• No results found

The causes and consequences of corruption

N/A
N/A
Protected

Academic year: 2021

Share "The causes and consequences of corruption"

Copied!
204
0
0

Loading.... (view fulltext now)

Full text

(1)   . THE CAUSES AND CONSEQUENCES OF CORRUPTION      . A THESIS PRESENTED IN FULFILMENT OF THE REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY IN ECONOMICS  . March 2011. BIN DONG        .   School of Economics and Finance Faculty of Business Queensland University of Technology Gardens Point Campus Brisbane Australia.  .

(2)  . Acknowledgements  . I would like to express my gratitude to my principal supervisor Professor Benno Torgler, for walking me through the journey of PhD study and for being there at every step as a source of inspiration, motivation and moral support. Professor Torgler’s excellent supervision, invaluable guidance, suggestions, corrections and empirical skills have helped shape much of this thesis. I would also like to extend deepest appreciation to my associate supervisor Professor Uwe Dulleck for his invaluable guidance and encouragement throughout this study. I extend my thanks to Dr. David Johnston for offering invaluable comments and suggestions for empirical analysis. I also extend special thanks to the members of the administrative staff of the School of Economics and Finance. I am extremely grateful to China Scholarship Council and Queensland University of technology for jointly providing me the financial support for my PhD study. I am deeply indebted to my wife Jin Kang, who has been the motivational force in my life, and thank her for her patience, understanding and invaluable support during the PhD study. I am grateful to my mother, brother for their selfless and unreserved support over the years. Finally I would like to thank Ms Ying Zhou, Mr Tony Beatton and Mr Markus Schaffner for their assistance and support.. i   .

(3)

(4)  . Abstract  . This thesis comprehensively studies the causes and consequences of corruption in both crosscountry and within-country contexts, mainly focusing on China. The thesis commences by extensively investigating the causes of corruption. Using the standard economic approach, this study finds that in China regions with more anti-corruption efforts, higher education attainment, Anglo-American historic influence, higher openness, more access to media, higher relative wages of government employees, and a greater representation of women in legislature are markedly less corrupt; while the social heterogeneity, deregulation and abundance of resources, substantially breed regional corruption. Moreover, fiscal decentralization is discovered to depress corruption significantly. This study also observes a positive relationship between corruption and the economic development in current China that is mainly driven by the transition to a market economy. Focusing on the influence of political institutions on corruption, the thesis then provides evidence that a high level of political interest helps to reduce corruption within a society, while the effect of democracy upon corruption depends on property rights protection and income distribution. With the social economic approach, however, the thesis presents both cross-country and within-country evidence that the social interaction plays an important role in determining corruption. The thesis then continues by comprehensively evaluating the consequences of corruption in China. The study provides evidence that corruption can simultaneously have both positive and negative effects on economic development. And it also observes that corruption considerably increases the income inequality in China. Furthermore this study finds that corruption in China significantly distorts public expenditures. Local corruption is also observed to substantially reduce FDI in Chinese regions. Finally the study documents that corruption substantially aggravates pollution probably through a loosening of the environmental regulation, and that it also modifies the effects of trade openness and FDI on the stringency of environmental policy. Overall, this thesis adds to the current literature by a number of novel findings concerning both the causes and the consequences of corruption. Key Words: corruption, causes, consequences, China, democracy, social interaction, political interest, economic development. iii   .

(5)  . Table of Contents Acknowledgements ................................................................................................................................. i  Statement of Original Authorship ......................................................................................................... ii  Abstract ................................................................................................................................................. iii  Table of Contents .................................................................................................................................. iv  List of Tables ......................................................................................................................................... vii  List of Figures ......................................................................................................................................... ix  Chapter One . Introduction .................................................................................................................. 1 . 1.1 Motivation of Thesis ..................................................................................................................... 1  1.2 Content of Thesis .......................................................................................................................... 2  1.2.1 Causes of Corruption .............................................................................................................. 2  1.2.2 Consequences of Corruption ................................................................................................. 7  1.2.3 Methodology issues ............................................................................................................... 8  1.3 Structure of Thesis ....................................................................................................................... 10  Chapter Two . Economic Determinants of Corruption: Chinese Evidence ....................................... 12 . 2.1 Introduction ................................................................................................................................ 12  2.2 Determinants of Corruption ........................................................................................................ 15  2.3 Empirical Analysis ........................................................................................................................ 18  2.3.1 Province‐level Analysis ......................................................................................................... 20  2.3.2 City‐level Analysis ................................................................................................................. 35  2.4 Conclusion ................................................................................................................................... 38  Appendix ........................................................................................................................................... 41  Chapter Three . Political Interest and Corruption: Cross‐country Evidence .................................. 44 . 3.1 Introduction ................................................................................................................................ 44  3.2 Political Interest .......................................................................................................................... 45  3.2.1 Theoretical Considerations .................................................................................................. 45  3.2.2 A Simple Model .................................................................................................................... 46  3.3 Data ............................................................................................................................................. 48  3.3.1 Dependent Variables ............................................................................................................ 48  3.3.2 Measuring Political Interest ................................................................................................. 50  3.3.3 Further Independent Variables ............................................................................................ 51  3.4 Empirical Evidence ...................................................................................................................... 56  3.4.1 International Evidence ......................................................................................................... 57  iv   .

