Content

2 Bathtub Model of Corruption in:

Ina Kubbe

Corruption in Europe, page 53 - 98

Is it all about Democracy?

1. Edition 2015, ISBN print: 978-3-8487-2347-8, ISBN online: 978-3-8452-6451-6, https://doi.org/10.5771/9783845264516-53

Series: Comparative Politics - Vergleichende Politikwissenschaft, vol. 6

Bibliographic information
Bathtub Model of Corruption Panel and Multilevel Analysis: A New Framework Corruption is a multilevel phenomenon that has to be examined at various levels of analysis. Initially, it takes place between certain individuals who act corruptly at the micro level (individual level). In addition, the aggregation of corrupt individual actions leads to certain levels of corruption at the macro level (country level) where corruption is usually measured by certain macro indices (e.g. Corruption Perceptions Index, Control of Corruption Index). The review of theoretical approaches and empirical studies demonstrates that it is essential to combine certain theoretical perspectives and empirical findings from different disciplines to build a solid foundation for analyzing the causes of corruption on different levels and across and within certain countries. It has been indicated, that neither economic nor sociological approaches by themselves can completely capture the complex phenomenon of corruption and its causes. While economists usually tend to “under-socialize” actors and their decisions, assuming that individuals are always acting in pursuit of their own profit, sociologists in contrast “over-socialize” decision-making, suggesting that they essentially act on behalf of social norms and traditions. In sum, they can explain only some aspects of the emergence of corrupt behavior but cannot cover the entire width of possible explanations. In this study, therefore, economic and sociological approaches are considered and used as theoretical approximation rather than extensive explanation models. Moreover, a review of the literature has shown that theoretical approaches on the one hand and empirical analyses on the other hand have developed independently from each other in corruption research. In fact, a range of literature concentrates on the theoretical and conceptual analysis of corruption while other articles explain corruption empirically, but with less theoretical background. In this regard, Coleman’s and Esser’s bathtub model (1990 / 1993) is a model that allows a combination of certain approaches to explain the causes of corruption. It integrates certain theoretical perspectives and enables scholars to focus on country and individual characteristics when studying corruption. The above mentioned theoretical approaches, namely econo- 2 2.1 53 mic and sociological such as rational-choice, sociological and historical institutionalism, and cultural approaches, and empirical studies and, thus, corresponding variables can be included in this model of corruption and fill with empirical data. Bathtub Model of Corruption Coleman’s bathtub model was originally developed to reconstruct and illustrate Weber’s analysis of the relationship between Protestant ethic and the development of capitalism (Weber, 2005). It is also known as the “Coleman boat” and offers an analytical framework for the analysis of what causes corruption. It includes the potential to analyze the complexity of corruption at a micro level, considering individual’s social behavior and social norms, as well as at a macro level, taking country characteristics and institutional settings into account. Therefore, the Coleman bathtub serves as an explanatory model for social phenomena by detailing macro-micro-macro cycles and allows integrating both, economic-rational and sociological considerations. More precisely, Coleman distinguishes between the level of action (micro level) and the system level (macro level). He links these two levels of analysis by illustrating the causal relationships that leads from the macro to the micro level and back to the macro level. While the macro level refers to the respective social situation, the micro level describes actors and their subjective perceptions, motivations and actions that, in turn, aggregate on the macro level (Coleman, 1990). In this context, Coleman (1990, p. 28) states that “the only action takes place at the level of individual actor, and the ‘system level’ exists solely as emergent properties characterizing the system of action as a whole. It is only in this sense that there is behavior of the system. Nevertheless, system-level properties will result, so propositions may be generated at the level of system.” Thereby, the “bathtub floor” represents the level of individual behavior that helps to explain macro level processes at the “water surface”. This includes that phenomena of the macro level influence system outcomes such as corruption (collective explanandum) through their effect on individuals’ orientations and behavior at the micro level (actor’s action). According to this, causal 2 Bathtub Model of Corruption 54 influence for macro factors can only work through disaggregated effects at the micro level. The Basic Model of Sociological Explanation Source: Coleman (1990) and Esser (1993a) Coleman’s bathtub model builds on the basic model of sociological explanation of social processes. The starting point of this model which is based on rational-choice approaches is the assumption that the explanation itself requires three different analytical steps (Esser, 1993b, pp. 8–10). First, the researcher has to reconstruct the “logic of situation” (Esser, 1993b, p. 8) for typical actors in typical situations (see Figure 2). Following Esser, actors are restricted, resourceful, evaluating, expecting, maximizing individuals. This implies a synthesis of the homo economicus and homo sociologicus and refers to the concept of bounded rationality that assumes that perfectly rational decisions are often not feasible in practice because of the finite computational resources available for making them (Simon, 1991). In this context, the structural conditions are descriptively connected with the actors of a bounded system of interaction (arrow 1). At this point, scholars usually formulate hypotheses that bridge the actor’s situation and the variables of applied theoretical approaches. Secondly, starting from this description, a theory of action is used to explain human behavior. Thus, an individual’s action is a consequence of a particular “logic of selection” (Esser, 1993b, p. 8) that is specified by the assumptions of the used theory (arrow 2). According to (Esser, 1993b, p. 9), human behavior is, in this way, conceptualized as an “intentional choice between options, one that is oriented to the respective situation. It therefore takes into account the human capacity for creativity, reflection Figure 2: 2.1 Panel and Multilevel Analysis: A New Framework 55 and empathy, as well as the significance of scarcity and opportunity costs of action.” The selection process itself can be decomposed into three steps: cognition of the situation, evaluation of the consequences of certain actions, and selection of a particular action on the basis of a specific rule (Esser, 1993b; Lindenberg, 1989). The third and last step involves the “logic of aggregation” (Esser, 1993b, p. 8). Individual actions are aggregated so that a social explanandum follows, for example, through formal derivations or partial definitions (arrow 3). Thereby, macro-sociological phenomena are reconstructed as unintended consequences of individual actions in respective situations (see also Hernes, 1976; Coleman, 1990). Based on Coleman’s and Esser’s model, I designed a heuristic model that combines the macro and micro level and considers corruption as individual action, taking particularly country and personal characteristics into account. Thus, the model resolves the duality between “under- and oversocialization” of both economic and sociological approaches and allows not only rational-economic but also social-value-based interpretations and explanations of corruption. Figure 3 portrays the bathtub model of Coleman and Esser and illustrates its application for the following analyses of corruption. The Bathtub Model of Corruption Source: Coleman (1990); Esser (1993a) Figure 3: 2 Bathtub Model of Corruption 56 As visualized in figure 3, situation 1 at the macro level presents the starting point of the model and illustrates characteristics of certain countries and its societies such as the economic, political, socio-cultural or historical context. According to the theoretical approaches introduced and the empirical studies reviewed, I assume for Europe that economic variables, for instance, a country’s economic development, political indicators such as the degree of democracy, socio-cultural variables such as a society’s dominant religion, and historical variables such as a country’s duration of democracy or communist past, affect the country-level. These variables describe the characteristics of certain countries and its societies at the macro level and, in turn, may influence individual actions at the micro level and vice versa. Therefore, certain characteristics of countries and societies can shape a certain “situational logic” (Esser, 1993a) by creating specific social frameworks for actions of individuals. Individuals, in turn, act under these particular social structural and institutional conditions and according to a specific social logic at the micro level. In fact, decisions of actors can be analyzed at the micro level based on rational-choice theory (corruption as rational behavior), but can also be conceived as individual behavior that is influenced by certain socio-demographic factors, cultural norms, values, attitudes or social relationships within countries and its institutions as well, according to sociological approaches. Based on these considerations, corruption can be interpreted as rational behavior but also as a way of life or a set of values and social norms that belong to a society’s culture. Therefore, individuals follow a certain logic of selection and decide to engage in corrupt activities either by virtue of rationally- or socially-based motives. At this point, it is worth stressing again that rational-choice assumptions in particular are primarily theoretical and are difficult to examine empirically. Experiments, for example, could support the exploration of rational behaviour of individuals. This work, however, pursues a quantitative approach from a comparative perspective, that struggles with an inadequate data situation in the field of individual rational behaviour. In this study, rational-choice assumptions are especially examined by contextual conditions of the country and individual level referring particularly to various resources. For this purpose, especially economic and socio-demographic factors are included in the analyses to explain the extent of corruption at both levels. Depending on economic factors such as a country’s economic development and specific socio-demographic factors such as 2.1 Panel and Multilevel Analysis: A New Framework 57 age, gender, an individual’s satisfaction with the financial situation and cultural variables rational-choices assumptions are taken into account. Finally, the aggregation of the individual actors' (corrupt) actions leads to situation 2 on the macro level, where corruption can be expressed, for example by criminal statistics or survey-based indices such as the Corruption Perceptions Index or Control of Corruption Index. The aggregation of the individual actors' actions can either reproduce social structures and behavioral patterns or modify them; in any case, it institutionalizes a social situation 2, which Esser (1993b, p. 8) calls the “collective explanandum”. Thus, levels of corruption are explained at the macro level via the aggregated individual actions at the micro level. In sum, this model combines economic and sociological approaches into an interdisciplinary framework and offers an integration and analysis of certain variables at different levels that may influence the extent of corruption. For the purpose of identifying the causes of corruption, specific situations at each level of the model can be filled by empirical data. Moreover, it allows to elicit and analyze the causes of corruption in European states over time as well as across and within countries. In the following chapters, this model serves as framework for analyzing the extent of corruption. Corruption in European Countries The analysis of the causes of corruption in European states is essential for several reasons. At first, descriptive analyses reveal that European states exhibit a wide spectrum of corrupt activity and are marked by large differences in terms of the extent of corruption (see also Charron, 2015). A closer investigation of patterns of how corruption has developed in recent years, clearly demonstrates that both new and established democracies in Western and Central and Eastern Europe show varying levels of corruption. For instance, countries such as Sweden (8.9) or the Netherlands (8.3) received very low degrees of corruption by Transparency International in 2013, ranking countries on a scale from zero (high corruption) to ten (low corruption), while states such as Romania (4.3) or the Ukraine (2.5) have received high scores of corrupt activities. Similar to countries in Central and Eastern Europe, states in Southern Europe such as Spain, Portugal, Greece or Italy have struggled with a number of corruption scandals. An initial examination of corruption development in Southern Europe also clearly demonstrates a continuous deterioration of corruptions scores in 2.2 2 Bathtub Model of Corruption 58 these countries. Since the beginning of the economic crises in 2007, corruption values have continuously increased in these countries. While in 2005, Spain received 7.0 points by Transparency International, its corruption score amounted to 5.9 in 2013, with a further declining tendency. The other Southern European countries show very similar developments. For instance, Portugal received 6.5 points in 2005 and 6.2 points in 2013, while Greece corruption score amounted to 4.3 points in 2005 and 4.0 points in 2013 (Corruption Perceptions Index, 2015). The reasons for this development still remain undiscovered. Due to this pronounced intra- and inter- European variation and continuously increasing corruption values in certain countries, the causes of corruption in