Journal of Civil Society ISSN: 1744-8689 (Print) 1744-8697 (Online) Journal homepage: http://www.tandfonline.com/loi/rcis20 Localization, Formalism, and Passivity: Symbols Shaping Bribe Offers in Azerbaijan Turkhan Sadigov To cite this article: Turkhan Sadigov (2017): Localization, Formalism, and Passivity: Symbols Shaping Bribe Offers in Azerbaijan, Journal of Civil Society, DOI: 10.1080/17448689.2017.1387716 To link to this article: http://dx.doi.org/10.1080/17448689.2017.1387716 Published online: 19 Oct 2017. Submit your article to this journal Article views: 73 View related articles View Crossmark data Full Terms & Conditions of access and use can be found at http://www.tandfonline.com/action/journalInformation?journalCode=rcis20 Download by: [UAE University] Date: 25 October 2017, At: 12:55 JOURNAL OF CIVIL SOCIETY, 2017 https://doi.org/10.1080/17448689.2017.1387716 Localization, Formalism, and Passivity: Symbols Shaping Bribe Offers in Azerbaijan Turkhan Sadigov Department of International Relations, Azerbaijan Tourism and Management University, Baku, Azerbaijan Downloaded by [UAE University] at 12:55 25 October 2017 ABSTRACT The article analyses a neglected side of corruption – citizen-initiated bribe offers in Azerbaijan. The main rationale for taking a citizen focus on corruption is to identify whether citizen choices support state-led anti-corruption initiatives. Drawing on a survey of 1002 respondents, the research shows that ‘tolerance of corruption’ is high among the population in Azerbaijan. Logistic regression results show that bribe offers among citizens are associated with localization and formal value orientations. This means that the respondents pro-actively use bribes and readily bend laws to their own local interests. Thus, bribe offers constitute around half of the entire variation in corruption in Azerbaijan. Therefore, a successful fight against corruption cannot be confined to the eradication of only bureaucratic bribe demand, but needs to take into account corruption choices of the population too. KEYWORDS Corruption; bribe offers; postSoviet politics; social capital; Azerbaijan Introduction Corruption is generally viewed as negatively affecting both political development (i.e., political participation, voting behaviour, and women’s representation) (Birch, 2009; Kostadinova, 2009) and the socio-economic situation of the country (lower growth) (see Stockemer, 2011; Stockemer, LaMontagne, & Scruggs, 2013). Pervasive corruption is an important feature of contemporary post-Soviet societies. Non-Baltic states of the region consistently rank close to the bottom of major annual corruption indices and reports, like those of Transparency International (TI), International Alert, World Bank, and others. Many analysts view the region among the most corrupt in the world (Holmes, 2012; Stiglitz, 2002, p. 148). A corruption study by the International Monetary Fund (IMF) defines the post-Soviet region as ‘markedly more corrupt than any other region of the world’ (Roaf, 2000, p. 1). Therefore, corruption is an important socio-economic problem of the region (World Bank, 2013). Among the most corrupt countries, having both of the mentioned features, Azerbaijan has recently made consistent progress in terms of its TI Corruption Perceptions Index (CPI) scores and ranking for the past eight years (see Table 1).1 According to TI’s The state of corruption report (McDevitt, 2015, p. 15), one of the causes of the recent improvement of the Azerbaijani CPI score is transferring of the provision of licencing and a range of other public services from public agencies to ASAN − a private CONTACT Turkhan Sadigov Azerbaijan email@example.com © 2017 Informa UK Limited, trading as Taylor & Francis Group Turkhan Sadigov, 822/23 Koroglu Karimov, Baku, AZ1172, 2 T. SADIGOV Table 1. TI CPI score dynamics of the lowest ranking resource-rich post-Soviet states (2004–2015). Years Azerbaijan Downloaded by [UAE University] at 12:55 25 October 2017 2004 1.9 2005 2.2 2006 2.4 2007 2.1 2008 1.9 2009 2.3 2010 2.4 2011 2.4 2012 27 2013 28 2014 29 2015 29 Source: McDevitt (2015). Russia Kazakhstan Uzbekistan Turkmenistan 2.8 2.4 2.5 2.3 2.1 2.2 2.1 2.4 28 28 27 29 2.2 2.6 2.6 2.1 2.2 2.7 2.9 2.7 28 26 29 28 2.3 2.2 2.1 1.7 1.8 1.7 1.6 1.6 17 17 18 19 2.0 1.8 2.2 2.0 1.8 1.8 1.6 1.6 17 17 17 18 company of a one-stop-shop model. While this progress is made through government initiative, how sustainable are these top-down changes? To what extent are these changes also echoed by the dynamics of the country’s civil society? Do the grassroots’ attitudes of the society support these state-led anti-corruption initiatives? A recent TI report points to ‘tolerance of corruption’ among the broader Azerbaijani population as one of the major problems in Azerbaijan (TI, 2014, p. 9). What is the degree of this corruption tolerance among the population? The research and practice of the anti-corruption fight in the region are dominated by a bureaucracy- rather than an ordinary citizen-centred approach. Many international organizations, policy-makers, and other practitioners of anti-corruption measures argue that the success in the fight against corruption in the post-Soviet space depends on the degree to which reforms and laws restrict misbehaviour of public officials. As Joel Moses notes Sachs and others, like Joseph Stiglitz from the World Bank, now bemoan their own failure to appreciate the depth of official corruption in many post-Soviet countries and the lack of any sense of the rule of law among the post-Soviet elites, a sense necessary to bring about … capitalism’ (Emphasis added; Moses, 2003, p. 13). Missing from this picture is the possibility that the initiation of corrupt behaviour emanates from ordinary citizens. Indeed, even ideal laws and governmental policies do not guarantee order, because order ultimately depends on the willingness of the population mass ‒ ordinary citizens ‒ to embrace good practices and integrate them into their daily behaviour. Much of the literature on corruption does not consider the impact of citizen responsibility for not offering a bribe on the incidence of corruption. From three major approaches to corruption, two – the institutionalist (Bardhan, 1997; Obydenkova & Libman, 2015) and economic approaches – only marginally focus on the role of bribe offers. Institutionalist approaches instead concentrate on the institutional opportunity structure (Rose-Ackermann, 1978) that gives bribe-demanding bureaucrats freedom of hand (RoseAckermann, 1997, 1999), while economic approaches view corruption as a result of personal cost–benefit calculations. Another strand − sociocultural theories (Mocan, 2004) − researches citizens’ initiation of bribes. However, there is a dearth of empirical research of bribe offers in particular. Therefore, my research focuses on bribe offers by the general population to fill in the existing gap in the corruption research. Downloaded by [UAE University] at 12:55 25 October 2017 JOURNAL OF CIVIL SOCIETY 3 This article takes a citizen-centred approach to corruption in the post-Soviet region. Corruption is defined in this study as the informalization of standardized laws, rules, and regulations by particularism (Ledeneva, 2006, 2013b; Mungiu-Pippidi, 2006) of local interests to provide and receive preferential treatment. A theoretical contribution of this work is that it fills a research gap in the literature which overwhelmingly focuses on bureaucrats’ misbehaviour in fighting corruption (e.g., arrests of officials), but not on how citizens offer bribes. The main subject of the work is the degree to which ordinary citizens (not occupying any public posts) are willing to support legality and socio-economic reforms in the country. To assess their readiness to support legality and reforms by ordinary citizens, my research aims to identify the causes of citizen-initiated bribe offers. Out of the many forms of corruption, the research focuses on bribery as ‘the most widespread form of corruption’ in the post-Soviet region (Temple & Petrov, 2004, p. 89). Bribery is also the best form of corruption through which to capture the attitudes and practices of ordinary citizens. While the initiation of other forms of corruption, like embezzlement, nepotism/clientelism or extortion, depends heavily on