(6)   3.4.2 Within‐Country Evidence ..................................................................................................... 70  3.5 Conclusion ................................................................................................................................... 71  Chapter Four  Democracy and Corruption: Cross‐country Evidence ............................................... 73  4.1 Introduction ................................................................................................................................ 73  4.2 Theoretical Model ....................................................................................................................... 75  4.2.1 Model ................................................................................................................................... 76  4.2.2 Economic Equilibrium .......................................................................................................... 77  4.2.3 Political Equilibrium ............................................................................................................. 78  4.3 Empirical Evidence ...................................................................................................................... 80  4.3.1 Methodology and Data ........................................................................................................ 81  4.3.2 Results .................................................................................................................................. 85  4.4 Conclusion ................................................................................................................................... 92  Appendix ........................................................................................................................................... 93  Chapter Five . Social interaction and Corruption: Cross‐country Evidence ..................................... 94 . 5.1 Introduction ................................................................................................................................. 94  5.2 Theoretical Foundation ................................................................................................................ 96  5.2.1 Background of the Model .................................................................................................... 98  5.2.2 A Simple Game ................................................................................................................... 100  5.2.3 Dynamics ............................................................................................................................ 102  5.2.4 Conditional Corruption – Discussion and Extensions ........................................................ 102  5.3 Data and Methodological Approach ......................................................................................... 103  5.3.1 Micro Analysis .................................................................................................................... 103  5.3.2 Macro Analysis ................................................................................................................... 107  5.4 Results ....................................................................................................................................... 107  5.4.1 Micro Level using the EVS .................................................................................................. 107  5.4.2 Micro Level using the WVS ................................................................................................ 117  5.4.3 Macro Level Using a Large Panel Data Set ......................................................................... 121  5.5 Conclusion ................................................................................................................................. 125  Chapter Six . Social interaction and Corruption: Within‐country Evidence ................................. 127 . 6.1 Introduction .............................................................................................................................. 127  6.2 Theoretical Model ..................................................................................................................... 129  6.3 Empirical Work .......................................................................................................................... 132  6.3.1 Data and Methodology ...................................................................................................... 132  6.3.2 Results ................................................................................................................................ 136  v   .

(7)   6.4 Conclusion ................................................................................................................................. 139  Appendix ......................................................................................................................................... 141  Chapter Seven . Consequences of Corruption: Chinese Evidence ................................................ 142 . 7.1 Introduction .............................................................................................................................. 142  7.2 Literature Review ...................................................................................................................... 145  7.3 Empirical Analysis ...................................................................................................................... 149  7.3.1 Data and Methodology ...................................................................................................... 149  7.3.2 Corruption, Economic Growth and Income Distribution ................................................... 153  7.3.3 Corruption and Foreign Direct Investment ........................................................................ 160  7.3.4 Corruption and Public Expenditures .................................................................................. 162  7.3.5 Corruption and the Environment ....................................................................................... 164  7.4 Conclusion ................................................................................................................................. 169  Appendix ......................................................................................................................................... 171  Chapter Eight . Conclusion ............................................................................................................ 172 . 8.1 Summary of Findings ................................................................................................................. 172  8.2 Policy Implications .................................................................................................................... 174  8.3 Further Research ....................................................................................................................... 175  References .......................................................................................................................................... 177 . vi   .

(8)  . List of Tables Table 2‐1 GDP (PPP) per capita of Chinese regions in 2008 (Intl. $) ..................................................... 14  Table 2‐2 Average annual registered cases on corruption across regions in China (1998‐2007) ......... 20  Table 2‐3 Corruption and its determinants in China: pooled OLS estimation ...................................... 32  Table 2‐4 Corruption and its determinants in China: fixed effects OLS estimation ............................. 33  Table 2‐5 Corruption and its determinants in China: fixed effects 2SLS estimation ............................ 34  Table 2‐6 Corruption and its determinants in Chinese cities ................................................................ 37  Table 2‐7 Data description .................................................................................................................... 41  Table 2‐8 Pairwise correlation coefficients between explanatory variables in the province‐level  analysis .................................................................................................................................................. 42  Table 2‐9 Pairwise correlation coefficients between explanatory variables in the city‐level analysis . 42  Table 2‐10 First‐Stage Regressions Based on Table 2‐5 ........................................................................ 43  Table 3‐1 Justifiability of corruption and political discussion ............................................................... 59  Table 3‐2 Perceived corruption and political discussion ...................................................................... 60  Table 3‐3 Justifiability of corruption and interest in politics ................................................................ 61  Table 3‐4 Perceived corruption and political interest .......................................................................... 62  Table 3‐5 Justifiability of corruption and important of politics in life .................................................. 63  Table 3‐6 Perceived corruption and importance of politics in life ....................................................... 64  Table 3‐7 2SLS results ........................................................................................................................... 67  Table 3‐8 Justifiability of corruption in Switzerland ............................................................................. 68  Table 3‐9 Perceived corruption in Switzerland ..................................................................................... 69  Table 4‐1  Descriptive Statistics ............................................................................................................ 84  Table 4‐2 Effect of democracy on corruption: review and implication (fixed effects results) ............. 86  Table 4‐3 Effect of democracy on corruption: fixed effect results ....................................................... 87  Table 4‐4 Effect of democracy on corruption: alternative measure of democracy.............................. 88  Table 4‐5 Effect of democracy on corruption: IV results ...................................................................... 89  Table 4‐6 Marginal effect of democracy on corruption ........................................................................ 91  Table 4‐7 Validity of instrument: Muslim ............................................................................................. 93  Table 5‐1 Influence of conditional corruption (EVS) ........................................................................... 110  Table 5‐2 Robustness test and the influence of conditional corruption using micro and macro proxies  (EVS) .................................................................................................................................................... 113  Table 5‐3 2SLS results (EVS) ................................................................................................................ 115  Table 5‐4 Causality discussion (Filtering) ............................................................................................ 116  Table 5‐5 Conditional corruption using WVS ...................................................................................... 119  Table 5‐6 2SLS results (WVS) .............................................................................................................. 120  Table 5‐7 Causality discussion filtering with WVS Data ...................................................................... 121  Table 5‐8 Evidence at the macro level ................................................................................................ 125  Table 6‐1 Literature summary ............................................................................................................ 129  Table 6‐2 Variables description, 1998—2007 ..................................................................................... 134  Table 6‐3 Corruption and social interaction ....................................................................................... 138  Table 6‐4 Average annual registered cases on corruption per capita across regions in China (1998‐ 2007) ................................................................................................................................................... 141  vii   .

(9)   Table 6‐5 Pairwise correlation coefficients between variables .......................................................... 141  Table 7‐1 GDP (PPP) per capita of Chinese regions in 2008 (Intl. $) ................................................... 144  Table 7‐2 Average annual registered cases on corruption per capita across regions in China (1998‐ 2007) ................................................................................................................................................... 150  Table 7‐3 Effect of corruption on economic growth: cross‐province evidence .................................. 154  Table 7‐4 Effect of corruption on economic growth: cross‐city evidence .......................................... 158  Table 7‐5 Effects of corruption on income inequality ........................................................................ 159  Table 7‐6 Effect of corruption rate on inbound FDI ............................................................................ 161  Table 7‐7 Effects of corruption on public expenditures ..................................................................... 163  Table 7‐8 Effect of corruption and the environment .......................................................................... 165  Table 7‐9 Effect of corruption on environment policy ....................................................................... 168  Table 7‐10 Data description ................................................................................................................ 171 .                                   viii   .