European countries need to be analysed. Secondly, there is no other region in the world but wider Europe where young democracies (e.g. Slovenia, Estonia) and well-established old democracies (e.g. Greece, United Kingdom, France) as well as authoritarian systems (e.g. Belarus, Ukraine) are located so closely to each other. All the more, Europe has a unique history. After the end of the Cold War and the transformation of communist countries to young democracies, the political, economic and socio-cultural situation in Europe has changed considerably, notably in the manifestation of corruption (Holmes, 2006). These countries in Central and Eastern Europe have experienced a long way of political transition that have possibly strongly affected the extent of corruption in Europe as a whole. In this context, Kostadinova (2012, p. 26) describes that “Simultaneous political, economic, and social reforms take place in the absence of structures, knowledge, and experience to maintain and solidify the efforts […]. The complex transformations, along with disputes over borders and land and interethnic conflict, made the postcommunist systems vulnerable to the spread of organized crime and improper enrichment” (similar Xin and Rudel, 2004; Holmes, 2009). The fact that a huge group of these countries experienced a systematic shock with the collapse of communism and that some of them became member of the European Union while others did not, created somewhat like a natural experiment with sufficient variation in the treatments. This variation is well reflected in a country’s years under communist rule and years of EU-membership. In a nutshell, the impact of systematic shock and variation in post-shock treatment can be studied in Europe and nowhere else. Furthermore, Europe presents the continent with the longest history of nation-states, and rule of law as one of the oldest traditional 2.2 Corruption in European Countries 59 European constitutional principles. When studying corruption in Europe, this has to be taken into consideration. Fourthly, the review of the empirical literature on corruption illustrates that previous research has often concentrated on countries in a global context and that in most of these studies, European countries were included in large samples with, for example, African states (Billger and Goel, 2009; Littvay and Donica, 2011). Yet, African states, for instance, are characterized by different regional indicators such as the economic development, traditions and norms than countries in Europe. For example, in the African context, corruption is socially embedded in so-called “'logics' of negotiations, solidarity, gift-giving, predatory authority and redistributive accumulation“, as Sardan (1999, p. 25) illustrates (similar Egbue, 2006; Mbaku, 2007). This example should demonstrate that using large samples including countries in a global context could lead to a biased analysis of corruption from a comparative perspective (selection bias). In fact, in this way, region-specific characteristics that might be important for the explanation of the extent of corruption are neglected. Moreover, it must be taken into account that the perception of corruption varies from society to society and depends on several factors such as historical developments or culture and traditions of countries. As a result, global samples are not qualified to find out the specific determinants of corruption among European states. Therefore, to find out area-specific determinants of corruption it is necessary to focus on individual regions such as Europe rather than on global samples. Finally, the existing research on corruption in European states shows a research gap, especially from a quantitative point of view. Currently, the primary focus is concentrated on qualitative studies such as the book by Heidenheimer and Johnston (2009) that includes some case-analyses on Europe. These qualitative studies are indispensable and play a central role in the research of the manifold phenomenon corruption. However, these qualitative studies are strongly case-related and less comparative, so that eligible indicators of corruption in Europe cannot systematically be classified and generalised. Moreover, the causes of corruption in European states have never been sufficiently studied from a comparative perspective. As a result, there is still no overall explanatory model of the causes of corruption – neither by cross-sectional nor by panel or multilevel analyses. 2 Bathtub Model of Corruption 60 Measuring Corruption: The Dependent Variable Reconsidered Corruption cannot be measured directly. It is secretive by nature and frequently takes place in hidden and unofficial settings because all participants are highly interested in keeping their corrupt actions secret. Even the victims of corruption are often unaware that they have indirectly participated. However, corruption is not always illegal, but it is often a matter of ethics and perceptions that primarily depend on certain area-specific social norms and traditions (Sandholtz and Koetzle, 2000). The traditional approach of directly measuring corruption through selfreporting is unlikely to be reliable. Such an approach under-measures corruption due to socially desirable responses and it is still difficult to identify the actors involved as well as the frequency of corrupt actions. In general, people do not want to admit that they act corruptly, and it can be assumed that this will always be the case. Nevertheless, while it is difficult to capture corruption, it is not impossible. Several ways of measuring corruption exist which allow some generalizations and can produce quantifiable results. The most common strategy – and so far probably the only way forward – for measuring corruption is in an indirect way. Here, data on corruption usually consist of subjective assessments of the corruption levels in different countries. Currently, the following sources are available: – Questionnaire-based surveys (such as the Corruption Perceptions Index developed by Transparency International or the Control of Corruption Index offered by the World Bank; the World Values Survey). – Reports on corruption that are available from published sources such as newspapers. – Case studies of corrupt agencies such as tax administrations, customs, and the police. – Experimental studies The most frequently applied way of measuring corruption is the use of questionnaire-based surveys that are also based in reports from published sources. These approaches include certain macro and micro level indices such as the Control of Corruption Index, the Corruption Perceptions Index or the International Country Risk Guide. At the individual level, the most widespread method is the use of national survey data which obtain corruption perception and experience data (e.g. World Values Survey). Contrary to these approaches, case-studies of certain corrupt actors such as police officers are often used to investigate individual cases. However, to mea- 2.3 2.3 Measuring Corruption: The Dependent Variable Reconsidered 61 sure corruption in general and for comparative reasons they are not ineligible. Recently, experimental studies testing corruption empirically have grown rapidly and present a rather new methodological field (Renner, 2004; Dušek et al., 2005; Bobkova and Egbert, 2012). Most of these experiments focus on the relationship between corruption, gender and culture and have already offered important micro level results. Some of them will be discussed in more detail in the following chapters. The Aggregate Level: Macro Level Indices Fortunately, in recent years, several organizations such as Transparency International and the World Bank have developed corruption perceptionbased indices based on expert reports across a wide range of countries to quantify the extent of corruption.37 These different quantitative indices have enabled scholars to study corruption in an empirical way and offer a wealth of information facilitating them to show a number of important results. They present statistical aggregation of secondary data and have been developed to raise awareness about the problem of corruption. In the following section, the best known and most frequently used are presented: Control of Corruption Index The Control of Corruption Index (CoC) is developed by the World Bank. It is part of the Governance Index and includes six dimensions: 1. voice and accountability, 2. political stability and no violence, 3. government effectiveness, 4. regulatory quality, 5. rule of law and 6. control of corruption. From 1996 to 2002, the index was published biannually. It covers almost 220 countries and is now published every year. The World Bank’s definition of corruption is basically the same as the one used by Transparency International: “Corruption is the misuse of power entrusted in public officials for private gain.” The CoC Index is based on over 100 individual variables that measure perceptions of 2.3.1 37 Before the development of certain perception-based indices, there were no statistics and quantifications of corruption. Measuring the degrees of corruption had to be based on anecdotes and intuition (Scott, 1972). 2 Bathtub Model of Corruption 62 governance drawn from 25 data sources and constructed by 18 different organisations. It “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” (World Bank, 2015). The World Bank Group compiles and summarizes this data from 30 sources and reports the views and experiences of citizens, commercial business information providers, experts in the public and private as well as NGO sectors from around the world. Based on these survey results, every nation is assigned a corruption control score ranging from -2.50 to 2.50, where lower scores (closer to -2.50) indicate low control of corruption, and higher scores (closer to 2.50) indicate high corruption control (Worldwide Governance Indicators, 2015). Corruption Perceptions Index The Corruption Perceptions Index compiled by Transparency International has become one of the most reliable and widely used indicators of corruption around the world. The meta-index was first launched in 1995 and ranks almost 200 countries based on the degree to which corruption is perceived among public officials and politicians. The CPI is a composite index drawing on 14 different polls and surveys from seven independent institutions and is carried out among business people and country experts. It also includes surveys of local residents and expatriates who rank countries on a scale from zero (high corruption) to ten (low corruption), according to the level of perceived corruption. One of the main advantages of this index it is reliability. Johnston (2001b, p. 161) claims: “Reliability is the strongest point of the CPI. Rather than employing just one or a few indicators, the data reflect the views of thousands of individuals who encounter corruption in differing ways in range of countries, and are gathered in a variety of ways.” In 2012, Transparency International updated the methodology of the Corruption Perceptions Index which means that the score better capture changes in the perception of corruption over time. These changes also include that the following CPI scores are not comparable with the scores the years before. The CPI since 2012 is presented on a 0-100 scale. However, as Lancaster and Montinola (2001) evaluate the CPI in great detail and conclude that while no measure is perfect, the Corruption Perceptions Index appears to be a rather robust one. Despite these limitations, 2.3 Measuring Corruption: The Dependent Variable Reconsidered 63 I use the CPI to measure corruption at the macro level over time because it is still the most valid and reliable data that is often used to demonstrate how the corruption levels of countries have improved or deteriorated. Yet, I am very aware of the weaknesses of using these indices. International Country Risk Guide Since the early 1980s, the International Country Risk Guide dataset (ICRG) has covered almost 150 countries on a monthly basis. It includes 22 variables of political, economic and financial risk ratings. For each subcategory, a separate index is created. The used sources also include CoCand CPI-data and are based on perceived corruption by a large number of country experts. The variable corruption within the political system is part of the political risk rating38 that aims to provide a mean of assessing political stability on the basis of subjective analysis of available, comparable information, while the economic and financial risk assessments are made on the basis of objective data. The measurement of corruption refers to “actual or potential corruption in the form of excessive patronage, nepotism, job reservations, ‘favor-for-favors’, secret party funding, and suspiciously close ties between politics and business“ (International Country Risk Guide, 2015). This is done by assigning risk points to a pre-set group of factors that are termed political risk components. The minimum number of points that can be assigned to each component is zero, while the maximum number of points depends on the fixed weight that component is given in the overall political risk assessment. In each case, the lower the risk point total, the higher the risk, and the higher the risk point total, the lower the risk. In fact, the ICRG assesses the risk of firms to invest in a country rather than the level of public sector corruption. That is why this index is not used in the following analysis to measure corruption.39 38 The political risk rating further includes government stability, socioeconomic conditions, investment profile, internal conflict, external conflict, military in politics, religious tensions, law and order, ethnic tensions, democratic accountability and bureaucracy quality. 