirty years. However, comparing Sweden with the United States, he argues that inequality is less problematic if the economy is rising in general. “As long as everyone was getting richer, rising inequality did not seem to matter much” (Inglehart, 2018, p. 23) If the real income of the working class is declining, however, support for authoritarianism rises. In the last decades, the majority of people in the US and other developed nations have experienced declining real income. In addition, inequality has risen, which affects the support for democracy negatively (Inglehart, 2016, p. 22). Furthermore, “resurgent authoritarianism”, the growing international assertiveness of authoritarian regimes, such as Iran, China, and Russia, is fueling the problems of democracy around the world. Their existence has shown that there are alternatives to democracy and that liberal democracy is not the only form of government. Especially China shows that rampant economic growth without democracy as a government or democratic institutions is possible (Plattner, 2017, p. 9).
This is somewhat in line with what Foa & Mounk argue in their articles, who state that
“enthusiasm for liberal democracy has fallen while openness to illiberal authoritarian alternatives to democracy has risen” (Foa & Mounk, 2017b, p. 3). However, they claim that
“democratic deconsolidation” is stronger among younger citizens. Undemocratic tendencies are
12
rising among higher-earning individuals and many western countries, such as Italy, Spain, Greece, Portugal, and the United States, experience a loss of faith in performance and a rise in levels of corruption (Foa & Mounk, 2016, p. 13).
They argue that already simple descriptive statistics might foreshadow democratic deconsolidation in a country. For example, in surveys, the highest level that democracy was considered “fairly bad” or “very bad”, was in Russia in 1995, four years before the election of Vladimir Putin (Foa & Mounk, 2017b, p. 16). To give another example, surveys show that in Venezuela already before Hugo Chavez’s election in 1995, 22.5 percent of the population preferred an authoritarian government. 46.3 percent stated that democracy “does not solve the problems of the country” and 81.3 percent thought that a strong leader would be good for the country (Foa & Mounk, 2017a, p. 11).
Empirically it has been proven that the attitudes of the younger generation are indeed an indicator in which direction the political system might go. Low institutional trust among the youth leads to a weakening of democratic regimes. The absolute measure of youth trust in a political institution and “the corresponding measure built on the trust index are positively associated with change in liberal democracy […]” (Kwak et al., 2020, p. 1381). “The lower trust in political institutions among young people, the larger the decrease in level of democracy […] in the future […]” (Kwak et al., 2020, p. 1384).
Empirically speaking, Chong & Gradstein (2018) studied which factors were relevant in citizens supporting strong leaders which are not restrained by constitutions or the legislative.
Using the World Value Survey, they analyzed over 70 countries all over the world, independent of each country’s political system, and found out that certain characteristics are relevant. On an individual level, strong leadership is favored by people with less individual income independent of the effect of attitudes towards democracy. Furthermore, education, as well as age are negatively correlated to support for strong leadership, whereas self-reported religious citizens are positively correlated. No significant results have been found regarding gender. In addition, they found that income inequality increases the support for a strong leader. Institutional characteristics matter as well. Support for a strong leader is higher when there are ethnic tensions or political instability (Chong & Gradstein, 2018, p. 1268).
Wike & Fetterolf (2018) analyzed a global survey conducted by the Pew Research Center. They conducted a 38-nation survey, asking respondents about their attitudes towards representative democracy, direct democracy, rule by experts, military rule, and rule by a strong leader without
13
institutional constraints. Their findings show that, despite the fact that democracy receives a wide general appeal, there is significant support for nondemocratic alternatives and autocracy in many nations, even in consolidated democracies. They found out that ideological beliefs, such as being politically on the right spectrum are positively correlated with support for non- democratic approaches. In addition, there is a positive correlation between the support for populist parties and support for alternatives to democracy. Education was negatively correlated:
lower levels of education lead to less support for representative democracy. Citizens with lower income are more likely to support a strong leader unrestrained by democratic institutions.