public officials, bribery can be initiated in equal measure both by officials (demands of bribes) and by ordinary citizens (offer of bribes). We analyse bribe offering among the population of Azerbaijan through the findings of the field survey conducted by the author among 1002 people in Azerbaijan from February to August 2015. We draw on relevant theories explaining bribe offers by citizens to identify 14 hypotheses about citizen choice of bribe offers. Overall, the regression also shows that citizen-initiated bribing accounts for approximately half of the variation in the incidence of population-initiated corruption. Therefore, in order to fight corruption effectively, policy-makers should not focus exclusively on ‘bribe demands’, but pay attention to ‘bribe offers’ from the population too. Alternative theoretical approaches Localization of laws/particularism One of the major findings of the corruption research is that individualization of purportedly universal and standardized rules, laws, and regulations by citizens is a reliable factor predicting those citizens’ attitudes and practices of corruption. Ledeneva (1998, 2003, 2011, 2013a) discusses selective and individualized ad hoc application of laws, predominance of informal rather than formal transactions − all this contributing to endemic corruption. Some authors call this drive to localization of laws ‘particularism’, which ‘runs directly counter to universalism, the norm and practice of individualistic societies, where equal treatment applies to everyone regardless of the group to which one belongs’ (Original emphasis; Mungiu-Pippidi, 2006, p. 88). A number of authors (Banfield, 1958; Harrison, 1997) point out that corruption is related to localization through widespread perception of formal laws as relativistic: For individuals, the occurrence of corruption or the lack of it depends very much on minute situational and local criteria − specific circumstances and social context in which a transaction happens. A clear line between legality and illegal acts becomes blurred. This uncertainty creates propitious grounds for selective application of laws, and active abuse of regulations, including bribing and nepotism. 4 T. SADIGOV Localization/particularization of laws is also visible in the discussion of the prevalence of personal traits of a public officer over the rules regulating his/her duties: Under the contradictory constraints imposed by formal rules and informal norms, relationships matter more than rules for individual survival … Personal loyalty … implies ‘personalized’ service – inclusive of display of emotional attachment, personal attention and friendly favors – and compliance that prioritizes informal commands over formal instructions. (Ledeneva, 2013a, p. 39) Downloaded by [UAE University] at 12:55 25 October 2017 A public officer in such case becomes a patron able to solve the problems of his/her clients more effectively than formal laws or the entire system of public administration. Social relations hinge on personal loyalty rather than loyalty to vaguely understood public good. We derive the following hypotheses from this discussion: Hypothesis 1: The greater a respondent’s perception that corruption depends on specific situational context and circumstances, the more s/he is likely to offer a bribe. Hypothesis 2: If a respondent views personality of a bureaucrat rather than formal laws as being more important in defining functions, duties, and responsibilities of a public office in Azerbaijan, then the respondent is more likely to offer a bribe. Citizen activism and passivity Citizen bribe offers might be affected by their socio-political activism. Social passivity and fatalism lead to low interest in social progress and the lack of a sense of personal responsibility in standing up to social ills like corruption (Banfield, 1958, pp. 109, 112). If a person thinks that the success of the fight against corruption depends on government rather than on ordinary people’s readiness to stand up to bribes, or, at least, refrain from bribing, then this respondent is more likely to offer a bribe, because of the lack of a sense of personal responsibility for causing corruption (Harrison, 1997; Putnam, 1993). This respondent is more likely to take a passive role of an observer: S/he might blame government for corruption, and argue that the ‘system made me do it’ (Karklins, 2005). The appeal to symbolic transcendental forces is closely tied to social passivity of a person. If success in a society like that of Montegrano depends on luck or divine intervention (symbolic realm) then, of course, people perceive appeals to saints and other manipulation of symbolic realm as the main avenue for achieving their material ends (Banfield, 1958, p. 109). Fatalism, with its social passivity and the appeal to symbolic transcendental forces, leads to low interest in social progress, and the lack of a sense of personal responsibility in standing up ‘here and now’ to social ills, like corruption. Therefore, if a respondent thinks that the cause of social problems is moral and ideological weakness of a society, s/he is more likely to have a low interest in investing efforts and personal application to oppose corruption. We derive the following hypotheses from this discussion: Hypothesis 3: The more a respondent supports the idea that the eradication of corruption depends on the government rather than citizens’ refusal to offer bribes, the more a respondent is likely to offer a bribe. JOURNAL OF CIVIL SOCIETY 5 Hypothesis 4: The greater a respondent’s perception that the government should be like parent (patrimonial view) rather than a public servant (contractarian view), the more a respondent is likely to offer a bribe. Hypothesis 5: The higher a respondent’s support for the idea that the root-cause of all socioeconomic problems is moral and ideological weakness of a society, the more likely s/he is to offer a bribe. Downloaded by [UAE University] at 12:55 25 October 2017 Decoupling of symbol and its content Socio-economic research has long established that closed systems, formed around ascriptive and symbolic factors, make corruption more probable (Banfield, 1958; Putnam, 1993). According to several theories of corruption, symbolic capital (excessive focus on status (Maine, 1861), exclusivity, and privilege) is associated with bribe offers. Predominance of symbolic (status, prestige) rather content (investment in science, education, and innovation) values leads to an exclusivist social framework, susceptible more to corruption rather than merit-based interaction. Engvall argues that ‘when individuals invest in public offices for clear profit-making reasons, it also comes at the expense of investments in productive commercial activities; finances that could flow into private sector businesses are instead circulating unofficially among state officials’ (Engvall, 2015, p. 15). Social and economic relations, thus, are increasingly revolving around the distribution and redistribution of rents on the basis of exclusivist status markers (Harrison, 2000, p. 299) rather than the production of wealth in a merit-based environment, regulated by inclusive standardized norms. As a result, rigid limits placed on economic activity and productive investment by status and prestige translate into higher probability of corruption (Banfield, 1958). Markers of status and prestige, thus, contribute to the spread of corruption by creating segregated exclusivist nepotistic circles of in-group belonging, new aristocracy of glamour and status, traditionalist habits of contradistinction between favoured ones and the outsiders. According to Huntington, ‘corruption … results when the absence of mobility opportunities outside politics, combined with weak and inflexible political institutions, channels energies into politically deviant behavior’ (Huntington, 1968, p. 66). This exclusivism runs counter to modern society’s definition of legality as based on egalitarian, standardized, uniform, formal, and de-individualized ‘contractual’ relations. This exclusivism is especially corruption-prone in the post-Soviet region, where the legal system is weak to enforce laws impartially. External focus − the belief that the key to the solution of the most important problems facing a society lies outside of