39 The Bribe Payers Index (BPI), developed by Transparency International, is another macro indices used to measure corruption. In fact, it “evaluates the supply side of corruption – the likelihood of firms from the world’s industrialised countries to bribe abroad” (Bribe Payers Index, 2015). First launched in 1999, it is based on 2 Bathtub Model of Corruption 64 As illustrated in table 1, the CoC-, CPI- and the ICRG-data are highly correlated with each other. That means they can be regarded as valid instruments measuring corruption in similar ways. Thus, it does not matter which index is used to analyze corruption. However, the highest correlations can be observed between the CoC and the CPI (r=0.97). Overall, the correlation coefficients can be observed in the following table 1. In the following analysis the CPI is primarily used to measure corruption at the macro level. In contrast to the Control of Corruption Index that has been published biannually from 1996 to 2002, the CPI has been collected every year since 1995 and has become one of the most reliable and widely used indicators of corruption around the world (Rose-Ackerman, 1999; Treisman, 2000; Sandholtz and Koetzle, 2000). Correlation Matrix of certain Corruption Scores Control of Corruption Corruption Perceptions Index ICRG Indicator of Quality of Government Control of Corruption Pearson's correlation 1 .97*** .90*** Significance (two-tailed) .000 .000 N 2037 1452 1370 Corruption Perceptions Index Pearson's correlation .97*** 1 .89*** Significance (two-tailed) .000 .000 N 1452 1906 1382 ICRG Indicator of Quality of Government Pearson's correlation .90*** .89*** 1 Significance (two-tailed) .000 .000 N 1370 1382 3272 Source: Own calculations. Table 1: the views of more than 3000 business executives worldwide and measures to what extent companies engage in bribery when doing foreign business. In 2011, the BPI scored almost 30 countries and territories of the world’s leading economies selected by four criteria: trade openness, measured by the values of their foreign direct investment outflows; G20 membership, the value of their exports; and their regional significance (Bribe Payers Index, 2015). The results based on using this index, however, are primarily focused on analyzing corruption in the private, and not the public sector, that I concentrate on in the following considerations. Furthermore, the BPI provides data only for a few countries. For these reasons, the BPI is not included in this analysis. As expected, my analysis has indicated that the BPI does not correlate with the other corruption macro indices. 2.3 Measuring Corruption: The Dependent Variable Reconsidered 65 Critical Evaluation of Using Macro Corruption-Indices Although, macro corruption indices present a useful instrument to measure corruption, they are flawed and limited in a number of ways. Working with these data, these limitations have to be accounted for and results have to be interpreted cautiously. First, the macro corruption indices do not reflect the actual level in a country because they only measure the perception of corruption rather than actual occurrence of corrupt activities. They consist of highly subjective evaluations that, in turn, can be influenced by a wide range of different factors such as delays and incompetence. For instance, it is assumed that actual experiences with corruption are not always truthfully and reported in their entirety. Often, respondents may not feel comfortable admitting openly that they have bribed a public official such as a judge or a court staff. Moreover, they may be reluctant to criticize their own institution or profession by indicating corruption among colleagues. For these reasons, it is advised to consider both ‘experience’ and ‘perception’ (Langseth, 2006, p. 35). Golden and Picci (2005, p. 39) suggest that “Respondents directly involved in corruption may have incentives to underreport such involvement, and those not involved typically lack accurate information. This is an intrinsic weakness to measuring corruption with survey information, especially when the surveys do not ask about firsthand experiences with corruption, but merely ‘perceptions’ of it.” Second, the indices use aggregate information on corruption from different surveys. For countries where information from as many as 14 surveys is available, the scoring is likely to be more reliable for countries whose score based on a smaller number of available surveys. For the CPI index, for example, at least three primary surveys or sources for corruption need to be available for each country to be included in the index. This is likely to generate systematic biases in the different datasets, possibly making the indices more reliable for developed nations than for less developed ones (Golden and Picci, 2005). Moreover, due to the aggregate nature of the macro data researchers can make only a few conclusions about the relationship between corruption and individual actors. Corrupt actions take place between individuals, however, these indices present statistical aggregation of secondary data to measure the extent of corruption. Therefore, most importantly, macro indices cannot satisfactory explain within-country variations in corruption on a conceptual level. Consequently, the indices cannot reflect variation in 2.3.2 2 Bathtub Model of Corruption 66 different sectors or regions of each country. Montinola and Jackman (2002, p. 12) point out that “Researchers should thus bear in mind that a country’s score on the CPI indicates the average level of corruption among the institutions within a country explicitly or implicitly considered by respondents of each of the different surveys. Since each survey may be measuring corruption in a different sector of a country, TI’s use of varying numbers of surveys for different countries in any given year presents a potential methodological weakness. Researchers engaged in cross-country analyses using the CPI might consider replicating their work with other corruption indices to ensure robust findings.” Further criticisms refer to the cross-cultural portability of the concept of corruption. In fact, the perception of corruption varies from country to country, implying that an action that is perceived as corrupt in European states might be considered as ordinary practice in developing nations. Different cultures and various perceptions lead to different understandings of corruption and may distort the survey data (problem of functional equivalence). For example, Moroff and Blechinger (2009) illustrate that comparisons of corruption using notions of culture only to equivocate about the normative status of particular actions often underestimate the complexity and subtlety of corruption and culture alike. They use a comparative dataset describing the ways corruption events are covered in the news media of several democracies and demonstrate that different countries’ journalistic reports on corruption place significantly different amounts of emphasis on various aspects of corruption. In addition, most of the macro corruption-indices do not reflect the activity of individuals who abstain from corrupt activities in their home country, but engage in them abroad. A large number of examples exist illustrating that public officials act correctly in their own system, but pay bribes abroad to achieve particular goals (Salbu, 2001). Additionally, the reliability of these indices may diminish over time (Golden and Picci, 2005). Respondents who participate in corrupt actions might tend to disguise their involvements, whereas people who are not involved are not able to give accurate information.40 As a result, it is difficult to use them in time series and pooled analyses. In Lambsdorff’s 40 Moreover Treisman attributes that “Ratings by international business people and experts, disproportionately drawn from developed Western countries, might be influenced by Western preconceptions or by the raters’ greater familiarity with certain cultures. Some of the organizations that prepare corruption ratings might also 2.3 Measuring Corruption: The Dependent Variable Reconsidered 67 words: ”year-to-year changes may not only result from a changing performance of a country […] changes can result from the different methodologies […] not necessarily from actual changes” (Lambsdorff, 2005, p. 1). For instance, a country’s scores on the Control of Corruption Index change because of three factors that compound interpreting the scores: changes in the underlying data, the addition of new data sources, and changes in the weights that are used to aggregate the micro-data. Therefore, by using macro level indices such as the Corruption Perceptions Index or the Control of Corruption Index, researchers have to carefully take into account that these indices provide an “annual snapshot” of different viewpoints, “ […] with less of a focus on year-to year trends” (Lambsdorff, 2005, p. 1). Following a panel-data research design, I am very aware of this limitation. Yet, there is still no corruption data available that can be used to demonstrate how countries change their corruption scores. Lastly, the indices do not distinguish between various forms and typologies of corruption, such as fraud or embezzlement. For instance, the survey-based data are often limited to passive corruption and are not dealing with active corruption. Moreover, Galtung (2006) argues that the definition of corruption used by Transparency International does not differentiate between corruption in different branches of civil services or in political party financing. In his article “Measuring the Immeasurable: Boundaries and Functions of (Macro) Corruption Indices” (Galtung, 2006) illustrates the limitations of the CPI grouped under seven headings, briefly summarizing the critique: 1. Only Punishing the Takers, not the Givers or Abetters 2. Irregular and Uncontrolled Country Coverage 3. Biased Sample: more than 90% of the world is missing 4. Imprecise and Sometimes Ignorant Sources 5. Far Too Narrow and Imprecise a Definition of Corruption 6. Does not Measure Trends: Cannot Reward Genuine Reformers 7. Guilty by Association – Aid Conditionality Galtung (2006, p. 17) concludes that “In particular, the CPI needs to be complemented by other indicators to address vital aspects of the subject that a single index can never hope to capture.” Nonetheless, despite their limitations, the indices have produced a challenging new generation of have ideological axes to grind. For all these reasons, ‘perceived corruption’ may reflect many other things besides the phenomenon itself” (Treisman 2007, p. 215). 2 Bathtub Model of Corruption 68 quantitative research that is successfully used and recommended by a lot of researchers (Treisman, 2000; Lancaster and Montinola, 200141; Møller and Skaaning, 2009). In sum, even though the macro level indices are flawed and limited in a number of ways, they present one of the most accessible and valid methods to analyze corruption. The data is available for a long period of time and includes a large country sample. Furthermore, it has the best construct validity and cross-country comparability and while no measure of corruption will ever be free of problems, and while the current scales have enriched the literature and research in major ways, there are ways to further improve their quality (Littvay and Donica, 2011). In the following analysis, I use the Corruption Perceptions Index of Transparency International as the prime data source for measuring corruption at the macro level. Comparatively, the CPI offers most data on corruption. Additionally, to check the robustness of measurement the Control of Corruption Index of the World Bank was applied as an additional control variable in previous analysis of corruption (see Kubbe, 2013). The Individual Level: Micro Level Indices Similar to macro level indices, scholars have attempted to find some indirect ways to identify corruption level at the individual level. These investigative approaches are manifold and very creative. They generally include the use of national survey and experimental data which obtain corruption experience data at the micro level, convictions from the judiciary, and public expenditure tracking surveys such as data on infrastructure projects. The most widespread method is, however, the use of survey data of individuals. Concentrating on survey observations, Boylan and Long (2003) and Alt and Lassen (2003) studied state house reporters’ perceptions of corruption in the United States. State house reporters who are members of the press 2.3.3 41 Lancaster and Montinola (2001)’s article “Comparative Political Corruption: Issues of Operationalization and Measurement” gives a great overview of the advantages and disadvantages of using corruption indices. See also Johnston (2009) who gives in his article “Measuring the New Corruption Rankings: Implications for Analysis and Reform” a critical analysis of the different corruption indices. He especially focuses upon the CPI. 2.3 Measuring Corruption: The Dependent Variable Reconsidered 69 and cover state government activities were asked to assess their state in terms of corruption levels of all government employees. Boylan and Long (2003) have demonstrated that these corruption measures are reliable and valid by illustrating that federal corruption prosecutions are positively correlated with both corruption and prosecutorial effort. In a similar vein, Seligson (2002) used national survey data of over 9.000 individuals that are conducted in four Latin American countries. They measured corruption by asking eight questions about participants’ experience with corruption over the year prior to the survey. These included: “(1) being stopped by a police officer for a trumped-up infraction of the law; (2) being asked to pay a bribe to a police officer; (3) observing a bribe being paid to a police officer; (4) observing a bribe being paid to a public official; (5) being asked to pay a bribe to a public official; (6) being asked to pay an illegal fee to expedite a transaction at the municipal government; (7) being asked to pay a bribe at work; and (8) being asked to pay a bribe in the court system” (Seligson, 2002, p. 388). Likewise, Mocan (2008) use survey data from more than 55.000 individuals from 30 countries that provide information about direct experiences with bribery. Individuals were asked whether any government official such as a government workers, police officers, or inspectors in that country have asked or expected them to pay a bribe for their services during the previous year. Subsequently, Mocan (2008) created an aggregate corruption index at the macro level using information provided by more than 90.000 individuals in the data set. Likewise, Atkinson and Seiferling (2006) measured corruption by using the item “How widespread do you think bribe taking and corruption is in this country?” of the World Values Survey from a representative sample of over 30.000 respondents in 33 countries. By adding national economic and cultural variables they created a data set of both micro and macro level information. Overall, they