(10)  . List of Figures Figure 1‐1 Thesis Structure ................................................................................................................... 11  Figure 2‐1 Determinants of corruption in China ................................................................................... 38  Figure 4‐1 Relationship between democracy and corruption .............................................................. 85  Figure 5‐1 Correlation between Justifiability of Corruption and Perceived Corruption ....................... 96  Figure 5‐2 Description of the Corruption Game ................................................................................. 101  Figure 5‐3 Correlation between Justifiability of Corruption and Control of Corruption .................... 106  Figure 7‐1 Marginal effects of trade openness on environmental stringency conditional on corruption  ............................................................................................................................................................ 166                                         . ix   .

(11)  . Chapter One. Introduction.  . 1.1 Motivation of Thesis Corruption, understood as ‘‘abuse of public office for private gain’’ is a persistent feature in human societies throughout time and space. Contemporaneous corruption scandals not only occur in developing countries such as Nigeria, India, and China where corruption is regarded as a norm, but also in developed economies such as France, Britain and America. The sale of parliamentary seats in ‘rotten boroughs’ in England before the Reform Act of 1832 1 and ‘machine politics’ in larger cities in America in the late 19th and early 20th century2 are two famous historical examples. Even in Scandinavian countries, like Sweden and Norway, which are supposedly free-from-corruption, managers of state owned companies have been found to take bribes. Corruption in the public sector is viewed as the major obstacle to economic development (Kaufmann, 1997). Solid evidence (for example, Mauro 1995, and World Bank, 1997) demonstrates the pernicious effects of corruption upon, among other things, investment, economic growth, environmental quality and therefore social welfare. In effect, a country is adversely affected by the existence of corruption, and therefore anti-corruption policies are important. Reducing corruption requires a precise understanding of its causes and consequences. The development of effective anti-corruption policies is based on a thorough investigation of corruption within and across countries. However, in current research the causes and consequences of corruption remain poorly understood and are broadly disputed. As a result, it is difficult for governments to design coherent policies to control corruption. This study provides new insight into the causes and consequences of corruption. We explore the discussed factors in a within country environment to provide evidence outside the US and in a more controlled environment, and also provide within-country and cross-country evidence at the micro level to explore new theories in the area of corruption such as conditional corruption. In detail the thesis first empirically examines the theoretical causes of corruption suggested in literature in both cross-country and within-country contexts. The                                                              1.  Pearce, R. and Stearn, R., 2000. Access to History, Government and Reform: Britain 1815-1918 (Second Edition). Hodder & Stoughton. 2  Clifford, T. P., 1975. The Political Machine: An American Institution. Vantage Press.. 1   .

(12)  . author robustly identifies the effects of its economic, political and social determinants on corruption, employing solid statistical tools dealing with causal relationship between observed factors. Secondly, this study comprehensively investigates the various consequences of corruption, focusing on China, the largest developing country central to world economy. According to my knowledge, there are few studies on the consequences of Chinese corruption. In summary, this study is expected to make a substantial contribution to the research of the causes and consequences of corruption, and therefore to add to effective policy guidelines to curb corruption.. 1.2 Content of Thesis This section provides a succinct portrait of the thesis. Initially we briefly review previous literature to build up the logical framework of the thesis (specific literature reviews are provided in each of the individual chapters that follow). Then based on the framework that has been introduced, the main findings of the thesis are then presented. 1.2.1 Causes of Corruption There has been a wave of empirical studies on the causes and consequences of corruption in recent years. With respect to the causes of corruption, this study, similar to Bardhan (2006), points out that there are generally two different approaches to research the causes of corruption, namely the standard economic approach and also the social economic approach. The standard economic approach emphasizes incentives and punishments in corrupt acts following Becker’s analytical framework (1968). According to this approach, there are three prerequisites necessary for the incidence of corruption (Jain, 2001). First, bureaucrats have discretionary power. Second, this power is associated with economic rents. Finally, the deterrence to corruption, as a function of the probability of being caught and the penalty for the corrupt act, is adequately low. The first two preconditions determine the benefit of corruption, while the last precondition influences the cost of corruption. Many studies adopting this approach concentrate upon economic conditions and policies influencing the cost and/or benefit of corruption. Literature shows that regulation and decentralization are the main determinants of the discretionary power of a government. Economic rents, on the other hand, increase with natural resource abundance, but decrease with economic competition proxied by trade openness. All of these factors are observed to substantially affect the benefit of corruption (e.g. Ades and Di Tella, 1999; Fisman and Gatti, 2002a, b).. 2   .

(13)  . The deterrence of corruption is a joint function of the possibility of being detected and the punishment once caught. High levels of economic development, education attainment and media access have been documented to reduce corruption by raising the possibility that corrupt acts are detected (for example, Treisman, 2000). Historical influence also plays an important role in corruption (see also, Treisman, 2000). Furthermore corruption has also been found to be negatively correlated with female representation in politics, possibly because women may feel a larger probability of being caught in an act of corruption (e.g. Dollar et al., 2001). Social and economic heterogeneity is also an indirect determinant of the probability of corrupt acts being caught. For example, ethnical fractionalization is believed to promote corruption since corrupt officials may be protected by their own ethnic groups for political reasons (see also Treisman, 2000). Finally, the relatively high wage of the public sector implies a high opportunity cost when officials are ousted due to corruption. As a proxy for the punishment, the (relative) wage of the public sector is found to be negatively associated with the corruption level (e.g. Van Rijkeghem and Weder, 2001) Studies on the causes of corruption by and large perform cross-national analyses using subjective survey data (for example, Treisman, 2000; Fisman and Gatti, 2002a). This kind of study, although fruitful, cannot circumvent two problems. Firstly, subjective survey data might be biased as Treisman (2007) argues: “the data do not measure corruption itself but opinion about its prevalence” (p. 215). Secondly, cross-country analysis often suffers from omitted variable bias. Substantial unobservable or unmeasurable differences in institution and culture between countries make cross-country results problematic. The disadvantages experienced by cross-national studies can be avoided if we use within-country objective data, since objective data do not suffer the bias of subjective data. Furthermore, homogeneity within a country also mitigates the omitted variable bias troubling the cross-country analysis. However, current within-country data are only proxies for corruption since corruption is actually secretive and hence difficult to measure directly. Goel and Nelson (1998), Fisman and Gatti (2002b), and Glaeser and Saks (2006) utilize the objective data: the number of public officials convicted for abuse of public office as an indicator of the actual levels of corruption in American states. However, this indicator may also reflect the anti-corruption efforts of local judiciary. As Lambsdorff (2005) point out “the appropriateness of such data as a proxy for corruption has thus been widely disputed” (p.1). Therefore the ideal strategy might be to investigate the causes of corruption with both the cross-country analysis using subjective data and the within-country analysis using objective data to get complementary results, which also makes empirical findings robust. 3   .