Regarding age, they find few consistent differences between younger and older adults. Negative views regarding the economy lead to less satisfaction and less commitment to representative democracy (Wike & Fetterolf, 2018).
It has also been proven with a set of experimental studies, that higher levels of inequality lead to higher support for a strong leader. The mechanism underlying this correlation is the perception of anomie. Anomie is “a state of society characterized by social dysfunction and chaos in which society provides little moral guidance to its citizens” (Sprong et al. 2019, p.
1627). In other words, anomie arises, when individuals perceive that there is a breakdown of society, socially and politically. They collectively share the view that others are not following moral principles and cannot be trusted. Furthermore, they perceive government and political leaders as illegitimate and ineffective (Sprong et al., 2019, p.1628). People are then more likely to prefer a strong leader, who takes charge and fixes the problems, the more they perceive that there is a breakdown of society. This result is contested, however, as Aruguete & McCutcheon (2021), replicating Sprong et al.’s study, find different results, stating that economic inequality does indeed lead to perceived anomie, yet none of these variables predict the desire for strong leadership. Similarly, Kakkar & Sivanathan (2017) set out the hypothesis that people desire a strong leader when they have to face “a situational threat of economic uncertainty”. Drawing on Hogg’s uncertainty theory
“Individuals are motivated to reduce uncertainty, an aversive state often perceived as a threat. […] For instance, when faced with uncertainty, individuals support groups that are perceived as more agentic, i.e., capable of taking radical actions against others, and endorse leaders who are perceived as nonprototypical and action oriented in hopes that such actions would lead to the reduction of uncertainty” (Kakkar & Sivanthan, 2017, p.
6734).
14
Using three cross-national studies, they show that their hypothesis is robust and that when faced with economic uncertainty, citizens seek an external agent, who is perceived to be dominant, authoritative, and decisive, who helps them overcome the psychological threat and regain control (Kakkar & Sivanathan, 2017, p. 6738).
Another factor which influences support for a strong leader is perceived moral polarization.
Higher levels lead to higher support. Furthermore, there is evidence “linking perceived moral polarization with perceived anomie in society (defined as the perceived breakdown of social fabric and leadership) and with the rise in support for leaders with both a conservative/authoritarian and progressive/democratic style” (Crimston et al., 2021, p. 1).
For democracy to survive, Classen (2020) argues that it also needs public support among the population and finds weak empirical evidence for his hypothesis (Claassen, 2020, p. 118). If there are low levels of support, there is the risk that democracies fail to consolidate or descend to autocracy. Even established consolidated democracies might become deconsolidated (Claassen, 2020, p. 131).
Lavric & Bieber (2021) found in their analysis that the support for a strong political leader, who is not restricted by parliament or elections in the Western Balkans has increased and is positively correlated to low income and unemployment. However, they found no correlation to citizens’ level of education. Interestingly, they could empirically prove that citizens do not think that having a strong leader, who does not have to bother with parliament and elections is conflicting with democracy but rather that they are compatible. “On the contrary, a strong leader might be seen as a welcome or, under circumstances of high economic insecurity, even essential supplement to a democratic system” (Lavric & Bieber, 2021, p. 24). However, the authors also argue that the rise of authoritarian, illiberal political systems cannot solely be explained by popular support for strong leaders. The authoritarian leaders in some of these countries came to power not based on authoritarian promises or introducing a strong leader but rather by promising to combat corruption. Thus, the popular support of a strong leader and the rise in authoritarian regimes are two “interlinked responses to the weakness of democratic institutions”
(Lavric & Bieber, 2021, p. 25).
Lastly, Rohac et al. (2017) analyze what the drivers of support for authoritarian populist parties are in Europe. Their findings show that there is a strong link between corruption and support for right-wing authoritarian populist parties. They show that higher levels of corruption and inequality lead to higher levels of support for right-wing populists. In their opinion, the
15
mechanism “through which corruption and cronyism provide a boost to populist parties is the undermining of voters’ trust in politics”, which can be even more amplified by the authoritarian populists themselves, once they arrive in power (Rohac et al., 2017, p.393).