found that public perceptions of corruption highly correlate with expert reviews such as the CPI of Transparency International. The correlation between their created index and the Kaufmann, Kray Zoido- Lobaton Governance indicators 1998 was r = -.81 (Atkinson and Seiferling, 2006). As an alternative way to survey data, Glaeser and Saks (2006) focus on convictions from the U.S. Justice Department Report to Congress on the Activities and Operations of the Public Integrity Section for the period of 1976-2002. They assume that the number of public officials convicted for misuse of public office might be an indicator for the objective level of corruption in a country. Similarly, Meier and Holbrook (1992) illustrate the 2 Bathtub Model of Corruption 70 effect of education and urban concentration on corruption in American states. They gathered data by the U.S. Department of Justice's (1988) Public Integrity Section and used the number of public officials in each state who are convicted for violating laws involving corruption per one hundred elected officials in that state. Similarly, Goel and Nelson (1998), claiming that state intervention and public spending foster rent-seeking activities and consequently corruption, also use state-level observations of convictions for public office abuse. Public office abuse includes offenses such as accepting bribes to accelerate the duties of officials or to grant preferential treatment in awarding government contracts. The data set consists of annual observations for the 50 U.S. states from the U.S. Department of Justice over the period from 1983 to 1987 (similar Fisman and Gatti, 2002). Finally, Goel and Nelson (1998) relate this corruption variable to the total real per capita expenditures of local governments and show a significant, positive relationship between the variables. However, Lambsdorff (1999, p. 1) argues that “the correlation might be explained differently. As governments increase their spending, the judiciary branch may also be allocated more funding, resulting in higher conviction rates. In this case, conviction rates are not an adequate indicator for the actual incidence of corruption, but rather, reflect the quality of the judiciary.” Another alternative of measuring corruption at the micro level has been conducted by Golden and Picci (2005) who suggest a proxy for corruption including “the difference between the amounts of physically existing public infrastructure (roads, schools, hospitals, etc.) and the amounts of money cumulatively allocated by government to create these public works. Where the difference between the two is larger, more money is being lost to fraud, embezzlement, waste, and mismanagement; in other words, corruption is greater” (Golden and Picci, 2005, p. 37). They created this index with subnational data from 95 provinces and 20 regions in Italy. However, this approach is virtually impossible when determining an accurate pricing of actual infrastructure costs, especially in a comparative perspective. Additionally, experimental research on corruption has grown in the last years and it is still in its infancy (Cameron et al., 2009; Banuri and Eckel 2012). Prior experimental studies have focused on individual determinants of corruption and consider the influence of an individual’s gender (Frank et al., 2011; Rivas, 2013), religion (Armantier and Boly, 2010), culture, amount of wages (Azfar and Nelson 2007; van Veldhuizen 2011), the amount of bribe, level of monitoring and punishment (Frank and Schulze, 2000; Schulze and Frank, 2003). However, like any other method also ex- 2.3 Measuring Corruption: The Dependent Variable Reconsidered 71 perimental research suffers from some methodological weaknesses such as the artificialness of experimental settings, validity and ethics. Moreover, there is still no experimental data available for several countries and individuals that can be used for comparative analyses. Nevertheless, they present a complementary approach to other social science methods and studies that research on corruption. Critical Evaluation of Using Micro Corruption-Indices Overall, scholars have been able to find some creative ways to identify corruption levels at the individual level. Yet, as the described studies and their certain approaches illustrate, there is still no adequate instrument to measure corruption at the micro level in a direct way. Consequently, these indices have to deal with some restrictions that relate particularly to their predications. It reveals that measuring corruption at the micro level deals with similar problems as capturing corruption at the macro level. These limitations particularly refer to socially desirable responses and subjective evaluations of corruption. For instance, due to socially desirable responses, survey or self-reporting data are not necessarily reliable. Such data also measure only the perception and not the actual level of corruption and include, therefore, highly subjective evaluations. Moreover, they cannot exactly identify the actors who are involved in corruption as well as the frequency of corrupt actions. Mainly, they are limited to capture passive corruption and are not dealing with active corruption. For instance, it is difficult to find out through asking direct questions, who has taken the role of a principal, agent or client, in other words, who is the provider of bribery, who receives the money or who is the middleman. Almost nobody admits that in a transparent and honest way. Moreover, these micro indices can often not distinguish between various forms and types of corruption such as petty or grand corruption. In addition, criticisms also include cross-culturally diverse corruption, implying that various perceptions of corrupt activities based on different cultural background lead to different understandings of corruption. Another way to measure corruption at the micro level is the use of so called objective measures such as criminal indictments or public expenditure tracking surveys (Boylan and Long, 2003). However, by using these objective measures problems of validity often arise. For instance, criminal 2.3.4 2 Bathtub Model of Corruption 72 indictments only tend to indicate the effectiveness of the judicial system rather than the actual level of the corruption in a country. Boylan and Long (2003) also state that, while data of federal prosecution offer information about the level of corruption, the number of prosecutions also illustrate the priority or amount of effort devoted to prosecution of public officials, which also varies by state. Additionally, the number of public officials convicted also includes convictions that are not specifically related to corruption. Nevertheless, this investigative approach should not be underestimated, implying that it is one possible strategy or an additional instrument to narrow down the measurement of corruption. However, contrary to macro indices, data from the micro level can reflect the activity of corrupt individuals and variation of corruption in different sectors or regions of each country. Nonetheless, despite their limitations, the use of corruption micro data is justified because the actual level of corruption in a country is difficult to investigate. Moreover, Charron (2015) shows a remarkably high consistency between actual reported corruption, as well as citizen and expert perceptions of corruption. Using survey data based on 85 000 European respondents in 24 countries, he concludes that “although existing corruption measures certainly have their share of problems, concerns regarding the validity and bias of perceptions have, perhaps, been overstated” (Charron 2015, p. 1). In sum, micro level approaches present a valid method to explore corruption. In most cases, micro level data is easily available and often includes large country samples, such as the World Values Survey. Although embryonic, corruption surveys appear to be a promising tools to measure corrupt activities. Therefore, combining macro and micro data presents an appropriate strategy to find out the causes of corruption in a comprehensive way. Variables and Research Hypotheses: Aggregate and Individual Levels Combined The Macro Level: Economic, Political, Socio-Cultural and Historical Factors Following theoretical and empirical arguments from the previous corruption literature and the bathtub model that serves as a heuristic instrument, the emergence of corruption can be explained by a number of country 2.4 2.4.1 2.4 Variables and Research Hypotheses: Aggregate and Individual Levels Combined 73 characteristics. In the following, the indicators for explaining the causes and variations of corruption are categorized into economic, political, socio-cultural and historical context factors. Even though, previous studies have analysed and discussed some of these variables and their effects on corruption, the current research has not taken all of these variables into account for the European states. In the following section, I describe and justify, how I derive these variables and hypotheses from the theoretical and empirical corruption literature that I will use in the subsequent analyses. It should be stressed, that these factors are also linked with a country’s development of democracy. To avoid an overestimation of the model, I reduce the independent variable to as few as possible. In previous research, I have included further variables that are excluded in this study because of multicollinearity, no significant relations or using proxy variables (see Kubbe, 2013). Economic Factors Economic factors include a number of different parameters. Following a rational-choice logic, authors assert that economic benefits are the root of most forms of corruption in modern societies and that this strong incentive of controlling economic resources or power motivates corruption (e.g. Rose-Ackerman, 1999; Basu, 2006). In the following section, I will discuss economic indicators and its relationships more closely and point out possible correlations with corruption scores. They include, in particular, a country’s level of economic development and its EU-membership. Economic Development Many studies show that socio-economic development, or modernization writ large, is the most decisive determinant of corruption. Usually measured by GDP, it strongly reduces the development of a society’s corruption (Husted, 1999; Shabbir and Anwar, 2007; Littvay and Donica, 2011). For instance, Gerring and Thacker (2005, p. 242) assert that “Wealthier countries are likelier to be less corrupt, and less corrupt nations are also more apt to perform well economically.” It is even claimed that the degree of a country’s economic development holds most of the explanatory power of the various corruption indicators (Treisman, 2000; Paldam, 2002). Basu 2.4.1.1 2 Bathtub Model of Corruption 74 (2006) asserts that economic benefits are the root of most forms of corruption in modern societies and serves as a strong incentive of controlling economic resources. Data on corruption development in Europe show, that although it is increasing over time, corruption is still lower compared to developing economies. In most cases, the least corrupt countries are also economically well developed (e.g. Denmark, Norway, Sweden) (MacDonald and Majeed, 2011. This implies that individuals who are in good economic situations are not dependent on bribery payments. It has also been demonstrated that a country’s economic development is closely linked to its democratic development (Przeworski, 2008), implying that more economically developed countries show higher levels of democracy that, in turn, might reduce a country’s level of corruption. The link between economic development and democracy is considered “one of the most powerful and robust relationships in the study of comparative national development” (Diamond 1992, p. 110). For Europe, I hypothesize that the extent of corruption will be higher in countries with lower levels of economic development. EU-Membership Following economic and sociological approaches, I suppose that a country’s international involvements affect its extent of corruption. In particular, countries that are more integrated into Western international networks of exchange, communication, and organization, are more exposed to both economic and normative pressures against corruption (Sandholtz and Gray 2003; Kostadinova 2012). On the one hand, international integration can offer economic incentives, altering the costs and benefits of engaging in corrupt acts for various actors. On the other hand, a country’s participation in international organizations affects corruption levels in a normative way, by creating channels and informal rules for the diffusion and absorption of anti-corruption norms to other member countries. Using data from approximately 150 countries (Sandholtz and Gray, 2003) empirically demonstrate that greater degrees of international integration lead to lower levels of corruption. They measured international integration primarily by the total number of years of membership in several Western international organizations such as the International Monetary Fund (IMF), the United Nations (UN), the General Agreement on Trade and Tariffs, the World Trade Organization (WTO) or the Organization for Economic Co-operation and 2.4 Variables and Research Hypotheses: Aggregate and Individual Levels Combined 75 Development (OECD).42 They theorize that such international involvements affect the degree of corruption in a country by offering economic incentives, altering the costs and benefits of engaging in corrupt acts for various actors, on the one hand, on the other hand, in a normative way, by creating norms and values in international society that stigmatize corruption. In other words, by considering countries as rational actors that follow economic incentives, these assumptions either refer to rational-choice approaches, or apply to cultural explanations by observing the prevailing norms and values in international societies. In particular, countries that are more integrated into Western international networks are more exposed to both economic and normative pressures against corruption. The authors assume that the more a country is tied into international networks of exchange, communication, and organization, the lower its extent of corruption is likely to be, because participation in international organizations creates channels for the diffusion and absorption of international anti-corruption norms. The analysis by Sandholtz and Gray (2003) strongly confirms this expectation. In a previous study using a multivariate model, Sandholtz and Koetzle (2000) also found that the lower the degree of integration in the international economy, the higher the extent of corruption. They measure the degree of international integration, particularly by the involvement in trade and illustrate that greater trade, involvement influences both the political-economic structure of opportunities and the cultural norms of a country. Kostadinova (2012) confirms these results by providing evidence that integration in the European Union has had a significant influence on reducing a country’s extent of corruption. Using data from Eastern Europe, she demonstrates that countries, especially in Eastern Europe, that had not signed agreements for integration with the EU until 1996 have higher degrees of corruption values than countries that have signed the agreements. Thus, her data confirms that “the mobilizing effect of association and probable EU membership is independent from domestic political and economic factors” (Kostadinova, 2012, p. 56). For Europe, I posit that the extent of corruption will be higher, if the country is not a member of the European Union. 