The impact of economic freedom on corruption: different patterns for rich and poor countries

European Journal of Political Economy Vol. 19 (2003) 605 620 www.elsevier.com/locate/econbase The impact of economic freedom on corruption: different patterns for rich and poor countries P. Graeff a, *, G. Mehlkop b a Department of Sociology, University of Bonn, Adenauer Allee 98a, 53113 Bonn, Germany b Department of Macrosociology, Technical University of Dresden, Mommsenstr. 13, 01069 Dresden, Germany Received 15 February 2002; received in revised form 28 July 2002; accepted 1 August 2002 Abstract This paper investigates the impact of various components of economic freedom on corruption. Some aspects of economic freedom appear to deter corruption while others do not. We identify a stable pattern of aspects of economic freedom influencing corruption that differs depending on whether countries are rich or poor. This implies that there is a strong relation between economic freedom and corruption. This relation depends on a country s level of development. Contrary to expectations, we find that some types of regulation reduce corruption. D 2003 Elsevier B.V. All rights reserved. JEL classification: H10; H11; H50; K20; O5 Keywords: Corruption; Economic freedom; Economic development; Government 1. Introduction Few papers examine the relation between economic freedom and corruption. The empirical results of these studies are consistently the same: the more freedom, the lower the level of corruption, implying that economic freedom is a deterrent to corruption (Chafuen and Guzmán, 2000; Paldam, 2002). Although economic freedom and corruption are phenomena for which the government of a country is an important factor, 1 from a theoretical point of view one would not * Corresponding author. Tel.: +49-0228-73-84-26; fax: +49-0228-73-84-30. E-mail address: ups20004@uni-bonn.de (P. Graeff). 1 Since data for the empirical analysis are only available for corruption amongst public officials, we narrow the scope of the paper inasmuch as we shall refer to corruption concerning public officials only. This may hedge the results of the analysis since corruption occurs outside the government too. 0176-2680/03/$ - see front matter D 2003 Elsevier B.V. All rights reserved. doi:10.1016/s0176-2680(03)00015-6

606 P. Graeff, G. Mehlkop / European Journal of Political Economy 19 (2003) 605 620 expect an empirical link between economic freedom and corruption to arise automatically. On the one hand, a government can impose restrictions on free trade via taxes or licences, which creates the opportunity for civil servants to take bribes or to engage in similar activities. To avoid those restrictions, some people will be willing to pay a bribe to supply civil servants with what they demand. Corruption can also develop when obstructions to economic freedom are imposed (Rose-Ackermann, 1999). On the other hand, a free economy can also increase the pressure to use illegal methods in order to be a step ahead of competitors. This can become more likely if business involves trading with countries that are more prone to use corrupt means of trading. Furthermore, there are different types of corruption depending on the norms and traditions in a society, the level of development and many other factors. It is questionable as to whether economic freedom always influences corruption in the same way, more or less independently of the type of society. Both corruption and economic freedom are multidimensional phenomena (Ades and Di Tella, 1997). According to the empirical result that economic freedom reduces corruption, it is important to examine which aspects of economic freedom influence corruption. While corruption is usually measured as a certain type of illegal behaviour (e.g. of public servants), the most frequently used indices of economic freedom (see Gwartney et al., 2000; Messick, 1996; O Driscoll et al., 2000) comprise various components. The Fraser Index 2000, for example, consists of 23 components that are classified into seven areas covering different aspects of economic freedom. To test the presumed effect of economic freedom on corruption, one has to take a closer look at the performance of these aspects. This paper investigates the impact of economic freedom on corruption (on the basis of the Fraser Index). To answer the question of whether economic freedom always affects corruption in the same way, the effect is calculated for poor countries and for rich countries separately. The paper is organised as follows: in Section 2, we explain the indices of economic freedom and corruption for the empirical study. In Section 3, we demonstrate that there is a pattern of aspects of economic freedom influencing corruption which is different for poor and rich countries. In Section 4, we test the stability of these patterns. Finally, in Section 5 we discuss the results. 2. Measurement of economic freedom and corruption We analyse the impact of economic freedom on corruption by means of the Index of Economic Freedom by Gwartney et al. (2000) and the Corruption Perception Index (CPI) published by Transparency International Berlin. The CPI is the compilation of different survey data. It combines assessments of previous years to reduce sharp variation in scoring which [...] might be due to high-level political scandals that affect perceptions, but do not reflect actual changing levels of corruption (Lambsdorff, 1999, p. 5). The CPI measures corruption from 0 (highest degree of corruption) to 10 (no corruption). A country is taken up in the CPI if at least three surveys are available for the assessment of its level of