this society rather than in hard work within domestic realm – is another important variable that is closely tied to the probability of corruption. Corruption-prone communities are more inclined to focus on the manipulation of external realm (lying beyond their societies), rather than engagement in improvements in their domestic sphere (Banfield, 1958; Gellner, 1983; Schein, 2010). In this worldview, external reality, while much more overwhelming and powerful, is more important for the fate of an agent than the internal/domestic sphere. As a result, most of the attention of both elites and the population is attached not to the domestic realm but to external reality, full of threats and promises. This, in turn, leads to the neglect of diligence, meritocracy, technical and instrumental savvy supplanted by the public relations manipulation of external 6 T. SADIGOV audiences and governments, and the shamanic application of other symbolic measures (lavish spending on international contests, publicity stunts, and glamour) to affect developments within the domestic realm. As a result, state statistics are manipulated to project a better ‘image’, budget funds spent on ‘white elephant’ projects are often misappropriated, and the contents of the laws are manipulated to achieve desirable symbolic effects, thus paving the way for corruption. We derive the following hypotheses from this discussion: Hypothesis 6: The higher the share of investment that a respondent is willing to spend on the markers of status and prestige (expensive car plates, phone numbers, and jewellery), the higher is the probability that s/he offers a bribe. Downloaded by [UAE University] at 12:55 25 October 2017 Hypothesis 7: The higher a respondent’s perception that external reality is more important than consistent work in the domestic realm, the higher is the probability that s/he offers a bribe. Personal factors of social and material capital There is an influential corpus of literature that finds the causes of corruption in varying levels of social capital, like associational membership, trust in others, and peculiarities of a belief system. A number of corruption researchers (Banfield, 1958; Harrison, 1997, 2000; Putnam, 1993) argue that a web of civic gatherings and organizations increases interpersonal trust in a society. The more a person prefers bridging social organizations (belonging to which is based on a commonality of interests) over bonding organizations (membership in which is defined by the commonality of ancestry, blood, origin, and other ascriptive qualities), the less s/he is likely to prefer nepotism and corruption as methods to solve social problems. On the other hand, weak associational membership leads to a gradual shrinkage of the social circle to the immediate kin, thus boosting the role of nepotism and corruption in a society. Economic theories agree with Putnam’s and Banfield’s emphasis on the inter-relation between trust and corruption. They point out that disparity in wealth distribution is said to undermine social cohesion of societies and ultimately lead to corruption (Rothstein, 1998; Rothstein & Uslaner, 2005). These authors take interpersonal trust as a proxy to measure social cohesion and its impact on corruption. Some theorists argue that following current events through news and other media outlets is an important proxy of citizen activism, which is negatively associated with corruption (Gudkov & Dubinin, 2001; Putnam, 1993). Putnam argues that following news reflects involvement of ordinary citizens into community life and their symbolic investment in communal good – aspects that make a citizen less likely to breach communal laws through resorting to corruption. There is, however, a difference among different types of media in affecting citizen activism. While the internet gives a person a greater chance to pro-actively choose, compare, and analyse news content from different sources, discuss and reflect on it in blogs and social media, news on TV is more likely to instil reactive and unresponsive consumption of information (Gudkov & Dubinin, 2001). Therefore, we include in our analysis a dummy variable, denoting whether a person follows any news or not. Another important personal factor that shapes citizen susceptibility to corruption is income level. Various theories point to poverty as a primary cause of bribe offers by Downloaded by [UAE University] at 12:55 25 October 2017 JOURNAL OF CIVIL SOCIETY 7 ordinary citizens (Kostadinova, 2012; Sindzingre, 2011). Using relatively small bribe offers, the poor population gains in exchange comparatively greater social benefits and welfare funds, and thus are able to live better. Thus, the population uses bribe offers to survive. Another example of citizen-initiated corruption is vote selling (Domínguez, Greene, Lawson, & Moreno, 2014; You, 2015). Poor strata of population exchange votes for immediate payments rather than thinking about longer term policy implications of their ballot selling. Green argues that ‘poverty fuels corruption, as starving people find it difficult not to sell their votes for a bag of flour’ (Green, 2012, pp. 85–86). As a result, using relatively small bribe offers to public servants, the poor population gains in exchange comparatively greater social benefits (pension payments that citizens may not be entitled to, welfare funds, etc.) and thus are able to survive. Various ideological belief systems may impose moral, spiritual, and other symbolist restraints on individuals’ behaviour and choices regarding various social phenomena, including corruption. Since ideological systems aim at boosting social stability and progress through discipline and integrity, they invariably should treat corruption as a threat, because corruption puts individual interest above a collective one. For instance, such diverse and at times mutually exclusive ideologies as Communism, Liberalism, and Islam forbid corruption. Therefore, our expectation is that an abidance by a systematized belief system (be it religious or secularist) that prohibits corruption will negatively affect the actual incidence of bribing among those respondents who follow these ideological frameworks. We derive the following hypotheses from this discussion: Hypothesis 8: Higher frequency of associational membership leads to lower probability of a bribe offer by a respondent. Hypothesis 9: The higher a respondent’s trust in others, the lower is the probability that s/he will offer a bribe. Hypothesis 10: If a respondent follows news on any media, s/he has lower probability of offering a bribe. Hypothesis 11: The higher a respondent’s monthly income, the less s/he is likely to offer a bribe. Hypothesis 12: Abidance by a systematized principle system (be it religious or secular ideology) should make a respondent less prone to offer a bribe. Institutional factors A range of theories find causes of corruption in gaps in the institutional environment (Shleifer & Vishny, 1993). To this group also belong researchers who attribute the pervasiveness of corruption to a society’s level of institutionalization (Huntington, 1968, p. 66). These theories pay particular attention to the choices and decisions that institutions create for actors involved in corruption. Robert Klitgaard argues that corruption is almost always defined by a formula: Corruption = Monopoly + Discretion − Accountability (C = M + D − A) (Klitgaard, 1998, p. 4). If an authority has a monopoly over some service or product and an unlimited discretion to use its monopoly, complemented by a low level of accountability to any other authority, then the probability is very high that Downloaded by [UAE University] at 12:55 25 October 2017 8 T. SADIGOV corruption will flourish in such system. We can use Klitgaard’s formula to measure the impact of citizen discretion (what is the real margin of choice that a citizen has to avoid paying bribes, if s/he does not want to pay?) and accountability (how likely is it that a citizen will get punished for offering a bribe?) on the citizens’ bribe offer probability. Institutions can also affect corruption through socialization mechanisms. According to Huntington, ‘corruption … results when the absence of mobility opportunities outside politics, combined with weak and inflexible political institutions, channels