(14)  . While there have been enormous cross-country studies, papers on the causes of corruption using within-country data are few, and most of them are working with US data. This thesis hence initially contributes to literature with a study on the causes of corruption in China. A study of China has a unique advantage. Firstly, it is helpful to understand corruption in developing and transitional economies where it is one of the central issues. Secondly, China is well-suited for a study on corruption. On the one hand, China is fairly homogenous in institutions, culture, and social structure. This helps us to mitigate the omitted variable bias in empirical analysis. On the other hand, there are great economic differences between the eastern and western provinces in China, which might make findings of corruption in China more generalizable on a global level. Chapter 2 in the thesis adopts a standard economic approach to explore the causes of corruption in China using two different data sets, namely a province-level data set and a citylevel data set, to obtain robust results. China is a key player in the world economy and will gain even further importance in the future. This study examines almost all cross-country findings in a Chinese context using the regional number of registered cases on corruption as a measure of corruption. Besides confirming most cross-country findings in a more controlled setting, this study adds to literature in several ways. Firstly, this chapter use behavioural variables3 rather than attitudinal variables (perceptions of corruption) to proxy for corruption. Secondly anti-corruption efforts are always controlled in this study to isolate the component of anticorruption efforts in our corruption measure though many studies including the small amount of studies using a behavioural proxy for corruption have neglected this. Thirdly, this study identifies a positive relationship between corruption and economic development (and marketization) in China due to the transition process in China. Fourthly, this study provides some novel within-country evidence such as the negative effect of British historic influence, the positive effect of natural resource abundance, and the negative effect of female representation in politics on corruption. Fifthly, this study presents solid evidence that even in a nondemocratic country the access to controlled media still checks corruption. There are a number of studies concentrating on the influence of political institutions on corruption using the standard economic approach. Good political institutions help to control and monitor the government and therefore reduce corruption. There are actually two kinds of political institutions: Formal institutions such as democracy and informal ones such as                                                              3.  The registered cases on corruption in procurator’s offices of provinces, and the average ratios of the travel and entertainment costs relative to the sales of investigated firms in Chinese cities in the survey conducted by World Bank in 2005. . 4   .

(15)  . political interest. Econometrically, the effect of formal institutions like democracy on corruption can only be analysed at the macro level in the cross-country context, while political interest is useful to analyse (informal) institutions at the micro/individual level. Democracy is theoretically supposed to reduce corruption mainly because political competition may provide checks against corruption. “In democratic systems, competitors for office have an incentive to discover and publicize the incumbent’s misuse of office whenever an election beckons.” (Treisman, 2000, p. 404) This therefore raises the possibility that corrupt acts can be detected. However, the relationship between corruption and democracy is empirically found to be complex. Besides the linear relationship mentioned above, a quadratic relationship between these factors, is also supported by several theoretical and empirical articles (e.g. Mohtadi and Roe, 2003, Rock, 2007). Moreover, some scholars such as Treisman (2000) suggest that it may take a long time for democracy to substantially reduce corruption. Further evidence is clearly necessary. It is worth noting however that variation in political institution within a country is not large enough in many cases for economists to identify the relationship between democracy and corruption. This thesis therefore mainly utilizes cross-country data sets to examine the influence of political institutions on corruption. The thesis, for the first time ever, investigates in Chapter 3 the relationship between political interests: an informal aspect of political institution and corruption since citizens’ political interest contributes to the probability of their being involved in the political process (Verba et al., 1995). Innovatively, this study uses the micro-level data from the World Values Survey to explore the impact of political interest represented by three different proxies on both the perception of corruption and the justifiability of corruption reflecting the social norm of corruption. It is worth noting that unlike the macro-level analysis which is popular in the corruption study, the micro-level study is able to measure the individual characteristics and induce robust relationships due the large amount of observations. Furthermore, as can be seen below, it allows researchers to explore new theories such as conditional corruption. Specifically, this study first performs a cross-country analysis with a huge data set, and then runs a within-country analysis focusing on Switzerland to check the robustness of crosscountry results. Both analyses clearly show that a high level of political interest helps to reduce the level of corruption within a society. In Chapter 4 the thesis shed new light on the relationship between democracy and corruption. To disentangle the actual linkage from previously mixed evidence, this study first establishes a new political economy model demonstrating that the effect of democracy on corruption is conditional on income distribution and property rights protection. Then with a 5   .