Taking all the literature together, we can see that there are certain effects of corruption, which increase the levels of support for a strong leader or support for authoritarianism. I would argue that corruption produces many of these effects. It leads to higher levels of inequality (see Gupta, 2002), lower levels of economic performance (see Glaeser & Saks, 2006), lower levels of political trust (see Hakhverdian & Mayne, 2012), confidence in parliament (see Pellegata &
Memoli, 2016), more negative perception of government and the political system (see Anderson
& Tverdova, 2003) and lower levels of satisfaction of democracy (see Habibov et al., 2019).
All of these factors are said to influence the support for authoritarianism. I would argue that once corruption becomes systemic and high levels occur, people’s trust in institutions faints.
These low levels of trust can be cushioned by a strong economic performance (see van Erkel &
van der Meer, 2016; Anderson & Tverdova, 2003) however in the long run studies show that corruption negatively affects economic performance as well (see Gleaser & Saks, 2006). High levels of inequality and dissatisfaction with democracy are also factors which might push people to support an authoritarian leader. All these factors combined can lead to a status of anomie, insecurity, and helplessness. Citizens start seeking a strong leader, who is not bound by (perceived corrupt) institutions and other members of parliament, and who can throw out the corrupt elite and bring them the security they desire. However, I would not argue that citizens want to throw out the general concept of democracy per se. More so, they come to a state of
“anything but this”, where they are willing to take undemocratic steps, to “fix” their country.
For all these reasons mentioned above we conduct the following hypothesis which I test in this paper:
Higher levels of corruption in a country lead to higher levels of support for authoritarianism.
16
3. Data and Methods
Our hypothesis argues that high levels of corruption in a country lead to citizens’ support for an authoritarian leader. It links the contextual level of countries at certain points in time to the individual level of citizens. In order to analyze the relationship between levels of corruption and the desire for strong leadership, a repeated measure of long periods of time was necessary.
For this reason, the European Value Survey (EVS) was used. The European Value Survey is a cross-national, large-scale, and repeated cross-sectional survey research program which focuses on basic human values. It shows the beliefs, insights, values, opinions, and ideas of the citizens of Europe and focuses specifically on the topics of family, work, perceptions of life, society, morality, politics, and national identity (European Value Survey, 2022). It offers a (non-panel) longitudinal setup, which covers a time span from 1981 to 2017 and has a core questionnaire repeated over time. Since 1981 and 2017 five waves have been conducted. Since our dependent variable, the desire for strong leadership (unrestrained by parliament or elections), is only asked starting from 1999, and control variables such as the Gini-Coefficient are widely available only since the beginning of this century, we use three waves, 1999, 2008, and 2017. For our analysis, I only select those countries which are currently part of the European Union and have participated in each of those three waves. The following 20 countries are included in the analysis: Austria, Bulgaria, Croatia, Czech Republic, Denmark, Estonia, Finland, France, Germany, Hungary, Italy, Lithuania, Netherlands, Poland, Portugal, Romania, Slovakia, Slovenia, Spain, and Sweden. Belgium, Cyprus, Greece, Ireland, Latvia, Luxemburg, and Malta have been excluded due to one or more missing participation for those three waves. The final sample size of all three datasets merged together is about 88.000 respondents.
3.1. Dependent variable
The desire for authoritarianism is measured by the question: “Having a strong leader who does not have to bother with parliament and elections.” with the following possible response categories: “Very good”, “Fairly good”, “Fairly bad”, “Very bad”, “No answer”, and “Don’t know”. We remove the possible answers “No answer” and “Don’t know” and set them as missing values. This question gives a clear overview of the respondents’ beliefs about
17
tendencies toward authoritarianism with the important aspect the leader does NOT have to bother with parliament and elections, which is a clear sign of authoritarian tendencies.