energies into politically deviant behavior’ (Huntington, 1968, p. 66). Since post-Soviet institutions have long been exposing citizens to corruption (Ledeneva, 2013a), institutional theories can argue that the population’s bribe offers are explained by the level of their institutional socialization of corruption. Since socialization is based on subjective perception, asking respondents about specifics of their socialization might yield distorted data. Therefore, we focus on the result of the exposure to corrupt institutions – the inclination of people to manipulate and bend laws. Institutional theories may argue that if institutions are dominated by ‘legal nihilism,’ this in turn makes people more likely to abuse laws and offer bribes. Hypothesis 13: If a respondent has a choice to avoid bribes, then s/he is less likely to offer a bribe. Hypothesis 14: The more a respondent uses informal methods instead of the appeal to formal laws to solve daily problems, the more s/he is likely to offer a bribe. Data and methods To study the incidence and characteristics of bribe offers among the population in Azerbaijan, we draw on an original, nationally representative, face-to-face survey of 1002 respondents. The survey was conducted by the author in February–August 2015. The margin of error for the survey is 5%. Using mixed stratified snowball sampling, the survey sampling covered all of the economic regions of Azerbaijan. The division of the respondents by region of residence, age, and gender reflects the relative shares of Azerbaijani population, according to the figures of the State Statistical Committee of Azerbaijan.2 The primary focus of the research is bribing as ‘the most widespread form of corruption’ in Azerbaijan (Temple & Petrov, 2004, p. 89). The fieldwork was shaped to identify the impact of relevant independent variables, as well as specific measures of both perception and practice of bribe offers among the respondents. Another method is in-depth interviewing of Azerbaijani citizens. We draw on 10 in-depth interviews with people from different social backgrounds (identified based on gender, age, education, place of residence, and income). The interviews were held in Azerbaijan in May−September 2015 by the author of the article. These interviews consisted of 15 semi-structured and open-ended questions for discussion of various aspects of corruption and bribe offers. Variables This research seeks to identify factors explaining bribe offers of the population. Therefore, as our dependent variable, we used an annual incidence of bribe offers among the respondents − a categorical dependent variable. We did not consider the frequency of bribe Downloaded by [UAE University] at 12:55 25 October 2017 JOURNAL OF CIVIL SOCIETY 9 offers, because it depends on a range of highly individual circumstances. For instance, if a person falls ill and turns to hospital, s/he may end up paying varying number of informal payments depending on which hospital s/he chooses. Or the type of business that an entrepreneur is engaged in puts him in a situation, when he pays more frequently, but less per bribe, in comparison to an entrepreneur in quite another sphere of business. So, the frequency of bribe payments is a very unreliable measurement of the propensity of the respondents to engage in bribe offers. Therefore, we chose instead to focus on the incidence of bribe offers (whether a respondent offered a bribe at all, or did not, for the past 12 months). The incidence of bribe offers measures all of the bribe offers initiated by a respondent, without difference as to whether these offers were received or not received by public officers, or whether these bribes satisfied these respondents’ needs, or failed to satisfy them after they offered a bribe. Given the dichotomous categorical nature of the dependent variable, we use logistic regression to analyse our data. We process our quantitative survey results to generate logistic probabilities that allow us to see how our independent variables affect the dependent variable. We start by recoding our dichotomous independent variables into dummy parameters. We have four independent variables that require recoding. These variables are: (1) the respondents’ attitudes towards whether formal role or informal personality is important in public administration (as a categorical variable, it has two categories: (a) Informal personality is more important; or (b) Regulations drawn from formal law are more important); (2) whether a respondent refers to some systematized set of principles (be it religious or secular) when making everyday choices, or her choices are not guided by any framework; (3) whether a respondent thinks that s/he has a choice to avoid bribes, or has no choice at all; (4) whether a respondent follows news through some media (be it TV, internet, radio or any other source) or does not follow news at all. All of these variables are bivariate ones, and therefore are transformed to a 0–1 scale. The remaining 10 factors are ratio variables not requiring transformation. We therefore create four dummy variables, recoding them in the following way: Personality of a bureaucrat (Relative importance of the personality of bureaucrats vis-àvis their formal responsibilities, outlined by law): . Personality is more important: 1 if personality is important, 0 ‒ if not important. Belief system (whether a person abides by a systematized set of principles while making daily choices): . Not guided by any framework: 1 if Yes, 0 ‒ otherwise. Choice to avoid bribes (whether a person thinks s/he has a choice to avoid bribes): . Has at least some choice to avoid: 1 if Yes, 0 ‒ otherwise. News following (whether a person follows news through some media, or does not follow news at all): 10 . T. SADIGOV Does not follow news at all: 1 if Yes, 0 ‒ if follows news through some media. The regression equation I use to test the model is the following: Yi = bo + li + ui + ai + ýi + b1 CIRCUMi + b2 LAWSi + b3 PATERNALi + b4 RESPONSIBILITYi + b5 MORALi + b6 EXTERNALi + b7 PRESTIGEi Downloaded by [UAE University] at 12:55 25 October 2017 + b8 MEMBERi + b9 INCOMEi + b10 TRUSTi , where Yi is the probability of a bribe offer by a respondent i, βo is the predicted constant of bribing for a respondent who adheres by at least one systematized set of principles while making decisions; has at least some choice to avoid bribes; follows news through at least one media; thinks that only formal laws and regulations guide bureaucrats’ behaviour; while having all of the rest of the independent variables equal 0. λi = 1 if respondent i thinks that a personality is more important than laws. λi = 0 otherwise (respondent i thinks that laws are more important than a personality of a public officer). θi = 1 if respondent i does not uphold any belief system when making decisions and choices. θi = 0 otherwise (respondent i abides by at least one belief system when making decisions and choices). αi = 1 if respondent i thinks that s/he has a choice to avoid bribes. αi = 0 otherwise (respondent i thinks that s/he has zero chance of avoiding bribe payment). ύi = 1 if respondent i does not follow news on any media. ύi = 0 otherwise (respondent i follows news through at least one media). β1CIRCUMi is the number (from 0 to 10) that reflects the respondent i’s position towards whether there is no clear-cut line between corruption and the lack of it. Corruption depends on a situation and is context-based (=0) or there is an absolute line between corruption and the lack of it in every situation (=10), or a number from 1 to 9 that correctly reflects the respondent i’s position. β2LAWSi is the number (from 0 to 10) that reflects the respondent i’s usage of laws to solve his/her problems (complete lack of appeal to law = 0; constant usage of laws = 10; mixed position regarding the usage of laws = a number from 1 to 9). β3PATERNALi is the number (from 0 to 10) that reflects the respondent i’s complete paternalistic (=0), fully contractarian (=10) or mixed (any number from 1 to 9) position regarding a government’s standing towards the population. β4RESPONSIBILITYi is the number (from 0 to 10) that reflects the respondent i’s position towards the idea that the eradication of corruption depends either solely on government (=0) or completely on the