(16)  . cross-national panel data, this study clearly shows that the previous empirical findings lose significance when considering the interactions between corruption, property rights protection and income distribution. The effect of democracy upon corruption is empirically observed to depend on the protection of property rights and income inequality. This thesis hence provides new theoretical and empirical evidence concerning the effect of democracy on corruption.     The social economic approach however insists that corruption to some extent arises from. social norms. It emphasizes the role of group dynamics as well as culture and history, in determining corruption. Put simply, corrupt people will feel less guilt (moral cost) if they find many others engaged in similar activities, and vice versa. Culture here coordinates the expectation on others’ behaviour in a society, while history provides the initial condition. Multiple equilibria are often expected in this circumstance. A society with an initially high corruption level may get ‘‘locked in’’ until a ‘big push’ similar to what happened in Hong Kong takes place (Aidt, 2003). Economists however have not focused on this approach heretofore. Only Goel and Nelson (2007) have studied the contagion of corruption in America. This thesis however attempts to fill the gap with both cross-country and within-country analyses. The study presented in Chapter 5, according to my knowledge, is the first crosscountry analysis studying the role of social interactions or social norms in corruption. This study builds first a behavioural model to innovatively argue that engaging in corruption results in a disutility of guilt. Guilt itself depends on a (current and past) perceived prevalence of corruption within a society. As a novelty the empirical section presents a large amount of evidence about the role of social interactions with two large micro level data sets and a large macro level panel data sets covering almost 20 years. The results clearly indicate that a willingness to be corrupt is influenced by the perceived activities of others, and the past level of corruption. The findings above therefore underscore the relevance of social interactions on the area of corruption. Furthermore the results also complement a large set of laboratory experimental studies (for example, Falk, Fischbacher and Gächter, 2003) that have studied conditional cooperation by providing evidence outside of a lab setting. As mentioned before, the study above may suffer both omitted variable bias and subjective data bias. Chapter 6 of this thesis then, from a different angle, turns to examine the role of social interaction on corruption within a Chinese context using province-level panel data. As a novelty, this study, unlike Goel and Nelson (2007), simultaneously investigates the impacts of both the corruption level of neighbours and the corruption level in the past upon contemporary corruption. Robust evidence is also presented that social interaction plays an 6   .

(17)  . important role in determining corruption rates in China. The thesis therefore contributes to literature both by cross-country and by within-country evidence on the relevance of social interaction in corruption. 1.2.2 Consequences of Corruption Corruption is believed to have a detrimental effect on economic development and hence social welfare. Many studies examine the relationship between corruption and economic growth since there is indeed a debate on the effect of corruption on economic growth. Some scholars (for example, Leff, 1964, and Huntington, 1968) argue that corruption may improve efficiency and hence promote economic growth by allowing enterprisers to circumvent cumbersome regulations with bribes especially in developing countries. However, the majority of literature insists that corruption lowers economic growth because it may reduce the incentive of private investment (Bardhan, 1997), distort public investment decisions (Tanzi and Davoodi, 1997), and induce talented people into rent-seeking activities (Murphy, Shleifer and Vishny, 1991). Most empirical studies indeed support the fact that corruption impedes economic growth mainly through channels of investment, openness and political instability (for example, Mauro, 1995, Mo, 2001). Specifically, corruption is found to reduce foreign direct investment (e.g. Wei, 2000a) because high corruption in host countries may imply high expropriation risk. Moreover, Fredriksson et al. (2003) show that corruption may influence FDI through another channel: environmental regulation. On the other hand, corruption may distort public investment. According to Mauro (1998), corrupt politicians may increase public expenditure easy to collect bribes, while decreasing expenditure providing fewer bribery opportunities. Furthermore he empirically observes that corruption significantly reduces public expenditure on education. Corruption also substantially affects income distribution. Gupta, Davoodi and AlonsoTerme (2002) find that corruption significantly increases income inequality, while Li et al. (2000) observe that corruption influences income inequality in a reversed U-shaped manner. The adverse effects of corruption on the environment are also documented in literature. Welsch (2004) found that corruption aggravates pollution especially in developing countries, while Cole (2007) provides seemingly contradicting evidence. More investigation is therefore needed. Pellegrini and Gerlagh (2006a, b) however provide solid evidence that corruption has a substantially negative effect on the environment policy stringency, which may imply that corruption affects pollution mainly through environment policy making. Furthermore, both 7   .

(18)  . theoretical and empirical evidence has shown that corruption not only reduces the stringency of environmental policy but also modifies the effects of other determinants of environment policy (Fredriksson and Svensson, 2003, Damania et al., 2003, and Cole et al., 2006). In Chapter 7 this thesis, for the first time, comprehensively investigates consequences of corruption with complementary Chinese data sets and alternative corruption measures. The study contributes to existing literature in several ways. Firstly, it suggests a novel perspective of the influence of corruption on economic growth. Specifically, this study empirically finds that corruption has simultaneously both positive and negative effects on the economic growth of China. The overall impact of corruption might be the balance of the two effects, both of which may depend on institutional environments. Secondly, this study provides, novel within-country evidence that corruption increases income inequality, decreases FDI and distorts public spending in China. Thirdly, this study documents that corruption substantially aggravates pollution probably through loosening environmental regulation in China, and that it modifies the effects of trade openness and FDI on the stringency of environmental policy in a similar manner to that observed in literature to date. There are still two important issues to be addressed. First, one may notice there is an imbalance in regards to the number of papers on the causes of corruption compared to the consequences of corruption. The current research project actually starts with the study about the causes of Chinese corruption since a better understanding of the causes of corruption is first of all required before one is able to analyse the consequences of corruption. The author will however complete the project with more papers studying the consequences of Chinese corruption (for example, from the micro/individual perspective) in the near future. Second, it is worth noting that chapters 2 to 7 in this thesis are respectively built on independent papers submitted to or published in academic journals. In order to retain the completeness of the research project in each chapter, some overlap between chapters especially chapters concerning corruption in China is allowed. Furthermore, the writing styles of chapters in the thesis are a bit different due to the different requirements of journals the chapters submitted to. 1.2.3 Methodology issues This thesis attempts to reliably identify the causality between corruption and other relevant factors. The important methodology issues which are the key to identification are briefly discussed here. Detailed methodology will be described in the following chapters. The first issue we need to address is how to measure corruption. Indices of perceived 8   .