3.2. Independent variables
The main independent variable in this study is perceived levels of corruption. As mentioned in previous chapters measuring corruption proves increasingly difficult due to numerous problems. Corruption takes place behind the scenes and chances are that a lot of problems never come to light. The most prominent indicators for country-level corruption are the Corruption Perception Index (CPI) by Transparency International as well as the Control of Corruption index (CCI) by the World Bank. Both are composite indexes of corruption perceived by experts and businesspeople. However, for this study, I decided to use the Control of Corruption indicator by the World Bank and not the Corruption Perception Index by Transparency International. The reason for this is due to the fact that Transparency International updated its methodology of the Corruption Perception Index in 2012. Before 2012 the CPI was based on the perceptions of corruption in a country relative to other countries’ scores. For this reason, Transparency International itself states that scores before 2012 are not comparable over time (Transparency International, 2012). For this reason, I decided to use the Control of Corruption indicator, provided by the World Bank from its Worldwide Governance Indicators. The indicator is defined as follows:
“Control of Corruption captures perceptions of the extent to which public power is exercised for private gain, including both petty and grand forms of corruption, as well as ‘capture’ of the state by elites and private interests.
Estimate gives the country's score on the aggregate indicator, in units of a standard normal distribution, i.e. ranging from approximately -2.5 to 2.5.”
(The World Bank, 2022a).
The underlying data is gathered from numerous “survey institutes, think tanks, non- governmental organizations, international organizations, and private sector firms” (The World Bank, 2022b). A lower value estimates a higher level of corruption, while a high estimate of the CCI shows lower levels of corruption (Kaufmann et al., 2011). This indicator has the advantage that it did not change its methodology (only upgraded the underlying source data)
18
and can be used for broad cross-country comparisons as well as longitudinal trends over time (The World Bank, 2022c). While the World Bank does not recommend using it for short-run year-to-year changes, it is unproblematic over longer periods of time, for example, a decade.
The World Governance Indicators are useful to show trends in the respected countries (The World Bank, 2022b). Since our three waves have a nine-year gap in between them, we can safely use the Control of Corruption indicator.
The second independent variable, used as a control variable is the country’s levels of inequality.
As shown in previous chapters, corruption and inequality are highly linked together creating a so-called inequality trap. In order to assess, the inequality levels of a country we use the Gini- Coefficient provided by Eurostat. The Gini coefficient is a measure of the income distribution of a population (OurWorldInData, 2013).
“A Lorenz curve plots the cumulative percentages of total income received against the cumulative number of recipients, starting with the poorest individual or household. The Gini index measures the area between the Lorenz curve and a hypothetical line of absolute equality, expressed as a percentage of the maximum area under the line.” (The World Bank, 2022d).
It ranges from 0 to 100, whereas 0 means that incomes are perfectly equally distributed and a value of 100 means that incomes are perfectly unequally distributed, meaning one person has all income and everybody else has 0.
We use the Gini coefficient provided by Eurostat as Eurostat is one of the few, which offers consistent coverage of the Gini coefficient in EU countries, going back as far as 1995. However, since the data for the years before 2001 are incredibly scarce and contain a large number of missing values, I decided to use the Gini coefficient of 2001 instead of 1999, as it offers a value for nearly all countries used for this study, except for Croatia and Slovakia. While this procedure is not optimal, this should not temper our data as the Gini coefficient is historically rather sticky, and the difference between two years should not distort the outcome significantly.
The third independent, country-level control variable I use is the Gross Domestic Product per capita (at purchasing power parity) (GDP per capita (PPP)) provided by the World Bank. The indicator shows the
19
“per capita values for gross domestic product (GDP) expressed in current international dollars converted by purchasing power parity (PPP) conversion factor. GDP is the sum of gross value added by all resident producers in the country plus any product taxes and minus any subsidies not included in the value of the products” (The World Bank, 2022d).
I decided to use the GDP (PPP) (purchasing power parity) per capita instead of the GDP (nominal) per capita, as the GDP (nominal) per capita does not reflect differences between countries’ cost of living and inflation rate. The GDP (PPP) per capita is more useful for comparing the differences in living standards between countries, as it takes these into account.