42 Further indicators that are used to measure international integration are the gross foreign direct investment per capita, international telephone minutes per capita, international air freight per capita, international air passengers per capita and trade openness (Sandholtz and Gray, 2003). 2 Bathtub Model of Corruption 76 Further Economic Factors Moreover, other variables, such as a country’s rate of inflation, unemployment rate, civil service wages, WTO- and OECD-membership were originally included in the analysis (Kubbe, 2013). The rate of inflation that can be used as a proxy variable for economic development is more likely to enhance corruption over time. In contrast to this finding, a country’s unemployment rate, WTO- and OECD-membership are significant in a socalled “economic model of corruption”, but they do not remain robust in an “overall European-specific model” (Kubbe, 2013, p. 197). The variable civil service wages does not show a relationship with the extent of corruption at the macro level, suggesting that there is no relationship between the extent of corruption and government expenditures on wages and employer contributions. Furthermore, some economic variables had to be excluded because of multicollinearity with other indicators such as the degree of democracy. Moreover, I aim to provide a very sharp model of the factors that affect corruption. That is why I try to reduce my models to as few as possible indicators. Another important variable, that could not be included, is social inequality, usually measured by the gini-index. Previous research indicates a positive relationship between the extent of corruption and inequality. Uslaner (2010, pp. 48-49), for instance, claims that “Inequality leads to corruption because it leads to resentment of out-groups and enhanced in-group identity. Economic equality, in turn, promotes both optimism and the belief that we all have a shared fate across races, ethnic groups, and classes.” However, in this case, sufficient data on social inequality is not provided for all included European states and the respective investigation period. Political Factors Even if economic variables have been the most cited and probably the most influential explanations of corruption, the literature indicates that political factors, such as the degree of democracy and other related variables such as the rule of law, have to be considered. As already mentioned, especially economic approaches view the causes of corruption as deficiencies in the political system and its institutions. Thus, corruption is to be determined by political systems which are deficient in democratic power- 2.4.1.2 2.4 Variables and Research Hypotheses: Aggregate and Individual Levels Combined 77 sharing formulas, checks and balances, accountable and transparent institutions and procedures of the formal and ideal system of democratic governance. Therefore, corruption is conceived as a symptom of poorly functioning systems and as the failure of democracy, ethical leadership and good governance (Doig and Theobald, 2000; Paldam, 2002; Shah, 2007; Fjelde and Hegre, 2014). The political factors that are presumed to have an influence on the extent of corruption include the degree of democracy and the percentage of women in parliaments. Degree of Democracy In the context of political factors, the degree of democracy and the quality of governmental institutions might be one of the most important contributors to corruption. Simultaneously, the degree of democracy is one of the most complex variables because it includes several sets of institutions, procedures, and values that may significantly reduce the extent of corruption. Overall, there is plenty of literature on the democracy-corruption-nexus. In a nutshell, most researchers argue that more advanced democratic structures lead to a lower degree of corruption by creating anti-corruption norms (Saha et al., 2014). Hill (2003), for instance, provides evidence for a strong relationship between higher democratization and lower levels of corruption in the U.S. American political system. He also demonstrates that specific components of democratization may enhance transparency in government and that probity among public officials is a good predictor of lower corruption levels. Shah (2007, p. 242), who refers to case studies of the World Bank, asserts that corruption is high where the rule of law is weakly embedded: “Public sector corruption thrives where laws apply to some but not others and where enforcement of the law is often used as a device for furthering private interests rather protecting the public interest.” Moreover, he points out that “societies in which the level of public sector corruption is relatively low usually have strong institutions of participation and accountability that control abuses of power by public officials. These institutions are either created by the state itself (for example, electoral process, citizens` charter, bill of rights, auditors general, the judiciary, the legislature) or arise outside of formal state structures (for example, the news media and organized civic groups)” (Shah, 2007, pp. 242–243). Examining different 2 Bathtub Model of Corruption 78 states within the U.S., Alt and Lassen (2008) show that institutional separation of power may hinder the extent of corruption. They illustrate that divided government institutions provide a system of checks and balances between the executive and legislative branches and that elected, rather than appointed, state supreme court judges show lower degrees of corruption. The authors conclude that the effect of an accountable judiciary is stronger under a unified government, where the government cannot control itself. Similarly, Ades and Di Tella (1997) and Shah (2007) find that corruption is higher in countries where judicial institutions are not well developed, or are not independent of political influence. Using cross-country data, the analysis of Damania et al. (2004) tests the hypothesis that the institutions necessary to monitor and enforce compliance are weak in politically unstable regimes (see also Serra, 2006). In such countries, corruption is therefore more pervasive, and the compliance with regulations is low. They conclude that political instability decreases judicial efficiency and in turn fosters corruption. Thus, the effect of political instability on corruption is not direct, but occurs indirectly via its effect on the degree of judicial efficiency (similar Ali and Isse, 2003). However, democracy does not guarantee honest and transparent governments (Uslaner, 2002; Shen and Williamson, 2005). Treisman (2000) does not find a significant correlation between levels of corruption and democracy, but he shows that a long period of exposure to democracy appears necessary to reduce levels of corruption. Montinola and Jackman (2002) and Sung (2004), for example, illustrate that in general corruption seems to be typically lower in dictatorships than in partially democratized countries. However, with more complete democratization, usually reflected in the nature of elections and the effective power of elected legislators, countries show lower levels of corruption. In fact, the prospects for corruption are more likely to be lower in consolidated democracies than in non-democracies as well as in unstable or young democracies. The political process can most easily become corrupt at the early stages of transition to democracy. When democratic institutions are weak, public officials often use their entrusted power to obtain private gains for themselves and their business partners. Nur-tegin and Czap (2012) present similar results and find strong empirical evidence that democracies, even if they are politically unstable, have public officials who are less corrupt than officials in authoritarian countries. In this context, Uslaner (2002, p. 18) summarizes that “making countries more democratic does not seem to make them less 2.4 Variables and Research Hypotheses: Aggregate and Individual Levels Combined 79 corrupt”, while Kreuzer (1996, p. 110) mentions that “Democratisation does not reach a final destination of full democracy, but constitutes an ongoing process of shielding and weaning democracy from political corruption.” In a similar vein, Rose-Ackerman (1997, p. 40) states that “Democracy gives citizens a role in choosing their political leaders. Thus corrupt elected officials can be voted out of office. But democracy is not necessarily a cure for corruption.” Based on previous research, I assume that more advanced democratic structures lead to a lower extent of corruption in European countries. Women in Parliaments The impact of gender on corruption has long been neglected in corruption research. Swamy et al. (2001) and Dollar et al. (2001) were one of the first scholars who have analyzed the relationship between the percentage of women in the labor force as well as in parliaments and the extent of corruption. Using cross-country data Swamy et al. (2001) show that women are less involved in corrupt transactions and are less likely to condone bribe-taking than men. Moreover, they illustrate that a higher female labour participation leads to less corruption in general. Following Gottfredson and Hirschi (1990) and Paternoster and Sally (1996), Swamy et al. (2001) provide four arguments to explain this finding. First, women seem to be more honest or more risk-averse than men by nature, which may be because they feel that there is a greater probability of being caught. Second, they are typically more involved in raising children, an activity in which they practice honesty in order to teach their children appropriate values. Third, it is assumed that “women may feel more than men- the physically stronger sex, that laws exist to protect them and therefore be more willing to follow rules.” Fourth, “girls may be brought up to have higher levels of self-control than boys which affects their propensity to indulge in criminal behaviour” (Gottfredson and Hirschi, 1990, p. 149). Dollar et al. (2001), who reviews behavioral studies, show similar results and also conclude that women are more trustworthy and public-spirited than men. Using data on female involvement in government from the inter-parliamentary union’s survey (1945-1995), these authors have been able to show that greater representation of women in parliaments may lead to lower levels of corruption. 2 Bathtub Model of Corruption 80 However, both Treisman (2007) and Sung (2003) doubt the conclusions of Swamy et al. (2001) and Dollar et al. (2001), claiming that “The logic is somewhat unclear and seems to run into problems of ecological inference. […] those women who get elected to parliament or appointed ministers are unlikely to be typical of those surveyed. In any case, politicians of either gender are likely to act in office in ways not fully consistent with their answers to abstract questions about professional ethics; and even if most female politicians preferred lower corruption, how a marginal increase in the female share in either the legislature or government could be expected to produce lower corruption at ground level is unclear” (Treisman, 2007, p. 238). Sung (2003) also argues that the correlation is spurious, caused by other aspects of liberal democracy going along with the protection of women’s political rights. She claims that it is ‘fairer systems’, not women’s greater integrity that explains why corruption is lower where more women are in government. Alatas et al. (2009) also suggest that the gender differences found in the previous studies may not be nearly as universal as claimed and may be more culture-specific. They conducted experiments in Australia, India, Indonesia and Singapore and did not find any significant differences between the attitudes of men and women towards corruption. However, larger variations were found in women’s attitudes towards corruption across different countries compared to men in general, indicating a stronger cultural rather than gender-based explanation (similar Alhassan-Alolo, 2007). Other reasons might explain why women seem to be averse towards corruption, such as work seniority and positions. For example, Lambsdorff and Fink (2006) find that female transit police officers seem to be more difficult to bribe than their male colleagues. It is assumed that they are afraid of accepting bribes because they have only been admitted to the police force recently and therefore have less seniority than their male colleagues. Moral aversion to illegally accepting bribes may be another explanation. Rivas (2013) shows similar results by using experimental methodology. Experiments were conducted in Spain with undergraduate students of the University in Barcelona. He has found that the percentage of male that decided to offer a bribe to public officers at least once was 80%, while the percentage of women that did so was 65%. Moreover, the experiments have demonstrated that men offered bribes more frequently to 2.4 Variables and Research Hypotheses: Aggregate and Individual Levels Combined 81 female than to male officials, although they offered them lower bribes.43 In a similar vein, Frank et al. (2011) show in their corruption