(19)  . corruption such as the Corruption Perceptions Index published by Transparency International and a corruption rating constructed by the Political Risk Services have been often used to measure corruption in many cross-country studies. These indices are actually based on the subjective assessments of experts or survey respondents of the extent of corruption in various countries. The subjective indices are indeed closely correlated with each other although they are complied by different organizations with different methodologies, suggesting “that these different spyglasses are aimed at a common target” (Treisman 2007, p.216). Furthermore, the perceptions indices are proved to be highly correlated with a range of generally believed corruption determinants, indicating that they are “a helpful contribution to the understanding of real levels of corruption” (Lambsdorff 2004, p.6). However, as Treisman (2007) argue, corruption perception data actually reflect impressions of corruption intensity rather than corruption itself, meaning that the data are actually correlated with survey respondents’ beliefs and other social and economic conditions (see also, Knack 2006). Such data, therefore, cannot be convincingly used as dependent variables because their measurement error is associated with many other background characteristics that are affected by explanatory variables (Bertrand and Mullainathan 2001). The objective data set, usually within-country, however can eliminate this kind of subjective data bias. For example, using the number of public officials convicted for abuse of public office in American states as a proxy for corruption, Glaeser and Saks (2006) examine the causes of and consequences of corruption in America. However Lambsdorff (2004) argue that this kind of objective data may reflect the quality of the judiciary rather than the actual corruption level. The reasonable strategy in the research of corruption is therefore complementally using subjective and objective data to measure corruption. Corruption is econometrically a messy environment to analyse. According to Leamer (1983), sensitivity analysis is therefore required here to ensure the credibility of the current study. The detailed strategy that this thesis adopts is conducting a lot of robustness tests,  exploring the issue with different data sets, proxies for corruption, and specifications, and using micro and macro, within country and cross-country evidence, to show a robust picture. Besides, the endogeneity issue needs to be addressed when identifying the causality in the study. Two strategies are utilized to remove the potential endogeneity bias. The first strategy is, whenever possible, to control for unobserved individual characteristics influencing both corruption and relevant factors by including individual fixed effects in our panel regressions. As Mo (2001) points out, “Corruption is commonly considered to be an institutional problem that lasts for a long period” (p. 70). Fixed effect regressions therefore are suitable for the 9   .

(20)  . investigation of the relationship between corruption and other factors since the major source of potential bias in our regressions might be time-invariant historical factors. However, fixed effect regressions do not necessarily identify the causality between corruption and other relevant factors. Fixed effects regressions cannot guarantee the causality, since there might be time-varying omitted factors affecting both corruption and the relevant factors. Fixed effects are not a substitute for instrumental variables. The second strategy to address the endogeneity problem therefore is to adopt the instrumental variable approach to identify the causality between corruption and relevant factors in our (fixed effects) regressions. The key issue of the IV approach, the selection of instrument variables, will be discussed in detail as the analysis proceeds.. 1.3 Structure of Thesis As mentioned before, this thesis mainly focuses on corruption in China, the largest developing country in the world. According to the logic framework set up above, this study is organised as follows. The first part of the thesis is Chapter 1, where an introduction of the study is presented. The second part of the thesis is composed of Chapters 2, 3, 4, 5, and 6, where the causes of corruption especially in China are extensively investigated. The standard economic approach is first adopted in Chapter 2, 3 and 4. Chapter 2 examines the effects of the economic determinants of corruption suggested by cross-country studies in China with two different data sets. Chapter 3 and 4 analyse the political determinants of corruption with cross-country data sets. Specifically, Chapter 3 investigates the relationship between political interest and corruption, while Chapter 4 explores the effect of democracy upon corruption. Findings of these two chapters may help to predict the influence of democratisation on corruption in China although China now is an authoritarian country. Chapter 5 and 6, however, follow the social economic approach to examine the role of social factors in determining corruption. Both cross-country analyses and within-country analyses (China) are performed here to make the findings solid since few studies have been previously undertaken in this area. The third part of the thesis, namely Chapter 7, comprehensively investigates the adverse effects of corruption on the economic development in China at a macro level. Finally this thesis finishes with the fourth part, Chapter 8, which provides a summary of all the findings above and some concluding remarks. The detailed structure of this thesis is shown in Figure 1-1.. 10   .

(21)   Figure 1-1 Thesis Structure Introduction. INTRODUCTION.  . Chapter 1. Economic determinants of corruption Standard Economic Approach. Economic determinants of corruption: Chinese evidence. Chapter 2. Political determinants of corruption CAUSES OF CORRUPTION. Social Economic Approach. Chapter 3. Democracy and corruption: cross-country evidence. Chapter 4. Social determinants of corruption. CONSEQUENCES OF CORRUPTION. CONCLUSION. Social interaction and corruption: cross-country evidence. Chapter 5. Social interaction and corruption: Chinese evidence. Chapter 6. Consequences of corruption: Chinese evidence. Chapter 7. Conclusion. 11   . Political interest and corruption: cross-country evidence. Chapter 8.

(22)  . Chapter Two. Economic Determinants of Corruption: Chinese Evidence4.  . 2.1 Introduction Since the establishment of the People’s Republic of China in 1949, corruption has vexed the national leadership, which prior to 1978 attempted to control it primarily through mass movements but occasionally with severe deterrents like the 1952 execution of two senior officials, Qingshan Liu and Zishan Zhang. Since the 1978 launch of economic reform, however, corruption has become even more widespread and according to Liu (1983), exists at every level of China’s political system. As a result, the 1989 market price of coal, for example, was 674 percent of the subsidized price, other producer goods sell at prices substantially higher than those fixed by the state, and payoffs to ensure the supply of products at state prices are very common (Rose-Ackerman 1999). Corruption in the form of applicant bribery is also widespread in the area of enterprise licensing because industrial and commercial enterprises in China must obtain government authorization to operate (Manion 1996). Liu (1983) thus differentiates between three types of corruption: “corrupt acts such as embezzlement and bribes, which are common place among nations having a political system to speak of; … appropriation of public goods, illegal trade, and housing irregularity, [which result] from a breakdown in the central allocation system and [are] commonplace among socialist nations … [and the] rather peculiarly Chinese Communist [practices of] illegitimate feasting, feudal rites, false models, and illegal imprisonment and torture” (p. 603). Even the Chinese government has admitted that corruption “is now worse than during any other period since New China was founded in 1949. It has spread into the Party, into Government administration and into every part of society, including politics, economy, ideology and culture” (Liang 1994, p. 122). The seriousness of this problem is exemplified by the recent charges against two members of the Politburo, Xitong Chen and Liangyu Chen, who accepted huge bribes. Not surprisingly, such rampant corruption, which seems to be a distinct feature of contemporary China, the largest transitional and developing country, has generated much literature, especially in sociology and political science (e.g, White 1996; Wedeman 2004; Gong 2006). From an economics perspective, Yao (2002) argues that corruption in China is generated by the Chinese political system, which grants and protects privileges, and Cai et al.                                                              4.  This chapter is under revision for resubmission to the Public Choice. . 12   .