It is measured in current international dollars. Due to the fact that all three variables are considered sticky and do not move substantially from year to year, I decided not to include any form of a time lag.
At the individual level, we also test for other variables such as education, religion, income, and age. All of these three are proven to have an effect on the desire for strong leadership. They are measured in the European Value Survey in all three waves. Education is categorized as
“higher”, “medium”, and “lower” education. Religion is a categorical variable with the answer categories “Roman Catholic”, “Protestant”, “Free church/Non-conformist/Evangelical”, “Jew”,
“Muslim”, “Hindu”, “Buddhist”, “Orthodox”, and “Other”. Age is categorized into six categories “15-24 years”, “25-34 years”, “35-44 years”, “45-54 years”, “55-64 years”, and “65 and more years”. The variable household income is categorized into “Low”, “Middle”, and
“High”.
3.3. Methods
I conduct our empirical analysis via multiple multi-level regression analysis. The reason why we use a multi-level model is, that our model has two different levels of analysis, namely country-level and individual combinations. A multi-level model is best suited for this kind of analysis.
I start with a comparison of the median values of the main variable, namely support for a strong leader, and a simple bivariate regression between the Control of corruption indicator to gain a first overview. In a second step, a multivariate regression, which includes the control variables
20
shows more nuanced results. However, in order to also see the time-series dynamic, I further refine the model by conducting a time-series regression. As mentioned before, I decided not to include a time-lag, due to the fact corruption, as well as the country-level variables of GDP per capita and Gini coefficient tend to be rather sticky, and do not show significant changes over one year.
4. Results and analysis
Table 1 shows the differences in countries’ desire for strong, authoritarian leadership: The average desire for a strong leader who is not bothered by parliament and elections is the strongest in Romania (3.01), followed by Bulgaria (2.64) and Lithuania (2.55). Lowest on the spectrum are Austria (1.6), Denmark (1.62), Germany, and Sweden (both 1,64). Interestingly enough, the desire for a strong leader does not fluctuate strongly within each country. The highest fluctuation between 1999 and 2017 is in Croatia, where support has risen by 0.53. This means that (with some exceptions) a country’s desire for a strong leader stays in general rather rigid in each respective country with no major changes within 18 years. Furthermore, more than half of the countries used for this study have a higher value in 2017 compared to 1999. At a first glance, we cannot see any pattern (whether geographical, political, or historical), as to why countries have a higher level in the first place or why levels increased or decreased.
Table 2 shows the Control of Corruption indicator by the World Bank. Here we can see that the countries with the highest average levels of corruption over the years are Romania (-0.277), Bulgaria (-0.194), and Croatia (-0.149). Interestingly the two countries with the highest levels of corruption (Romania and Bulgaria) are the ones with the highest level of desire for a strong leader, which could be a first indication that the two variables are influencing each other. Of course, historical context is an important factor and could be the explanation.
21
Table 1: Descriptives for Support for an Authoritarian Leader
N 1999 2008 2017 Mean
Austria 4381 1.59 1.71 1.51 1.60
Bulgaria 3258 2.36 2.75 2.70 2.64
Croatia 3793 1.63 2.10 2.16 2.01
Czechia 4978 1.70 1.94 1.87 1.83
Denmark 5724 1.59 1.58 1.64 1.62
Estonia 3446 1.87 2.00 1.70 1.87
Finland 3170 1.98 1.62 1.61 1.73
France 4736 2.06 1.90 1.85 1.93
Germany 5832 1.62 1.63 1.67 1.64
Hungary 3803 1.80 2.00 1.81 1.88
Italy 5471 1.64 1.61 1.97 1.76
Lithuania 3347 2.59 2.53 2.54 2.55
Netherlands 4650 1.96 2.21 2.04 2.08
Poland 3602 1.94 1.90 1.71 1.84
Portugal 2890 2.24 2.50 2.56 2.45
Romania 3792 2.86 3.02 3.11 3.01
Slovakia 3813 1.85 1.65 1.96 1.82
Slovenia 3299 1.86 1.97 1.97 1.94
Spain 3372 1.89 1.72 1.85 1.81
Sweden 3148 1.75 1.55 1.62 1.64
Source: European Value Survey (1999, 2008, 2017); (1=very bad, 2=fairly bad, 3=fairly good, 4=very good).