population (=10) or mixed position (a number from 1 to 9). β5MORALi is the number (from 0 to 10) that reflects the respondent i’s position on the importance of the idea that the root-cause of all socio-economic problems is moral and ideological weakness of a society (0 = very important, 10 = not important at all, and from 1 to 9 = mixed importance). β6EXTERNALi is the number (from 0 to 10) that reflects the respondent i’s position towards the importance of external focus − the belief that the key to the solution of the most important domestic social problems lies outside of the Azerbaijani society (very important = 0, not important at all = 10, and from 1 to 9 = mixed importance). β7PRESTIGEi is the number in thousand of manats (from 0 to 100 thousand), denoting the respondent i’s willingness to spend on the markers of status and prestige. β8MEMBERi is the number that reflects the respondent i’s memberships in voluntary associations and organizations. β9INCOMEi is a respondent i’s monthly income from all sources. β10TRUSTi is a respondent i’s level of trust to strangers (from complete distrust (0) to complete trust (10), or a mixed result (between 1 and 9)). JOURNAL OF CIVIL SOCIETY 11 Since the least score in independent variables is 0, and this score is associated with a greater probability of bribe offers in our model, we predict a negative association of these interval independent variables with our dependent variable. Downloaded by [UAE University] at 12:55 25 October 2017 Descriptive highlights Naturally, in the context of this study, the first issue to be considered is whether respondents engage in bribing, and, if yes, how prevalent it is. The absolute majority (77.8%) of the respondents indicated that they offered a bribe at least once per past 12 calendar months. The majority of those that offered a bribe, offered it once (28.14%) for the past 12 months. However, 56.3% of survey participants indicated that they have a valid choice to avoid a bribe when a public servant presses for it. Those who offered two (20.0% of the respondents) and 3 bribes (9.8%) closely follow. This means that the majority of the respondents are not intensively engaged in bribing. The survey revealed that an average respondent views corruption as the most important problem that business faces in Azerbaijan. Specifically, one-third of the respondents (33.6%) pointed to corruption as the most important domestic problem that negatively affects the socio-economic development of Azerbaijan (see Table 2). The respondents view corruption as the single most important problem that needs to be tackled not only with respect to personal problems, but on a countrywide scale. Regression results Table 3 reflects logistic multiple regression results. I include pseudo R 2 and log-likelihoods into my results. Model 1 and Model 2 differ in only one variable − the self-reported choice of a respondent to avoid bribe payments. By withholding this variable, I measured the relative impact of a choice to avoid bribes on the overall probability of bribe payments by the respondents. As Models 1 and 2 show, all of our independent variables, except for five − the respondents’ abidance by a belief system, the level of support for paternalistic state, number of memberships in voluntary associations, news following, trust in others, and choice to Table 2. Most important domestic development problem faced by Azerbaijan according to respondents. Azerbaijan’s most important problem Corruption/red tape Moral decadence in society Low education level/technical knowledge Low salaries/low pensions Monopolism in the economy Lack of political stability/democracy High tax rates High crime rates Unemployment Harsh bank credit terms Weak technological development Infrastructural problems Do not know Total Frequency Percent 337 147 65 215 22 37 12 2 81 15 13 45 11 1002 33.63 14.67 6.49 21.46 2.20 3.69 1.20 0.20 8.08 1.50 1.30 4.49 1.10 100.00 12 T. SADIGOV Table 3. Impact of value orientations on bribe offers in Azerbaijan. Downloaded by [UAE University] at 12:55 25 October 2017 Logit-logged coefficients Model 1 Model 2 Personality more important than laws 1.817 (3.16)*** 1.812 (3.14)*** Adherence to a belief system 0.581 (1.16) 0.577 (1.15) Corruption depends on circumstances − 0.507 (−8.04)*** −0.502 (−7.98)*** Usage of laws −0.382 (−7.53)*** −0.382 (−7.49)*** Level of paternalism 0.050 (1.16) 0.047 (1.08) High personal responsibility for uprooting corruption − 0.087 (−1.88)* −0.088 (−1.90)** Cause of social problems lies in technical solutions rather than in weak −0.213 (−4.14)*** −0.211 (−4.09)*** morality Domestic vs. External orientation −0.146 (−2.86)** −0.151 (−2.92)*** Willingness to spend on Prestige 0.022 (3.03)*** 0.0233 (3.06)*** Associational membership 0.297 (0.94) 0.322 (1.01) Not following news 0.203 (0.59) 0.210 (0.61) Monthly income 0.001 (3.71)*** 0.001 (3.72)*** Trust to others −0.050 (−0.83) −0.050 (−0.82) Choice to avoid bribes – −0.516 (−0.83) Constant 4.853 (4.86)*** 5.331 (4.59)*** Observations 1002 1002 2 0.6866 0.6873 Pseudo R Log likelihood −166.450 −166.090 Notes: Absolute value of z-statistics in parentheses. *p < .10 (significant at 10% in a two-tailed t-test). **p < .05 (significant at 5%). ***p < .01 (significant at 1%.). N = 1002 face-to-face respondents. avoid bribes − are statistically significant at the 1% level. Perception by a respondent of an individual responsibility to uproot corruption is significant at the 6% level, and preference for external/internal orientation is significant at the 3% level. The introduction of a subjectively felt choice to avoid bribes (the variable itself not statistically significant in our model) only slightly affects the predictive power of the model. Overall, both models are quite strong, judged by their pseudo R 2 and log likelihood measures (the log likelihood in ‘iteration 0’ line equals = −531.183, while the final log likelihood is much closer to 0 and equals = −166.090). Predicted probability estimation also shows robust predictive power of our models (Table 4). Five statistically significant variables are positively associated with bribing: If a respondent thinks that the personality of a bureaucrat is more important than laws in public administration operation; if a respondent does not follow news; the more money a respondent spends on markers of prestige; and finally, the more a respondent earns ‒ the more s/ he is likely to offer a bribe. The remaining variables are negatively associated with bribing, also conforming to our model prediction: The more a respondent thinks that the occurrence or lack of corruption is defined by standardized norms rather than depends on specific circumstances; the greater her usage of laws to solve daily social problems; the higher her feeling of personal responsibility for the uprooting of corruption; the higher her perception that the cause of social problems lies in technical solutions rather than in weak ideology and morality; the higher her domestic (vs. external) orientation − the lower is the likelihood that s/he will engage in a bribe offer. Since in all of the independent variables with the range from 0 to 10, zero was associated with the most likelihood of bribing, the positive association between the constant and the dependent variables was also predicted by our model. The results lend support to our Hypotheses 1, 2, 3, 5, 6, 7, 10, and 14. Both of the localization/particularization factors are statistically significant, indicating that in the Azerbaijani context ordinary citizens use opportunities to hijack laws and regulations the same way as bureaucrats do. Symbolic factors are mostly significant in predicting corruption. JOURNAL OF CIVIL SOCIETY 13 Table 4. Predicted probability of a bribe offer by an average Azerbaijani respondent. logit: Predictions for BRIBED. Confidence intervals by delta method 95% Conf. Interval Pr(y = 1x): 0.9465 [0.9248, 0.9681] Pr(y = 0x): 0.0535 [0.0319, 0.0752] Downloaded by [UAE University] at 12:55 25 October 2017 Variables Personality more important than laws Adherence to a belief system Corruption depends on circumstances Usage of laws Level of paternalism High personal responsibility for uprooting corruption Cause of social problems lies in technical solutions rather than in weak morality Domestic vs. External