(23)  . (2009), using an innovative measure of corruption in Chinese firms, find that corruption significantly reduces firm productivity. Nevertheless, no empirical study yet exists that comprehensively analyses the economic underpinnings of corruption in China. Rather, the majority of extant studies on the causes of corruption are cross-national investigations that use subjective survey data. For instance, Treisman’s (2000) comprehensive cross-country study employs several indices of perceived corruption to explore the causes of corruption. These studies, although fruitful, are subject to the problems of subjective bias and omitted variable bias. First, as Treisman (2007) admits, corruption perception data actually reflect impressions of corruption intensity rather than corruption itself, meaning that the data are actually correlated with survey respondents’ beliefs and other social and economic conditions (see also, Knack 2006). Such data, therefore, cannot be convincingly used as dependent variables because their measurement error is associated with many other background characteristics that are affected by explanatory variables (Bertrand and Mullainathan 2001). Second, the substantial number of unobservable or unmeasurable differences in institutions and cultures between countries makes it difficult for cross-country analyses to solve the omitted variable bias. Admittedly, some cross-country analysts have attempted to bypass this bias by using fixed-effect regressions, however, as Treisman (2007) and Knack (2006) point out, the appropriateness of using some subjective corruption indices in longitudinal analyses remains questionable. Such disadvantages in cross-national research can certainly be mitigated by the use of an objective within-country data set that eliminates the subjective data bias and, despite some regional differences, provides a higher level of homogeneity that moderates the omitted variable bias to which cross-country analyses are subject. In this respect, studies of China have a unique advantage: China is a centralized country with unified legal and administrative systems and a fairly homogenous society dominated in most areas by Han ethnicity and Confucian values. This high degree of legal and social homogeneity helps to efficiently mitigate the omitted variable bias in any empirical analysis. On the other hand, as shown in Table 2-1, great economic differences exist between China’s rich eastern and poor western provinces. For example, in 2008, the GDP (PPP) per capita of Shanghai, which approximates that of Hungary, was about nine times higher than that of Guizhou province, which resembles that of Cameroon. Thus, a study of China can provide valuable insights into the causes of corruption in developing and transitional economies in which corruption is a central issue. Surprisingly, however, few studies on the causes of corruption employ within-country data and most that do are working with U.S. data. For instance, Goel and Nelson (1998) 13   .

(24)  . investigate the effect of government size on corruption using an American annual state-level data set, while Fisman and Gatti (2002b) use information on the mismatch between revenue generation and expenditure in American states to test the relationship between decentralization and corruption. More recent studies by Leeson and Sobel (2007) and Boettke et al. (2008) show that those American states struck most frequently by natural disasters attract more disaster relief, creating new opportunities for political corruption comparable to resource windfalls and therefore setting in motion rent-seeking activities. Boettke et al. (2008) stress that the Federal Emergency Management Agency relief “is especially corrosive in terms of corruption because of the chaotic atmosphere in which it is unavoidably deployed. In the case of a major disaster, the combination of billions of dollars of relief being dumped onto one location in only a short period of time, along with the confused and difficult-tomonitor environment in which these windfalls are dispensed, create incredible temptation for public officials to abuse their positions of power by corruptly appropriating relief funds” (p. 367). Internationally, Svensson (2003) use firm-level data from Uganda to explore the determinants of firm bribery payments, while Cai et al. (2009) employ firm-level data to examine the “micro” causes of corruption in China.                        Table 2-1 GDP (PPP) per capita of Chinese regions in 2008 (Intl. $) Beijing. 16577. Anhui. 3810. Chongqing. 4741. Tianjin. 14590. Fujian. 7922. Sichuan. 4044. Hebei. 6112. Jiangxi. 3887. Guizhou. 2321. Shanxi. 5365. Shandong. 8701. Yunnan. 3310. Inner Mongolia. 8472. Henan. 5153. Tibet. 3646. Liaoning. 8221. Hubei. 5223. Shaanxi. 4799. Jilin. 6184. 4608. Gansu. 3185. Heilongjiang. 5714. Guangdong. Hunan. 9886. Qinghai. 4573. Shanghai. 19232. Guangxi. 3936. Ningxia. 4706. Jiangsu. 10421. Hainan. 4517. Xinjiang. 5232. Zhejiang. 11102. In this chapter, we adopt both fixed-effect and instrumental variable (IV) approaches to identify the causes of corruption in China using different regional data sets. Besides confirming most cross-country findings in a more controlled setting, our study makes three important contributions to the literature. First, we identify a positive relationship between corruption and economic development in China, one that stems from the current transition process. Second, we obtain novel within-country evidence on the depressive effect of the 14   .

(25)  . Anglo-American colonial heritage, the contributory effect of abundant natural resources, and the depressive effect of female representation in politics on corruption. Third, we find that even in a non-democratic country, access to controlled media keeps corruption in check. The chapter is organized as follows: Section 2.2 reviews previous research on the causes of corruption, Section 2.3 empirically determines the causes of corruption in China, and Section 2.4 presents our concluding remarks.. 2.2 Determinants of Corruption Previous research has identified several possible causes of corruption, including political institutions, the judicial system and the cultural environment; however, as these factors are homogenous among Chinese regions, we focus here on other determinants. According to Jain (2001), there are three prerequisites for corruption: bureaucratic discretionary power, the association of this power with economic rents, and deterrence as a function of the probability of being caught and penalized. Whereas the first two preconditions determine the benefit of corruption, the third influences the cost of corruption; therefore, the regional characteristics that affect these preconditions determine its local incidence (Becker, 1968). Bureaucratic discretionary power over the allocation of resources is particularly important to the existence of corruption and, according to Rose-Ackerman (1978), frequently arises during the enforcement of regulations. That is, because bureaucrats can assign themselves the discretion to distribute resources when setting and implementing regulations, more regulations means more discretionary power and thus more incidences of corruption. In contrast, levels of corruption can be expected to decrease if controlled economies become more marketized. Governmental discretionary power can also be influenced by decentralization, although the relationship between decentralization and corruption is still being debated. According to Brennan and Buchanan (1980) and Weingast (1995), decentralization introduces competition between local governments, thereby reducing bureaucratic profits from corruption. For example, the mechanism of entry and exit in U.S. federal states provides a strong incentive to produce public services in accordance with individual preferences (Hirschman 1970) and can be a method of government control, as when exits threaten firms with higher mobility (Rose-Ackerman 1999). Nevertheless, because federalism and local autonomy combine with innovation, federalism can also serve as a laboratory for effective policy inventions (Oates 1999). On the other hand, Shleifer and Vishny (1993) argue that since decentralization causes the dispersion of government power, bureaucrats that are not coordinated will over-extract rents from firms. Likewise, smallness 15   .