experiments that women who are involved in potentially corrupt transactions are more likely to fail. They claim that “The reason is not that women are intrinsically more honest, but that they are more opportunistic when they have the chance to break an implicitly corrupt contract and less engaged in retaliating non-performance” (Frank et al., 2011, p. 59). Therefore, women are not more averse to corruption than men, but comparatively tend to react more strongly to a given risk of detection. This result is confirmed by a study of Agerberg (2014) using data on governance in Europe. The data is a result of a survey of corruption conducted at the regional level in the EU, in 2010 and 2013, among 23 European countries. The survey results show that on average, women perceive levels of corruption as worse, pay fewer bribes, have a lower tolerance for corrupt actions, but are also given fewer opportunities to engage in corrupt actions than men. The study also supports previous results revealing that higher gender equality influences corruption levels in a positive direction. For the European states, I expect that the extent of corruption will be lower in countries that have higher levels of female participation in parliaments. Further Political Factors Further variables that were initially included in this analysis encompass a country’s anti-corruption policy, degree of government centralization, degree of political competition, degree of public spending ratio, the rule of law in a country and freedom of the press (Kubbe, 2013). For instance, a country’s anti-corruption policy shows a significant relationship with the extent of corruption. My results suggest that the admission of countries into organizations with high anti-corruption standards such as the European Union seems to be an overall efficient anti-corruption instrument because international pressure tends to produce behavioral changes in countries regarding their corruption levels. However, in the following analysis, I do not include the variable because a country’s EU-membership seems to be 43 However, Mocan (2008) using data from almost 50 countries illustrates that males are more frequent targets of bribery than women, possibly because in most countries, males are still more active than females in the labour market for several reasons, and therefore, in turn, more often work as governmental officials. 2 Bathtub Model of Corruption 82 a good proxy for anti-corruption policies of states. The hypothesis that high degrees of government centralization lead to greater extents of corruption is confirmed in the so-called “political model of corruption” for European states, however, it does not remain robust in an overall model of factors (Kubbe 2013, p. 159). Surprisingly, the variables of the degree of political competition and public spending ratio are not significant in the European sample. The variables rule of law in a country and freedom of the press are not included in the analysis because these variables are strongly related to a country’s degree of democracy. Lederman et al. (2005) report that variables such as parliamentary systems, political stability and freedom of the press are associated with lower levels of corruption. Besley et al. (2002) and Brunetti and Weder (2003) also find a strong negative relationships between corruption and free press or unbribable media.44 They suggest that a free press might expose corrupt activities and serve as a restraint on public officials such as politicians (similar Mungiu-Pippidi, 2013). However, with regard to the appearance of multicollinearity, these variables could not be included in the analysis. Socio-Cultural Factors Socio-cultural factors have increased in importance in the research of corruption. Already in the 1960s and 1970s, the impact of socio-cultural factors on corruption was stressed. For instance, in the 1970s, Huntington (1968) claimed that corruption is to a large degree a cultural phenomenon. He argues that “Corruption may be more prevalent in some cultures than in others but in most cultures it seems to be most prevalent during the most intense phases of modernization.” He understands modernization as “a change in the basic values of the society” (Huntington 1968, p. 492). 2.4.1.3 44 For instance, using three different sources of corruption, Besley et al. (2002) find a robust negative correlation of corruption with foreign ownership of the media. They interpret their results as evidence that foreign ownership may be correlated with indicators that make the media a more effective information-generating instrument. Brunetti and Weder (2003) also indicate that freedom of the press might control corruption. They find a significant relationship between more press freedom and less corruption in cross-section of countries and present results suggesting that the direction of causation runs from higher press freedom to lower corruption (similar Ahrend, 2002 and Chowdhury, 2004). 2.4 Variables and Research Hypotheses: Aggregate and Individual Levels Combined 83 According to sociological approaches, it is now often assumed that mainly social structures and cultural values determine the extent of corruption (Paldam, 2002). For instance, a number of case studies have found that societies with strong family or clan-based loyalties show high levels of corruption (Theobald, 1990). At the macro level, this group of factors captures the dominant religion of a society (Catholicism, Orthodox, Protestantism, Islam) that may have an impact on the level of corruption. Religion Religiosity is an important indicator for explaining the emergence of corruption. Dreher et al. (2007, p. 448) theorizes that “religion may shape social attitudes towards social hierarchy and family values and thus determine the acceptability, or otherwise, of corrupt practices. In more hierarchical systems (for example, Catholicism, Orthodoxy and Islam), challenges to the status quo are less frequent than in more egalitarian or individualistic religions.” The relationship between religion and corrupt behavior is notably explored by Treisman (2000). He regresses corruption on the percentage of Protestants in the total population and concludes that a Protestant tradition appears to have a negative (though small) effect on corruption, by controlling for variables such as a country’s economic development. Similar results are presented by Bonaglia et al. (2001), Gerring and Thacker (2005) or Serra (2006). A more in-depth analysis of the impact of religion is provided by Paldam (2001), who identifies eleven different groups of religions and tests their impact on corruption. While in countries with a large fraction of Reform Christianity and Tribal religion, corruption is lower, higher levels of corruption can be found in countries with a large influence of Pre-Reform Christianity, Islam, Buddhism and Hinduism. However, the impact is only significant for Reform Christians (Protestants and Anglicans). Similarly, La Porta et al. (1999) illustrate that countries with a high proportion of Catholics or Muslims reduces a country’s quality of government and, by extension, may reduce the extent of corruption. In sum, studies demonstrate that countries with larger proportions of Protestants tend to be less corrupt than traditionally Catholic countries. Theoretically, this association is often ascribed to egalitarian and individualistic features of Protestantism that facilitate the extent to which office-holders are held accountable for their actions. Thus, compared to other religions such as the Orthodox and Catholic churches as well as Islam, Protestant 2 Bathtub Model of Corruption 84 societies show less hierarchy and are less prone to tolerance towards power abuses and corrupt behavior. Additionally, the Protestant church has traditionally been separated from the state and played a role of opposition to the abuses of the government (Treisman, 2000). The Puritan aspects related to this religious tradition could also have a corruptionpreventing effect on both providers and receivers of bribery (Skaaning, 2009). Moreover, Protestants are less embedded in social networks that seem to be a breeding ground for corruption in other religions (Lambsdorff, 2002). Likewise, “Corruption belongs to a sinister informal network of giving and taking, demanding a basic form of trust. There are no contracts or actionable agreements. Corruption flourishes in well-established networks, whether it is a matter of having long-standing connections to building authorities or long-term supply contracts with large corporations. Since both parties may be guilty of a punishable offense, there is trust on both sides” (Alemann, 2004, p. 33). However, the direction of causality remains unexplored. For Europe, I posit that countries with higher levels of Protestants are likely to be less corrupt. Further Socio-Cultural Factors Further variables that have been included in prior analyses encompass a society’s degree of ethno-linguistic fractionalization, level of education and degree of urbanization (Kubbe, 2013). My findings indicate that the degree of ethno-linguistic fractionalization and the level of education have no influence on the extent of corruption in European countries. While the variable degree of urbanization shows a negative relationship with the extent of corruption in a so-called “socio-cultural model of corruption”, it is not significant in an “overall European-specific model” (Kubbe, 2013, p. 145). Historical Factors Corruption has numerous historical roots. When it once reached a certain level, it has been difficult to quickly reduce it again (Kostadinova, 2012). Sociological and historical institutionalists particularly emphasize that historical developments of institutions people operate in, and certain cultural values and traditions that have developed over a number of years can af- 2.4.1.4 2.4 Variables and Research Hypotheses: Aggregate and Individual Levels Combined 85 fect the level of corruption (North, 1990b; Thelen, 1999). Consequently, a country’s degree of corruption can be considered as path-dependent or even as cultural heritage. That implies that corrupt actions in the past affect corrupt actions in the present and in the future. These assumptions also refer to variables such as the degree of democratic consolidation and the duration of democracy in a country. Therefore, the historical variables that are included in the analysis encompass the durability of democracy and the communist past of a country. Years of Democracy Another variable that is assumed as an explaining factor of corruption is the durability of democratic systems. However, studies concentrating on the relationship between corruption and a democratic system’s durability are also very rare. Treisman (2000) was one of the first researchers who observed a significant impact of the distant past on the degree of corruption and illustrates that a long duration of democracies seems to be necessary to significantly reduce corruption. He explored that states that had been democratic systems constantly since 1950 show lower degrees of corruption. In other words and as already described, longer established democracies are less corrupt. However, the opposite has not been shown, namely that countries with high levels of corruption experience weakened democratization or democratic breakdown. Similarly Treisman (2000, p. 439) claims that “What matters is whether or not it has been democratic for decades. The regression estimates suggest a painfully slow process by which democracy undermines the foundations of corruption. Those countries with at least 40 years of consecutive democracy behind them enjoyed a significant, though small, corruption dividend, and those with 20-30 years may also have benefited slightly.” Blake and Martin (2006) also show that longitudinal measures of democracy have a strong association with the level of corruption measuring by CPI data from 1996 to 2000. By examining the raw, interval data and by creating a series of threshold levels between 10 and 20 years as an empirical metric of democratic consolidation in regard to corruption control, they used the number of uninterrupted years of democracy as indicator of the presence of a democratic system. In a similar vein, Pellegrini and Gerlagh (2008) show a negative relationship between a medium-long exposure to uninterrupted democracy (30 years) 2 Bathtub Model of Corruption 86 and corruption, whereas political instability leads to an increase of corruption. Based on these findings, I expect that European states with longer democratic histories will have lower levels of corruption. Communist Past Previous research indicates that the legacy of a country’s communist past has a strong impact on country’s corruption level (Rose, 2001; Miller et al., 2001; Treisman, 2003; Møller and Skaaning, 2009). According to this regional context, Kostadinova (2012, p. 26) claims that ”Because of the multifaceted character of postcommunist transition, numerous opportunities emerged for illicit payments, patronage, alllocation of public contracts, black market interactions, and covert networks. These could spread and grow in the Eastern Europe societies, already suffering from endemic bribery and lack of elite integrity.” To be more specific, Kostadinova (2012, p. 26) argues that “A legacy of ‘informality’ inherited from the communist era enhanced the formation of exchange networks operating through contacts and privileges. In this volatile environment, too many public officials preferred to stay loyal to ‘their colleagues and agency than to the state and society more generally’” (see also Holmes, 2006 and Karklins, 2005). Likewise, Sandholtz and Taagepera (2005, p. 114) argue that “Postcommunist states are susceptible to corrupt practices both because of the heritage of economic decision-making under communist rule and because of the vulnerability of privatization schemes to corrupt influences.” Based on World Values Surveys conducted in 46 societies between 1995 and 2001, they have empirically shown that there is a positive relationship between high levels of corruption and exposure to communist regimes and the adoption of communist structures and institutions, including certain social norms and values. They suggest, that “Communism created structural incentives for engaging in corrupt behaviors, which became such a widespread fact of life that they became rooted in the culture in these societies - that is, the social