22
Table 2: Control of Corruption Indicator (Worldwide Governance Indicators)
1998 2008 2017 Mean
Austria 1.851 1.843 1.545 1.746
Bulgaria -0.172 -0.255 -0.153 -0.193
Croatia -0.556 -0.011 0.118 -0.149
Czechia 0.565 0.358 0.607 0.509
Denmark 2.256 2.393 2.239 2.296
Estonia 0.652 0.991 1.245 0.963
Finland 2.259 2.337 2.211 2.269
France 1.394 1.407 1.267 1.356
Germany 2.053 1.759 1.844 1.886
Hungary 0.766 0.472 0.125 0.454
Italy 0.518 0.270 0.211 0.333
Lithuania 0.319 0.139 0.556 0.339
Netherlands 2.147 2.118 1.792 2.019
Poland 0.819 0.461 0.729 0.670
Portugal 1.294 1.070 0.861 1.075
Romania -0.615 -0.143 -0.074 -0.277
Slovakia 0.089 0.362 0.132 0.194
Slovenia 1.213 0.982 0.816 1.004
Spain 1.399 1.190 0.548 1.046
Sweden 2.228 2.227 2.148 2.201
Source: World Bank
23
Furthermore, the three countries with the lowest levels of corruption Sweden (2.201), Finland (2.269), and Denmark (2.296) are also the ones with very low desires for a strong leader. One exception which we can see here is in the Netherlands, which is on average the 4th least corrupt country, however, has very high levels of support for a strong leader.
With Table 3 we move on to the multilevel tests. Model 1 shows the Control of Corruption indicator by the World Bank. Here we can see that the levels of corruption have the expected effect on the support for authoritarian leadership. Higher levels of corruption lead to higher levels of support for a strong leader, who is not constrained by parliament or elections. Model 2 adds control variables at the individual level. All of these variables are significant: older people, citizens with higher education, and citizens who live in a household with a higher income are less likely to support authoritarian leadership. For Religion, we see different results, however, as not all religions are significant. We find no significance for Buddhists, Jews, Hindus, and Orthodox (see Table 3). In addition, corruption stays robust and significant even with the inclusion of the individual level control variables.
With Model 3 we include the country-level indicators economic performance, and the inequality in a country with the GDP per capita (PPP) and the Gini coefficient. Adding these, indicators we can see that corruption stays robust and remains significant. Looking at the Gini coefficient, we can see that it affects support for authoritarianism as well. This result is expected, as corruption and inequality are deeply intertwined (see Uslaner 2017) and confirms the findings of Chong & Gradstein (2018). However, economic performance measured via the GDP per capita (PPP) seems to be insignificant. We cannot find any evidence that the GDP per capita (PPP) influences the support for an authoritarian leader. This is rather surprising, as we expected the economic performance to influence the support for authoritarianism as studies have shown that the effects of corruption can be reduced by economic performance and that economic performance influences also political trust (see Anderson & Tverdova, 2003). In general, however, we can accept the hypothesis that higher levels of corruption lead to higher support for a strong leader, who does not have to bother with parliament and elections on a 95%
confidence level. With Model 4 we check the interaction effects between levels of corruption and education. This would show us if the levels of corruption are moderated by education. Here our results are not significant. Higher levels of corruption do not affect higher or lower educated citizens more in leading them towards the support for an authoritarian leader. In other words, with our existing data, we cannot say that the levels of corruption affect higher educated citizens’ support for a strong leader more than lower educated ones.