orientation Willingness to spend on Prestige Associational membership Not following news Monthly income Trust to others Choice to avoid bribes x= 0.89720559 .91716567 4.5748503 3.5518962 3.9181637 4.2305389 3.754491 4.001996 27.090818 .23253493 .3003992 486.89222 4.3363273 .93013972 The respondents’ choices are tilted more towards the symbolic rather than content realm. This may indicate that the Azerbaijani social context is dominated by an exclusivist social framework, emphasizing in-group belonging more than merit-based openness. The exclusivist closed circles, with contradistinction between those who are favoured and outsiders, create a milieu susceptible more to corruption rather than to merit-based interaction. These circles and groups gather around conspicuous consumption, prestige, status, and spending on luxurious trinkets. The mean share of income that the respondents of our survey were ready to spend on items of status, prestige, and luxury (expensive car plates, phone numbers, and jewellery) on average amounted to 27.09% of all their annual income. Institutional factors show mixed effects on corruption. The choice to avoid bribes does not predict respondents’ practice of bribing: Although more than half do have a choice, more than three-quarters of them nevertheless paid a bribe in the past 12 months. This may indicate that respondents pro-actively use bribes to attain benefits, thus offering bribes voluntarily in the majority of the bribing cases. This is corroborated by descriptive data: The majority of those who offered a bribe did so for various utilitarian reasons (to save time and money – 35.2%, or to avoid potential red tape – 27.7%). Although income is among the statistically significant variables, this does not support our hypothesis 11. The hypothesis predicts a negative association between the variables (the less income, the more a respondent is likely to offer a bribe), while logistic regression shows a positive association between the variables (the higher the income, the more a respondent is likely to offer a bribe). Most of the hypotheses not supported by the regression are associated with respondents’ social capital. Specifically, such factors as associational membership, adherence to any belief system, following news on media, and interpersonal trust all are statistically insignificant. The results indicate that the factors of social capital and trust are not effective predictors of corruption in Azerbaijan. These results indicate that ascriptive identities and bonding social capital (family, kinship, clan belonging) play a defining role, while bridging Downloaded by [UAE University] at 12:55 25 October 2017 14 T. SADIGOV social ties are insignificant. Indeed, the Azerbaijani population has substantial trust in familial and kinship ties in contrast to wider and more formal ties of bridging nature. The analysis of Caucasus Research Resource Center (CRRC) data shows that the ‘levels of trust and association diminish … in relation to more distant, non-familial social networks … Thus, Azerbaijanis’ trust in family members, friends, and immediate neighbors is generally high, while their involvement in civic groups remains low’ (Liles, 2012). Additionally, such factors as a wide use of client–patron relations in political culture and the administration based on neo-patrimonialism (Guliyev, 2012, p. 119) also contribute to the reign of informality. As a result, the participation of the Azerbaijani population in formal institutions and organizations is almost imperceptible in comparison to wide employment of informal networks. Indeed, the results of my survey show that whereas only 18.9% of the respondents are members of at least one national association or organization, the share of those who contributed at least one hour of their time for the development of their local community is 39.5%. Research on small and medium entrepreneurs in Azerbaijan in 2011 revealed the lack of any significant association between the inclination of the entrepreneurs to pay bribes and their participation in formal nationwide associations, organizations, or movements, although there was a significant impact of participation of the respondents in informal local associations on the probability of bribe offers (Sadigov, 2013). The explanation might be that in Azerbaijani society local connections, local politics, personalized informal allegiance have a greater appeal for an individual in comparison to nationwide formal politics and nationwide associations. Social capital, trust, and associational membership have no meaningful effect on corruption because these factors are minor to non-existent in Azerbaijan. As a result, we see weaker social responsibility and a greater propensity to disregard social rules and to prefer nepotism and corruption among the respondents in Azerbaijan. None of the variables of our logistic regression model demonstrated the level of standard error exceeding 1. However, we test our model for multicollinearity to identify possible specification errors, if any. Table 5 displays correlations among our independent variables. As is evident from the results, correlation among our independent variables is moderate and varies between 0.01 (1%) and 0.4936 (49%). Additional collinearity checks also testify to a low interdependence among the independent variables. While small-to-moderate multicollinearity is fairly widespread, in cases ‘when severe multicollinearity occurs, the standard errors for the coefficients tend to be very large (inflated), and sometimes the estimated logistic regression coefficients can be highly unreliable’ (Chen, Ender, Mitchell, & Wells, 2012). Using STATA command collin, we obtain collinearity measures for our independent variables (Table 6). As seen from the output, all of the variables have VIF scores less than 2. The results testify to low collinearity among our independent variables. As a measure of model fit, we can calculate PRE (Proportional Reduction in Error) – a useful function that identifies the level of predictive leverage of logistic regression independent variable over dependent variable. PRE is an effective measure of model fit that provides measurement by compar[ing] the percentage of cases the independent variables predicted correctly to the percentage of cases that would have been correctly predicted if the researcher adopted a naïve Morality Choice Corruption vs. Domestic Not Trust to Belief depends on Usage Level of Personal Technical vs. External Associational following Monthly to avoid Personality system circumstances of laws paternalism responsibility solutions orientation Prestige membership news income others bribes Personality Belief system Corruption depends on circumstances Usage of laws Level of paternalism Personal responsibility Morality vs Technical solutions Domestic vs. External orientation Prestige Associational membership Not following news Monthly income Trust to others Choice to avoid bribes 1 0.304 −0.465 1 −0.318 1 −0.378 −0.308 −0.201 −0.344 −0.374 −0.198 −0.229 −0.051 −0.104 −0.233 0.494 0.346 0.180 0.379 0.480 0.186 −0.335 0.179 −0.145 −0.1236 −0.054 0.215 −0.340 0.118 −0.121 −0.1098 −0.026 −0.368 0.266 −0.188 0.020 0.201 0.030 1 0.281 0.172 0.393 0.418 1 0.202 0.238 0.284 1 0.137 0.162 1 0.411 −0.247 0.219 −0.101 −0.034 0.2155 0.064 −0.231 0.306 −0.172 0.164 0.1707 0.010 −0.184 0.096 −0.109 −0.012 0.1759 0.022 −0.229 0.158 −0.120 0.067 0.1478 0.054 1 −0.314 0.201 −0.107 0.083 0.1687 −0.012 1 −0.103 0.101 0.113 −0.176 0.032 1 −0.116 0.130 0.1582 0.044 1 −0.094 −0.040 0.0002 1 −0.043 −0.076 1 0.029 1 JOURNAL OF CIVIL SOCIETY Downloaded by [UAE University] at 12:55 25 October 2017 Table 5. Correlation matrix. 