norms and practices prevailing in communist societies. The transitions toward democracy and market economies have not yet erased this culture of corruption” (Sandholtz and Taagepera, 2005, p. 109). In a similar vein, Gerring and Thacker (2005) suggest that countries that do not have a history of socialist rule tend to exhibit lower levels of 2.4 Variables and Research Hypotheses: Aggregate and Individual Levels Combined 87 corruption. However, Treisman (2003) cannot confirm the communismcorruption nexus. He finds no significant difference between post-communist countries and claims that “The higher average level of corruption in postcommunist countries seems to have little to do with postcommunism per se. Although these countries may be corrupt in distinctive ways, they are not corrupt for distinctive reasons. They have bad governments, largely because they are poor and lack a post-war history of democracy” (Treisman, 2003, p. 22). Moreover, opinion polls show that people do not generally blame communism for current corruption problems. The level of corruption is viewed by most either as a part of the moral crisis of transition or as a result of the country’s culture (Hutchcroft, 1997). Skaaning (2009, p. 226) even assumes that ”as culture only changes slowly, the corrupt traditions have arguably survived the end of communist regimes. Communism is thus likely to have established a negative legacy. New bureaucracies were not created from scratch, large extents of the personnel carried over, and enterprises as well as private people in general had 'internalized' certain practices.” Their analysis indicates that a communist past has no significant influence on the level of corruption. Using data from the World Values Survey from 64 societies from 1981-2001, Moreno (2002) shows a negative relationship between corruption permissiveness and support for democracy. He also illustrates that there are important cross-national differences, suggesting that there is a cultural basis for the justification of corruption. For instance, the extent of corruption permissiveness is still higher in post-communist, Latin American and South Asian countries. He assumes that in young democracies, corruption may be observed as part of inherited practices from old authoritarian governments. Furthermore, he observes an increase in corruption permissiveness in Western societies. However, the influence on levels of corruption is difficult to determine as the measurement of communist legacies is relatively unclear. Based on these findings and reflections, I expect that the extent of corruption in countries will be higher, if the country has a communist past. Further Historical Factors In previous analyses, I included the variable “History of corruption” which plays an important role in explaining current corruption levels and their persistence. The variable is measured by a 4-years lagged dependent varia- 2 Bathtub Model of Corruption 88 ble, the transformed Corruption Perceptions Index. The rationale for including this variable is the assumption that corruption is obviously path dependent. More clearly, it is supposed that the persistence of corruption as the cultural heritage of a society has obviously influence on the extent of corruption of the following years. My analysis confirms this assumption and shows that a country’s “history of corruption” has a strong negative impact on a country’s extent of corruption. However, the coefficient of the extent of corruption and the history of corruption is very high (0.82) and absorbs a lot of explanatory power of an “overall European-specific-model” (Kubbe 2013, p. 156). Moreover, it is highly correlated with the variable “communist past”. For these reasons, the variable is not included in the following analysis. The Micro Level: Socio-demographic Characteristics, Values, Norms and Attitudes Besides a number of country characteristics, personal characteristics of individuals are also expected to impact the extent of corruption through the mechanisms discussed in the bathtub model. Scholars of sociological approaches generally argue that decisions about whether to engage in corrupt transactions are particularly influenced by social norms and cultural values. In this way, corruption is observed as a way of life, as a kind of tradition and as a set of values that is part of a society’s culture. Several researchers indicate high correlations between various societal values and the extent of corruption (Tanzi, 1994; Husted, 1999; Getz and Volkema, 2001; Banuri and Eckel, 2012). Overall, it is striking that only a small body of literature exists that has concentrated on the relationship between corrupt actions and individual characteristics. At the micro level, variables for corruption that are included in the analysis are categorized into socio-demographic factors such as gender, age, employment status, income level, values and norms such as interpersonal trust, and attitudes such as an individual’s satisfaction with financial situation and the justification of bribery. Even though, previous studies have analyzed and discussed some of these variables and their effects on corruption, the current research has not taken all of these variables into account, when analyzing corruption in Europe. These variables particularly include an individual’s level of income, individual’s satisfaction with the financial situation or the justification of bribery. In the following section, I describe and justify, how I derive the 2.4.2 2.4 Variables and Research Hypotheses: Aggregate and Individual Levels Combined 89 variables and hypotheses that I will use and examine in the subsequent analysis. Socio-Demographic Characteristics Gender As already discussed at the macro level (variable “women in parliaments”), the impact of gender on the extent of corruption should not be underestimated. In particular, Swamy et al. (2001) and Dollar et al. (2001) show that women are less involved in corrupt transactions and are less likely to condone bribe-taking than men, implying that women are more trustworthy, public-spirited and have higher norms regarding bribery than males (similar Gottfredson and Hirschi, 1990; Paternoster and Sally, 1996). This is confirmed by a number of articles (Hunt, 2004; Lambsdorff and Fink, 2006; Rivas, 2013; Agerberg, 2014). In addition, Torgler and Valev (2006, p. 17) illustrate that women are significantly less likely to agree that corruption and cheating on taxes can be justified. They claim that “Being a woman rather than a man increases the probability of stating that corruption or tax evasion is never justifiable between 5.8 and 7.1 percentage points." In their study, Torgler and Valev (2010) use the World Values Survey and the European Values Survey data covering eight Western European countries for the period from 1981 to 1999. Their results also remain robust after investigating different time periods and extending their analysis by several additional factors such as education, employment status or income. However, some authors doubt these results (Treisman, 2007). For instance, Sung (2003) or Alatas et al. (2009) illustrate by conducting experiments that the gender differences seems to be more culturespecific than caused by gender differences. Likewise, Alatas et al. (2009, p. 17) assume that “In the context of corruption, one possible explanation for the different gender effects that are observed in our data is the differing social roles of women across cultures. In relatively more patriarchal societies where women do not play as active a role in the public domain, women’s views on social issues may be influenced to a greater extent by men’s views. Hence, in such societies, one would expect to see less of a gender difference in behavior towards corruption in comparison to societies where women feel more comfortable in voicing their own opinions.” 2.4.2.1 2 Bathtub Model of Corruption 90 For the European states, I hypothesize that there is a significant relationship between perceived corruption and an individual’s gender. Age So far, the variable age has been considered only in a few studies examining the extent of corruption. For instance, Torgler and Valev (2006) find strong evidence that there is a significant relationship between corruption, which is measured by the World Values Survey item “justifiability of corruption” (1995-1997), and people’s age. In a panel analysis of 39 countries, they observe that all age groups from 30 to 65+ report a significantly lower justifiability of corruption than the reference group below 30. Referring to previous research such as Hirschi and Gottfredson (2000)45, Torgler and Valev (2006) indicate that an individual’s age is an important indicator of other illegal activities as well. They provide evidence that older people are less likely to view corruption as justifiable and illustrate that the age effect is robust across different social and cultural conditions. This corruption-age nexus is justified by the argument that older people tend to be more tax compliant and less likely to be involved in criminal activities. Hunt (2004) achieved similar results. Using data from 34 countries from the International Crime Victim Surveys, she finds a negative relationship between corruption and age, claiming that older people have had time to develop networks, and such networks, in turn, could lead to honesty. As a result, older people tend to bribe less than younger people. In this context, she suggests that “A higher probability of detection and a greater value of reputation within networks could lead to honesty rather than implicit quid pro quos, although there is no clear dividing line between the two. In the context of the links between crime and trust, trust should lead to honesty, rather than a network for mutually beneficial but possibly illegal exchange.” In a similar vein, Mocan (2008) illustrates that individuals who are 20 to 39 years of age are more likely to be asked for a bribe in comparison to those who are younger than 20. In contrast to that, individuals 45 Hirschi and Gottfredson (2000, p. 138) illustrate that age is negatively correlated with rule breaking. According to this result, they point out that “no fact about crime is more widely accepted by criminologist. Virtually all of them, of whatever theoretical persuasion, appear to operate with a common image of the age distribution. This distribution thus represents one of the brute factors of criminology.”. 2.4 Variables and Research Hypotheses: Aggregate and Individual Levels Combined 91 who are 60 years and older are less likely to get involved in corrupt transactions. Mocan (2008) concludes that older (possibly retired) individuals may have to deal with government rules and regulations less frequently. I assume that there is a significant relationship between perceived corruption and an individual’s age. Employment Status Based on rational-choice approaches, it can be strongly assumed that unemployed people tend to engage in corrupt actions, compared to individuals having a job. Implying that low or no income creates strong economic incentives to take some extra-money in form of bribery. Yet, only a few scholars have examined the relationship between corruption and individuals’ employment status. In particular, Torgler and Valev (2006) illustrate that self-employed and unemployed people have a lower tolerance for corrupt activities compared to other citizens. It lowers the probability for a self-employed person to state that accepting the bribe is never justifiable by 2.9 and for unemployed people by 5.3 percentage points. Torgler and Valev (2006) use a dummy variable for self-employed individuals “as they might be in the best position to invest in bribing and benefit from corruption” (Torgler and Valev, 2006, p. 16). They assume that such a position or a certain status, in turn, may influence the norms regarding bribery and state: “Being away from a job with its regular hours, restrictions, and compensations may increase the incentive to act illegally” (Torgler and Valev, 2006, p. 16). However, in their study from 2010, which concentrates on gender and public attitudes toward corruption and tax evasion, they do not find a statistically significant effect of the employment status on individuals’ justifiability of corruption. Yet, Mocan (2008) using micro level data shows that enhancing the unemployment rate increases the counts of bribery. He demonstrates that an increase of 1 percentage point in the male unemployment rate in the country leads to a rise of bribery by 0.06 percentage points. Macro level studies also show that that increased joblessness is associated with higher levels of corruption (e.g. Goel and Rich, 1989). On the basis of these findings, I posit that an individual’s employment status influences the extent of perceived corruption. 2 Bathtub Model of Corruption 92 Level of Income Similar to the variable “employment status” at the macro level, the level of an individuals’ income is likely to play a role in the extent of corruption at the micro level. However, there are only a few studies concentrating on the relationship between corrupt actions and the individuals’ level of income at the micro level. Similar to the assumption of the variable “employment status” it can be strongly assumed that people with small incomes tend to engage in corrupt actions. With regard to rational-choice approaches, it is assumed that people with low incomes may have greater incentives to engage in corrupt activities because of corruptions’ relatively high benefits. A low income creates challenges for making ends meet and is likely to create incentives for generating supplementary income. In addition, as already indicated at the macro level, lower income countries often have fewer financial resources for creating efficient law enforcement institutions, which make corruption less likely to be detected and punished (O'Connor and Fischer, 2012). Torgler and Valev (2006) confirm this. They indicate that people with a higher income are more likely to be asked for a bribe, as are those with a better education. Contrary to his, individuals with a lower income have lower social “stakes” or restrictions but are “[...] less in a position to take risks because of a high marginal utility loss (wealth reduction) if they are caught and penalized” (Torgler and Valev, 2006, p. 15). This relationship has been examined at the macro level by Haque and Ratna (1996), Montinola and Jackman (2002), Alt and Lassen (2003) or Gerring and Thacker (2005) who have demonstrated that countries with higher incomes tend to exhibit lower levels of corruption. However, the results are mixed. For instance, studies by Husted (1999), and Gurgur and Shah (2005) did not show a statistically significant relationship between civil service income and the extent of corruption. Similar to the assumption of my “employment status” hypothesis, I expect that there is a significant relationship between perceived corruption and an individual’s level of income. 