(26)  . and intimacy of local jurisdictions with patronage-ridden governments promote corrupt relationships (Rose-Ackerman 1999). In fact, Treisman (2000), using a dummy variable that reflects whether a state is federal, finds that federal states are seen as more corrupt. Fisman and Gatti (2002a), however, provide cross-country evidence that fiscal decentralization in government expenditure is significantly correlated with lower corruption. Using American data, they also identify a positive relationship between corruption and the proportion of a state’s expenditure derived from federal transfers (Fisman and Gatti 2002b). Obviously, rational individuals pay bribes only if they can reap a higher marginal benefit from doing so. Hence, economic rents related to discretionary powers are a necessary condition for corruption, but corruption is unlikely to be generated by discretionary powers without related rents. Indeed, Ades and Di Tella (1999) show that countries in which firms have higher rents tend to be more corrupt. One concentrated and easily expropriable activity of particularly high rents is natural resource exploitation (Sachs and Warner 2001), which echoes Leite and Weidmann’s (1999) empirical finding that the incidence of corruption depends significantly on natural resource abundance. Treisman (2000), on the other hand, finds no strong evidence that fuel and mineral exports are positively correlated with corruption level, although, intriguingly, Leeson and Sobel (2008) report that resource windfalls generated by disaster relief in frequently affected American states raise public corruption in much the same manner as rich natural resource endowments. Another source of economic rents is lack of competition: economic rents decrease when economic activities are marked by intensive competition. For instance, Ades and Di Tella (1996, 1999), using a country’s openness – measured by share of imports in the GDP – to indicate firms’ external competition, find that economic openness is negatively correlated with levels of corruption. Treisman (2000) also provides evidence of a negative association between the share of imports in GDP and corruption levels, a relationship similar to that recently identified by Gerring and Thacker (2005) between corruption and trade openness. Deterrence of corruption is a joint function of the probability of detection and punishment once caught, a probability affected by several factors. First, higher income levels accelerate the spread of education and democratic institutions and therefore enhance individuals’ political involvement. They consequently enable private individuals to better identify corrupt behaviours and punish official malfeasance. Hence, regions with richer and more educated citizens are assumedly less corrupt. In fact, according to Treisman (2007), the negative relationship between the incidence of corruption and income level is the strongest and most consistent finding of empirical studies on corruption (see, e.g., La Porta et al. 1999; 16   .

(27)  . Ades and Di Tella 1999; Treisman 2000). The probability of being caught also depends on the effectiveness of the country’s legal system. For instance, La Porta et al. (1999) argue that the common law systems in Britain and its former colonies are more effective in protecting property rights and enforcement than civil law systems, which would imply that the probabilities of corruption being exposed are higher in common law countries. Treisman (2000) does indeed find that, as expected, Britain and its former colonies have substantially lower levels of corruption than other countries; however, Pellegrini and Gerlagh (2008) find no such linkage. Economic and social heterogeneity may also be an indirect determinant of the probability of detection and thereby affect corruption. For example, You and Khagram (2005) argue that “the poor are more vulnerable to extortion and less able to monitor and hold the rich and powerful accountable as income inequality increases” (p. 136). Thus, income inequality enables the latter to abuse their power for private gain and, as the authors confirm through cross-country analysis, promotes higher levels of corruption. Husted (1999), on the other hand, finds no such relationship between income inequality and corruption. One social heterogeneity factor with the potential to promote corruption is ethnic fractionalization, which may lead to corrupt officials being protected for political reasons by their own ethnic groups. Nevertheless, neither Treisman (2000) nor Pellegrini and Gerlagh (2007) find strong evidence of such a linkage, although Glaeser and Saks’ (2006) results do show a positive correlation between corruption levels and racial division in U.S. states. Press freedom also plays an important role in corruption detection because independent journalists have incentives to investigate its presence or absence. Therefore, as a particular mechanism of external control, press freedom appears to reduce corruption: firms and individuals can reveal corrupt behaviour to a journalist and this possibility of media reporting increases the costs of corruption for bureaucrats (i.e., increases the probability of detection). In other words, the media can be seen as a platform for voicing complaints (Brunetti and Weder 2003), and, as Adsera et al. (2003) show, even the “free circulation of a daily newspaper” (their interaction term between a democratic measure and newspaper circulation) is negatively correlated with corruption (p. 455). Likewise, Brunetti and Weder (2003) show empirically that a high level of press freedom is associated with a low incidence of corruption, Chowdhury (2004) emphasizes that press freedom controls corruption via the channel of democracy, and Freille et al. (2007), using a modified extreme bounds analysis, provide evidence that the greater the press freedom, the lower the level of corruption.. 17   .

References

Related documents

After the people in the city of Troy went to sleep, the Greeks came out of the body of the horse.. They killed the lookout guards and lit a torch on top the wall to signal the

For eksempel påvirkes sammenlignbarheten av at man i Danmark kun har dekket heltidsstudenter i utvalget for Eurostudent (dette gjør at det er færre studenter over 30 år,

For this purpose a building is to be built on the site to house an exhibition on the history and rise of the Nazi Party (NSDAP) in North Friesland, the history of the

The purpose of this article is to show the need for recreation in the context of foreign leisure trips of administrative staff of selected universities in zachodniopomorskie

The items covered include fundraising, source of funding, investment volume, stage distribution of investment, industrial distribution of investment, the internal rate of return

 ICT   for  Sustainability:  Proceedings  of  the  First  International  Conference  on  Information  and  Communication   Technologies  for

When it is properly formulated, and after the completion of all course work (including all incompletes), the fulfillment of all language requirements, and successful completion of