15 16 T. SADIGOV Downloaded by [UAE University] at 12:55 25 October 2017 Table 6. Collinearity Diagnostics. Variables VIF SQRT VIF R-squared Eigenval Cond index Personality more important than laws Belief system Corruption depends on circumstances Usage of laws Contractarian vs. Paternalist position High personal responsibility for uprooting corruption Cause of social problems in technical solutions than in weak morality Domestic vs. External orientation Prestige Associational membership Not following news Monthly income Trust to others Choice to avoid bribes Mean VIF 1.53 1.27 1.82 1.53 1.32 1.11 1.24 1.13 1.35 1.24 1.15 1.05 0.6529 0.7867 0.5508 0.6522 0.7596 0.8998 0.3471 0.2133 0.4492 0.3478 0.2404 0.1002 9.8025 1.2714 0.7874 0.6611 0.5283 0.3911 1.0000 2.7767 3.5285 3.8508 4.3076 5.0065 1.37 1.17 0.7277 0.2723 0.3127 5.5993 1.54 1.28 1.28 1.07 1.12 1.11 1.02 1.31 1.24 1.13 1.13 1.03 1.06 1.05 1.01 0.6510 0.7796 0.7808 0.9341 0.8923 0.9019 0.9791 Condition number: 27.3196 0.3490 0.2204 0.2192 0.0659 0.1077 0.0981 0.0209 0.2752 0.2523 0.2289 0.1998 0.1466 0.0585 0.0131 5.9681 6.2336 6.5445 7.0037 8.1775 12.9448 27.3196 Tolerance Table 7. Proportional Reduction in Error results of the logistic regression model. Dependent variable: BRIBED Using cut-off score of .5 Classification table Predicted values of BRIBED 0 1 Total Actual values of BRIBED 0 1 188 35 223 18 761 779 Total 206 796 1002 Percent Correctly Predicted = 94.71%. Percent in Modal Category = 77.74%. Proportional Reduction in Error = 76.23%. Expected PCP = 91.04%. Expected PMC = 65.40%. Expected PRE = 74.11%. model that only used the central tendency of the dependent variable to make predictions. (Arceneaux, 2013) PRE reflects Michael Herron’s ‘expected percentage correctly predicted’ and ‘expected proportional reduction in error’ factors for limited dependent variable models. Thus, the employment of PRE would give us a better measure of how much the model improves with regard to random choice. Table 7 shows the PRE results for our logistic regression model. The results lend strong support for our model: PRE for the model is 76.2%, more or less consistent with the pseudo R 2 from the model (0.6873). The test shows 94.7% prediction correctness for our logistic model. Conclusion In recent years, Azerbaijan – one of the worst performers on TI CPI index – has made sustained progress. The improvement was mainly attributed to the top-down introduction by the government of measures that raise bureaucratic transparency. The main focus of this article has been an assessment of the readiness of Azerbaijani ordinary citizens to Downloaded by [UAE University] at 12:55 25 October 2017 JOURNAL OF CIVIL SOCIETY 17 sustain this improvement at the grassroots. This research has looked at a variety of relevant variables that might explain citizen attitudes to and practices of bribe offers. The scale of corruption among the population is substantial. The research identified that for the past 12 calendar months before the research, 77% of the respondents offered a bribe at least once. However, the intensity of bribe offers remains moderate (the majority of the respondents offered between 1 and 3 bribes). The population’s bribe offers constitute around half of the entire variation in corruption in Azerbaijan. Citizens generally show social passivity and are not ready to stand up to corruption. This may partly be explained by the argument that the Azerbaijani social context is dominated by an exclusivist social framework, emphasizing in-group belonging over merit-based openness. Indeed, ‘the boundary between formal and informal in Azerbaijan is very fluid and easily penetrable’ (Sayfutdinova, 2015, p. 9). As a result, we see weaker social responsibility, and greater propensity to disregard social rules, and prefer nepotism and corruption among the respondents in Azerbaijan. Most of the hypotheses not supported by the regression are associated with respondents’ social capital. Specifically, such factors as associational membership, adherence to any belief system, following news on media, and interpersonal trust all are statistically insignificant. The results indicate that the factors of social capital and trust are not effective predictors of corruption in Azerbaijan. The explanation might be that in Azerbaijani society local connections, local politics, and personalized informal allegiance have a greater appeal for an individual in comparison to nationwide formal politics and nationwide associations. The pseudo-R 2 of our logistic regression model is 0.6873, while PRE (Proportional Reduction in Error) is somewhat better (76.2%). The results show that our model accounts for more than half of the variation in corruption, while visibly improving with regard to random choice. Since our models focus mainly on the supply side of corruption, the remaining half of the variation in corruption may potentially be due to other factors that go beyond the focus of our research: Level of institutional opportunities for corruption; impact of indigenous traditions of integrity and neighbourhood; influence of the nature of the administrative and political system (authoritarian, democratic, mixed); and other variables usually associated with a bribe demand (i.e., bureaucracy-initiated corruption). The scale and the depth of the problem of citizen-initiated bribe offers indicate that a successful fight against corruption cannot be confined to the eradication of only bureaucratic bribe demand, but needs to take into account corruption choices of the population too. The research revealed that ordinary citizens, although blaming corruption for social ills, at the same time are widely using bribe offers to advance personal utilitarian interests. Corruption may be a symptom – a symptom of the ineffectiveness of formal institutions and rules – on top of being a serious impediment to socio-economic development. The Azerbaijani case of corruption is illustrative for several reasons. First, it shows that government anti-corruption initiatives may not be sustainable if not supported by the practice of the wider population. Since the local social and value context is an important factor mediating anti-corruption policies, such policies need to take account of longstanding collective values and institutions. Second, there is a big gap between the perception of corruption and the practice of it. While a citizen may decry corruption, s/he Downloaded by [UAE University] at 12:55 25 October 2017 18 T. SADIGOV simultaneously may offer bribes to officials. This shows the limited nature of corruption perception indices and research. The research findings indicate that citizen-initiated bribe offers account for a considerable share of corruption in the post-Soviet context. The population may acknowledge the problem as one of the biggest ones that the post-Soviet societies face. However, the same population, to an equal degree, also feeds corruption itself − by widely using bribes to buy the support of officials, simultaneously not standing up to corruption, and generally by disregarding laws, rules, and institutions. An important corollary of the research, then, is that institutions as framers of ‘the rules of game’ are not developed because, despite high demand for impartial institutions and the rule of law, there is no support for them within a wider society. On a policy dimension, the findings of the research do not bode well with the general thrust of anti-corruption campaigns that the post-Soviet governments have launched recently. For instance, the Azerbaijani anti-corruption campaign is characterized by a selective rather than holistic approach. Since the launch of the campaign, several high-profile officials and administrators have been either dismissed or downgraded (Rosenblum, 2011). According to reports, generally, all of the public officers were advised ‘to avoid irritating the population and to work effectively and build public trust’ (Ismayilova, 2011). However, this selective approach addresses only half of the problem (the demand of bribes), while leaving untouched the supply side of it. More importantly, it neglects the formidable argument that the problem of corruption is tightly woven into the social fabric of a post-Soviet society. Simple administrative measures focusing on bureaucracy do little to affect institutional, economic constraints, and sociocultural factors that citizens face as recent Russian and Georgian anti-corruption policies have vividly shown. The anti-corruption campaign may yield incomplete results in a country, where not only bureaucrats, but wide masses of the population are eager participants in bribing. Notes 1. 2004 is the first year in which TI CPI record was collected for all of the mentioned states. Pre2004 TI CPI reports lack scores for several resource-rich post-Soviet countries. 