2.4 Variables and Research Hypotheses: Aggregate and Individual Levels Combined 93 Values and Norms Level of Interpersonal Trust Previous research offers different theoretical considerations and contradicting empirical findings on the relationship between trust and corruption. Uslaner (2006) thouroughly investigated both variables trust and corruption and claims that even if they represent opposing moral values, the two are very strongly related. His cross-section results show a reciprocal connection between these two variables and “that the effect of corruption on trust is greater than the opposite causal claim (trust begets an honest political system)” (Uslaner, 2006, p. 3). Moreover, trust as a central component of social capital is a value expressing the belief that others are part of your moral community. Yet, some scholars are more hesitant to inject such a strongly moralistic interpretation into trust. From a rational-choice point of view, trust is simply based on the expectation that others behave predictably (Hardin, 2002). Trust is about certainty of expectations and lays the basis for cooperation with people who are not like yourself (Putnam, 1993; Uslaner, 2006). Among other empirical studies, Paldam and Svendsen (2001), Uslaner (2006) and You (2005) conclude that a strong negative relationship between corruption and interpersonal trust exists, implying that trusting societies have less people acting corruptly. Even though some researchers suggest that societies with high levels of trust also tend to be more tolerant towards corrupt practices. For instance, Moreno (2002) argues that high levels of interpersonal trust support corruption because trust plays an important role in the relationship between corrupt individuals who usually operate with high levels of interpersonal trust necessary to maintain their relationship (Della Porta, 2000). In this context, Rose-Ackerman (2001) illustrates that societies with greater levels of interpersonal trust also exhibit higher levels of corruption and donative transfers.This is attributed to the fact that interpersonal trust decreases the risk of disclosure in corrupt transactions. More precisely, individuals from societies where people commonly rely on informal contracts (which may or may not be corrupt) are common are more likely to enter informal contracts in the future. Similarly, Della Porta, (2000, p. 223) claims that “In all illegal systems of exchange, a high degree of trust and reciprocity is necessary among participants, so the internalization of some rules of the game is therefore necessary. A good reputation for respecting the terms of the illegal exchange, which participants often call ‘honesty’, 2.4.2.2 2 Bathtub Model of Corruption 94 is valued by the actors involved.” However, Rothstein and Stolle (2003, p. 12) opposes this finding: “The high degree of norm conformity that Della Porta depicts among those who are involved in corruption may be plausible, but this is a specific type of trust relations, that cannot be revealed to the outside world”. They claim that people involved in corruption need not really trust one another. It is rather a situation of “mutual deterrence”.46 For the European states, I expect that the level of interpersonal trust influences the extent of perceived corruption. Attitudes Satisfaction with Financial Situation Similar to the assumed relationship between corruption and the level of income at the macro level, it relates to an individual’s satisfaction with the financial situation. However, in contrast to the level of income, this variable relates to the subjective perception of one’s own financial situation. In this context, Torgler and Valev (2006) assume that people who are dissatisfied with their financial situation tend to be more willing to act illegally. Such discontentment, in turn, can create “a sense of distress, especially when there is a discrepancy between the actual and the desired financial situation. Thus, there may be a higher incentive to act illegally to reduce this gap” (Torgler and Valev, 2006, p. 7). Besides the study of Torgler and Valev (2006), there are no other analyses that include this variable. However, following economic approaches, implying that actors follow a rational-choice logic and are often motivated by material interests and commit or refrain from corrupt acts for tangible goods, this variable should be included in the following analysis. It can be assumed that people who are unsatisfied with their own financial situation strive for higher income and are also prepared to accept illegal payments. For Europe, I assume that 2.4.2.3 46 In this analysis institutional trust is not included as independent variable because it rather appears as consequence as a cause of corruption. In this context, Kubbe (2014) presents a causal model of corruption and trust, including corruption as a mediator of interpersonal and institutional trust. For the Western and Central and Eastern countries the model illustrates that increasing interpersonal trust enhances institutional trust. This is especially the case, when the degree of corruption is minimized. 2.4 Variables and Research Hypotheses: Aggregate and Individual Levels Combined 95 there is a significant relationship between perceived corruption and an individual’s satisfaction with the financial situation. Justification of Bribery In the following analysis, it is also assumed that people’s attitude towards illegal behaviour has an influence on the extent of corruption. More precisely, it is expected that people who are more tolerant towards corruption are more likely to act corruptly as well. Research examining this relationship is, however, very scarce. For instance, Moreno (2002, p. 2) suggests that “Corruption has a cultural side, and most societies have a certain degree of corruption permissiveness, with some of them being, on average, more likely to justify corrupt practices than others.” In his article he analyses data from the World Values Survey including 64 societies in four rounds of surveys conducted between 1981 and 2001. Out of this data, he constructed an index of corruption permissiveness measuring the extent to which people tend to justify certain practices that can be considered corrupt. He demonstrates that there are significant cross-national and crossregional variations in the permissiveness of corruption, suggesting that some societies justify corrupt acts based on cultural values. For instance, the level of corruption permissiveness was very high in post-Communist countries, followed by Latin American countries, and South Asian societies. Additionally, for the time period of 1995-2000 he finds an observable increase in corruption permissiveness in Western democracies as well, the most significant being in the United States (Moreno, 2002). He suggests that these attitudes towards corruption are strongly negatively associated with interpersonal trust and democratic attitudes such as the support for democracy that are important components of a democratic political culture. In his article “Everyone’s is doing it”, Green (1991) also identifies and analyses certain conditions47 that provide people a moral justification for 47 Green (1991) recommends the following “Conditions Permitting One to Engage in Harmful but Prevalent Behavior: 1. Refraining from this behavior will unavoidably cause you (or those you care for or for whom you are responsible) serious harm or loss. 2. Your engaging in this behavior will not also cause significantly more harm or loss to others. 2 Bathtub Model of Corruption 96 engaging in corrupt actions. He points out that the excuse “Everyone’s is doing it” is frequently used as a morally valid reason to explain why people behave corruptly. He applies these conditions to representative cases in business ethics and provides evidence that this is particularly true in business sectors, where competitive pressures are often very high.48 Green’s (1991) findings also suggest, that corrupt actions are often closely related to cultural norms and traditions given in certain societies. Similarly, this is substantiated by the study of Beck and Lee (2002) which analyses the attitudes of Russian police officers as perceived by the public and the media, to be open to using their public positions to obtain some extra-money, goods and services. Conducting surveys about beliefs and values referring to corruption among students and serving officers attending a police institute, they demonstrate that young police recruits in particular believe that corruption is often justifiable and morally acceptable under particular circumstances and rules. That, in turn, might have an influence on a country’s degree of democracy because the police as state representatives are perceived to play a key role in the development of democratic states (Beck and Lee, 2002). Furthermore, on the basis of interviews conducted in Kenya and Uganda, Persson et al. (2012) argue that the reason why people act corruptly, although condemning corruption, seem to be that they understand the situation as a collective action problem where 3. Your engaging in this behavior will not lead others to engage in it in ways that are equally or more harmful, and this would be true if your engaging in this behavior were to become public knowledge. 4. Your refraining from this behavior will not lead others to refrain from it, and this would be true if your refraining from this behavior were to become public knowledge. 5. You refraining from this behavior will not unavoidably lead others to engage in it in ways that are substantially more harmful than would have been the case had you chosen to engage in it yourself and this would be true if your refraining from this behavior were to become | public knowledge.” (Green, 1991, p. 77). 48 Green (1991, p. 90) concludes that “On the one hand, these conditions must be shown to reflect and adequately express the relevant considerations governing all moral choice, what I earlier called the basic ‘logic’ of the moral reasoning process. On the other hand, a list of justifying conditions must also adequately guide judgment through both familiar and novel cases for decision and it must do so without violating some of our firmest and most settled judgments about these cases. When these two sides of the task are adequately accomplished, we can say that an exercise of this sort is successful and that we are in a position of “reflective equilibrium” before the issues at hand.”. 2.4 Variables and Research Hypotheses: Aggregate and Individual Levels Combined 97 it makes little sense to be “the only one” that refrains from using or accepting bribes and other kick-backs. In other words, “it appears to be a coordination problem, where the equilibrium that emerges depends on shared expectations about others’ behavior” (Persson et al., 2012, p. 15). For the European states, I posit that the level of the justification of bribery influences the extent of perceived corruption. Further Socio-demographic Characteristics, Values, Norms and Attitudes In previous analyses, I also included the variable “societal values”, measured by data from the World Values Survey. They consist of a combination of (1) a liberating orientation, an emphasis on freedom of choice, (2) an egalitarian qualification of this liberating orientations as equal freedom of choice, or equality of opportunities (see also emancipative values in Welzel, 2013).49 Yet, my study reveals that societal values do not have an impact on a country’s perceived level of corruption (Kubbe, 2013). 49 Welzel (2013) has identified twelve items that are grouped into four domains of emancipatory orientations, covering an emphasis on autonomy, choice, equality and voice. While the emphases on autonomy and choice address more directly the liberating aspect of emancipation, the emphases on equality and voice address more directly the egalitarian aspect. 2 Bathtub Model of Corruption 98

Chapter Preview

References

Abstract

International studies often point to Europe for low levels of corruption. However, recent scandals in nearly all European states illustrate that corruption continues to be on the rise. The author investigates the causes of corruption in Europe. The analysis indicates that a country’s contextual conditions such as the economic development, the degree and duration of democracy or historical factors like the post-communist past strongly influence Europe’s level of corruption. Furthermore, corruption is likely experienced differently depending on interpersonal trust and the justification of bribery. The findings reveal that a bundle of factors adding up to a specific “democratic culture” hinders the growth of corruption by generating strong democratic institutions and fostering citizen norms and values aimed at monitoring and sanctioning corrupt actors. As a result, democracy promotion is the best remedy against corruption spread in Europe.

Zusammenfassung

Auch wenn europäische Staaten vergleichsweise geringe Korruptionswerte aufzeigen, verdeutlichen Skandale immer wieder, dass Korruption ein großes Problem darstellt, mit dem auch Europa stark zu kämpfen hat. Die Autorin untersucht daher die Ursachen von Korruption auf dem europäischen Kontinent. Verschiedene Analysen zeigen, dass Kontextfaktoren eines Landes wie dessen ökonomischer Entwicklungsstand, der Demokratisierungsgrad und die jeweilige Dauer oder historische Faktoren wie die kommunistische Vergangenheit das Auftreten von Korruption stark beeinflussen.

Darüber hinaus spielen interpersonales Vertrauen und die Rechtfertigung von Bestechungszahlungen eine erhebliche Rolle in der Wahrnehmung von Korruption. Insgesamt zeigen die Befunde, dass letztendlich eine „demokratische Kultur“ der Schlüssel im Kampf gegen Korruption in Europa ist. Diese fördert demokratische Institutionen sowie Normen und Werte, die darauf abzielen, korrupte Akteure zu kontrollieren und sanktionieren.