24
Table 3: Main results of the regression models
Model 1 Model 2 Model 3 Model 4
Constant 2.16 (0.09)** 2.67 (0.09)** 2.08 (0.23)**
Country-level determinants
Corruption -0.15 (0.07)* -0.15 (0.06)* -0.17 (0.06)**
Economic performance -0.00 (0.00)
Inequality 0.02 (0.01)**
Individual-level determinants
Age -0.01 (0.00)** -0.01 (0.00)**
Education -0.18 (0.00)** -0.18 (0.00)**
Income -0.05 (0.00)** -0.05 (0.00)**
Religion
Protestant -0.09 (0.01)** -0.09 (0.01)**
Free church/Evangelical -0.08 (0.04)* -0.08 (0.04)
Jew -0.07 (0.09) -0.09 (0.09)
Muslim 0.06 (0.03)* 0.05 (0.03)
Hindu 0.16 (0.14) 0.15 (0.14)
Buddhist -0.14 (0.13) -0.14 (0.13)
Orthodox 0.01 (0.02) 0.00 (0.02)
Other -0.12 (0.03)** -0.12 (0.03)**
No Religion -0.08 (0.01)** -0.09 (0.01)**
Corruption*Education -0.02 (0.01)
Notes: N (individuals): 76,154; N (country-waves): 3. *p < 0.05; **p < 0.01.
Source: European Value Survey (1999, 2008, 2017)
25
In this first step, we have seen that levels of corruption are indeed a significant factor for citizens’ desire for an authoritarian leader. Higher levels of corruption, as well as higher levels of inequality, lead to higher support for a strong leader, who is unrestrained by parliament and elections. We could not find any evidence, however, that the economic performance influences our dependent variable. All of our individual-level variables have significance as well. Younger citizens, the less educated and citizens coming from a household with a lower income all have stronger desires for an authoritarian leader. There is, however, no interaction between the levels of corruption and education.
What is missing in our model so far is to see whether it is actually differences between countries or because of changes in levels of corruption, economic performance, and inequality within a country over time, which leads to the support for an authoritarian leader. In a second step, I split the country-level indicators to test this. Surprisingly, as we can see in Model 5, once we split the country-level indicators, we do not reach significance for any of the macro-level variables, except for changes in inequality over time. This means that increasing levels of inequality over time within a country are a driving factor for the desire for a strong leader.
However, once split up, both levels of corruption as well as GDP per capita (PPP) do not show any significance. This result suggests that the negative effect of corruption is indeed significant, however, we cannot see whether it is because of within-country variance, cross-national variance, or an absolute effect. Lastly, we will check again for an interaction effect between education and corruption for the desire for an authoritarian leader. Model 6 shows that neither changes in levels of corruption over time nor differences between countries are interacting with the level of education. We cannot say that the changes in corruption within a country nor the differences between countries affect higher educated citizens’ support for a strong leader more than lower educated ones.
26
Table 4: Split up of country-level variables
Model 5 Model 6 Model 7
Constant 2.01 (0.73)**
Country-level determinants
Corruption (between countries) -0.06 (0.16) Corruption (within country over time) -0.09 (0.09) Economic performance (between countries) -0.00 (0.00) Economic performance (within country over time) 0.00 (0.00) Inequality (between countries) 0.03 (0.02) Inequality (within country over time) 0.02 (0.01)*
Individual-level determinants
Age -0.01 (0.00)**
Education -0.19 (0.00)**
Income -0.06 (0.00)**
Religion
Protestant -0.09 (0.01)**
Free church/Evangelical -0.08 (0.04)
Jew -0.12 (0.09)
Muslim 0.05 (0.03)
Hindu 0.15 (0.14)
Buddhist -0.13 (0.13)
Orthodox 0.01 (0.02)
Other -0.12 (0.03)**
No Religion -0.09 (0.01)**
Corruption (between countries)*Education -0.02 (0.02)
Corruption (within country over time)*Education 0.00 (0.05)
Notes: N (individuals): 76,154; N (country-waves): 3. *p < 0.05; **p < 0.01.
Source: European Value Survey (1999, 2008, 2017)