2. Information about sampling can be obtained from the author upon request. Disclosure statement No potential conflict of interest was reported by the author. Funding This work was supported by the Open Society Foundation. References Arceneaux, K. (2013). Generating a percent reduction in error (PRE) statistic using Stata. Retrieved from http://astro.temple.edu/~arceneau/pre.htm Banfield, E. (1958). The moral basis of a backward society. Glencoe, IL: The Free Press. Downloaded by [UAE University] at 12:55 25 October 2017 JOURNAL OF CIVIL SOCIETY 19 Bardhan, P. (1997). Corruption and development: A review of issues. Journal of Economic Literature, 35(3), 1320–1346. Birch, S. (2009). The case for compulsory voting. Public Policy Research, 16(1), 21–27. Chen, X., Ender, P., Mitchell, M., & Wells, C. (2012). Logistic regression with Stata. Retrieved from http://www.ats.ucla.edu/stat/stata/webbooks/logistic/chapter3/statalog3.htm Domínguez, J. I., Greene, K. F., Lawson, C. H., & Moreno, A. (Eds.). (2014). Mexico’s evolving democracy: A comparative study of the 2012 elections. Baltimore, MD: John Hopkins University Press. Engvall, J. (2015). The state as investment market: A framework for interpreting the post-soviet state in Eurasia. Governance: An International Journal of Policy, Administration and Institutions, 28, 25–40. Gellner, E. (1983). Nations and nationalism. Ithaca, NY: Cornell University Press. Green, D. (2012). From poverty to power: How active citizens and effective states can change the world. Oxford: Oxfam International. Gudkov, L., & Dubinin, B. (2001). Obshestvo telezriteley: massy i massoviye kommunikatsiyi v Rossiyi v kontse 90-x godov [A society of TV-viewers: The masses and media in Russia at the end of 1990s]. In Russian. Monitoring Obshestvennogo Mneniya #2 (52). March–April. Retrieved from http://ecsocman.hse.ru/data/197/991/1219/05gudkov-31-45.pdf Guliyev, F. (2012). Political elites in Azerbaijan. In A. Heinrich & H. Pleines (Eds.), Challenges of the Caspian resource boom. Domestic elites and policy-making (pp. 117–130). Houndmills: Palgrave Macmillan. Harrison, L. E. (1997). The Pan-American dream: Do Latin America’s cultural values discourage true partnership with the United States and Canada? New York, NY: Basic Books. Harrison, L. E. (2000). Promoting progressive cultural change. In L. E. Harrison & S. P. Huntington (Eds.), Culture matters: How values shape human progress (pp. 296–307). New York, NY: Basic Books. Holmes, L. (2012). Corruption in post-Soviet Russia. Global Change, Peace & Security, 24(2), 235–250. Huntington, S. (1968). Political order in changing societies. New Haven, CT: Yale University Press. Ismayilova, K. (2011, January 31). Azerbaijan: Protests in Egypt Are Reverberating in Baku. Eurasianet. Retrieved from http://www.eurasianet.org/node/62802 Karklins, R. (2005). The system made me do it: Corruption in post-communist societies. Armonk, NY: M.E. Sharpe. Klitgaard, R. (1998). International cooperation against corruption. Finance & Development, 35(1), 3–7. Kostadinova, T. (2009). Abstain or Rebel: Corruption perceptions and voting in East European elections. Politics and Policy, 37(4), 691–714. Kostadinova, T. (2012). Political corruption in Eastern Europe: Politics after communism. Boulder, CO: Lynne Rienner. Ledeneva, A. (1998). Russia’s economy of favors. Cambridge: Cambridge University Press. Ledeneva, A. (2003). Informal practices in changing societies: Comparing Chinese Guanxi and Russian Blat (Working Paper No. 45). Centre for the Study of Economic and Social Change in Europe: University College London. Ledeneva, A. (2006). How Russia really works. The informal practices that shaped post-Soviet politics and business. Ithaca, NY: Cornell University Press. Ledeneva, A. (2011). Open secrets and knowing smiles. East European Politics and Societies, 25(4), 720–736. Ledeneva, A. (2013a). Can Russia modernize? Sistema, power networks and informal governance in Putin’s Russia. New York, NY: Cambridge University Press. Ledeneva, A. (2013b). Sistema, power networks and informal governance in Putin’s Russia. The World Financial Review. Retrieved from http://www.worldfinancialreview.com/?p=1040 Liles, T. (2012). Trust and agency in Azerbaijan: Personal relationships versus civic institutions. CRRC - Social Science in the Caucasus Blog. Retrieved 15 November, 2012 & August 10, 2015, from http://crrc-caucasus.blogspot.com/2012/11/trust-and-agency-in-azerbaijan-personal. html Downloaded by [UAE University] at 12:55 25 October 2017 20 T. SADIGOV Maine, H. S. (1861). Ancient law. London: John Murray. McDevitt, A. (2015). The state of corruption: Armenia, Azerbaijan, Georgia, Moldova and Ukraine. Berlin: Transparency International. Mocan, N. (2004). What determines corruption? International evidence from micro data (National Bureau of Economic Research Working Paper No. 10460). Cambridge, MA: National Bureau of Economic Research (NBER). Moses, J. C. (2003). Dilemmas of transition in post-soviet countries. Chicago, IL: Burnham. Mungiu-Pippidi, A. (2006). Corruption: Diagnosis and treatment. Journal of Democracy, 17(3), 86– 99. Obydenkova, A., & Libman, A. (2015). Understanding the survival of post-communist corruption in contemporary Russia: The influence of historical legacies. Post-Soviet Affairs, 31(4), 304–338. Putnam, R. (1993). Making democracy work: Civic traditions in modern Italy. Princeton, NJ: Princeton University Press. Roaf, J. (2000). Corruption in Russia. European II Department, IMF Conference on Post-Election Strategy, Moscow, April 5–7. Retrieved from http://www.imf.org/external/pubs/ft/seminar/2000/ invest/pdf/roaf.pdf Rose-Ackermann, S. (1978). Corruption: A study in political economy. New York, NY: Academic Press. Rose-Ackermann, S. (1997). Corruption and development. Washington, DC: The World Bank: Annual Bank Conference on Development Economics. Rose-Ackermann, S. (1999). Corruption and government: Causes, consequences and reforms. Cambridge: Cambridge University Press. Rosenblum, S. (2011). How Serious is Azerbaijan’s Anti-Corruption Campaign? CACI Analyst. Retrieved from http://www.cacianalyst.org/?q=node/5506 Rothstein, B. (1998). Just institutions matter: The moral and political logic of the universal welfare state. Cambridge: Cambridge University Press. Rothstein, B., & Uslaner, E. (2005). All for all: Equality, corruption, and social trust. World Politics, 58(1), 41–72. Sadigov, T. (2013). Corruption and social responsibility: Bribe offers among small entrepreneurs in Azerbaijan. East European Politics, 30(1), 34–53. Sayfutdinova, L. (2015). Negotiating welfare with the informalizing state: Formal and informal practices among engineers in post-Soviet Azerbaijan. Journal of Eurasian Studies, 6(1), 24–33. Schein, E. (2010). Organizational culture and leadership (4th ed.). San Francisco, CA: Jossey-Bass. Shleifer, A., & Vishny, R. (1993). Corruption. The Quarterly Journal of Economics, 108(3), 599–617. Sindzingre, A. (2011). A comparative analysis of African and East Asian corruption. In A. J. Heidenheimer & M. Johnston (Eds.), Political corruption: Concepts and contexts (3rd ed., pp. 441–462). New Brunswick: Transaction. Stiglitz, J. E. (2002). Globalization and its discontents. New York, NY: W.W. Norton. Stockemer, D. (2011). Women’s parliamentary representation: Are women more highly represented in (consolidated) democracies than in non-democracies? Contemporary Politics, 15(4), 429–443. Stockemer, D., LaMontagne, B., & Scruggs, L. (2013). Bribes and ballots: The impact of corruption on voter turnout in democracies. International Political Science Review, 34(1), 74–90. Temple, P., & Petrov, G. (2004). Corruption in higher education: Some findings from the states of the former Soviet Union. Higher Education Management and Policy, 16(1), 83–99. Transparency International. (2014). Azerbaijan National Integrity System Assessment 2014. Retrieved from http://files.transparency.org/content/download/1330/10305/file/2014_ NISAzerbaijan_EN.pdf World Bank. (2013). Corruption is ‘public enemy number one’ in developing countries, says World Bank Group President Kim. Press Release. Retrieved from http://www.worldbank.org/en/news/ press-release/2013/12/19/corruption-developing-countries-world-bank-group-president-kim You, J.-S. (2015). Democracy, inequality and corruption: Korea, Taiwan and the Philippines compared. Cambridge: Cambridge University Press.