P. Graeff, G. Mehlkop / European Journal of Political Economy 19 (2003) 605 620 607 corruption. 2 The measurement of perceived corruption includes, however, assessments of the extent of illegal behaviour in a country in general. 3 Over the last 3 years, especially in 1999, many new countries have been included in the CPI. To extend the sample of countries, the average of the CPI data for every country over 1998 to 2000 has been used. The core ingredients of the Economic Freedom Index are personal choice, protection of private property and freedom of exchange. Individuals have economic freedom when their legally acquired property is protected from invasion by others and when they are free to use and exchange their property without violating the rights of other individuals (see Gwartney et al., 1996). This concept of economic freedom fits into the theoretical framework of methodological individualism which can be seen as a core feature of modern political economy since Adam Smith ([1789] 1976). In this theoretical approach, the institutional framework (economic freedom being a major part of it) within a society affects the incentives of individuals, the productive effort and the effectiveness of resource allocation (De Haan and Sturm, 2000; Jones, 1981; North and Thomas, 1973; Olson, 1996). Freedom of choice, freedom to supply any kind of goods and resources, fair competition in markets (reflected by Area I and II in the index), the availability of reliable money (Area III and IV), secure property rights (Area V), and freedom to trade with others (Area VI) and the allocation of capital by the market (Area VII) are indispensable elements of economic development and progress. 4 An institutional arrangement that is in conflict with economic freedom by restraining trade, increasing transaction costs, undermining property rights and producing uncertainty will reduce the incentives of rational individuals to engage in economic activities. Thus, in an economically free society, the fundamental functions of government are the protection of private property and the enforcement of contracts (De Haan and Sturm, 2000, p. 217), and the provision of a few collective goods (like national defence) that cannot be allocated by a private market (Buchanan, 1975). Anything less could violate the economic freedom of individuals. The expansion of government s size beyond these core activities will have a negative impact on a country s economy (see Gwartney et al., 1998, pp. 168 170) because of the excess burden of higher taxes and/or additional borrowing, for instance. When government s size expands, economic growth will decline. According to Schumpeter (1942), growth is a discovery process driven by entrepreneurs. While private markets reward innovation and progress but punish failure, adjustment to change is slower in the public sector. An expanding government becomes more and more 2 The authors of the CPI made commendable efforts to maximize the reliability of the index by combining various data sources. However, according to Lancaster and Montinola (2001, p. 10) [...] researchers should bear in mind that the index is not necessarily more accurate than any single poll. Determining which poll is most accurate is not possible. Confidence in the CPI assumes each poll is subject to random error. 3 In regression models with corruption as dependent variable it might be problematic to use independent variables that measure some kind of illegal behaviour, too. Area IV of the Economic Freedom Index (Gwartney et al., 2000) includes the element IVb Difference between the Official Exchange Rate and the Black Market Rate. A black market exchange rate premium is both an obstacle to trade and an indicator of unsound money. It is also an indicator for the extent of the shadow economy and illegal practices like corruption. To avoid the dependent variable being confounded with the independent variables we excluded Area IVb from the regressions. 4 For a more detailed description of the Index of Economic Freedom and the areas, see Appendix B.

608 P. Graeff, G. Mehlkop / European Journal of Political Economy 19 (2003) 605 620 Table 1 Correlations of the Economic Freedom areas and corruption (CPI) CPI Area I Area II Area III Area IVa Area V Area VI Area VII Economic Freedom CPI 1.000 0.546** 0.417** 0.513** 0.364** 0.709** 0.521** 0.625** 0.692** 0.000 0.000 0.000 0.004 0.000 0.000 0.000 0.000 86 85 85 85 85 88 76 85 85 Area I 1.000 0.040 0.154 0.043 0.509** 0.345** 0.195 0.169 0.716 0.158 0.696 0.000 0.002 0.073 0.123 85 85 85 85 84 76 85 85 Area II 1.000 0.309** 0.526** 0.352** 0.294** 0.714** 0.752** 0.004 0.000 0.001 0.010 0.000 0.000 85 85 85 84 76 85 85 Area III 1.000 0.287* 0.562** 0.191 0.563** 0.658** 0.008 0.000 0.099 0.000 0.000 85 85 84 76 85 85 Area IVa 1.000 0.362** 0.416** 0.628** 0.713** 0.001 0.000 0.000 0.000 85 84 76 85 85 Area V 1.000 0.430** 0.521** 0.716** 0.000 0.000 0.000 85 75 84 84 Area VI 1.000 0.478** 0.577** 0.000 0.000 76 76 76 Area VII 1.000 0.904** 0.000 85 85 Economic 1.000 Freedom 85 First entry in every cell: correlation coefficient, second entry: level of significance, third entry: number of observations; *p < 0.05; **p < 0.01, two-tailed test. involved in the redistribution of income and regulatory activities which, in turn, encourages individuals to seek income via government favours rather than through production in exchange for income. Gwartney et al. (2000) convert the raw data of the components to a 0 to 10 scale. Higher ratings are indicative of institutions and policies more consistent with economic freedom. The Index of Economic Freedom adds up the area scores to a summary index score. It can be expected that the correlation between the area scores and the summary index score (and to the CPI accordingly) is positive. For Area I (Size of Government) this is not true (see Table 1). 5 Area I correlates negatively with the CPI and the other areas. Since this area is indicative of the size of government, this correlation suggests that bigger governments are exposed to less corruption (a high index score indicates more freedom, i.e. little govern- 5 Note that the number of observations differs somewhat, because there are some missing values within the components or areas.

P. Graeff, G. Mehlkop / European Journal of Political Economy 19 (2003) 605 620 609 ment influence). This contrasts with the view of Becker and Becker (1997, p. 203) who argue that in a small state, public officials should find less opportunity to use public spending for private gain contrary to the general interest. This negative correlation between size of government and the CPI makes the summary index score of economic freedom inadequate for any further investigation of corruption. It is wrong to evaluate the relationship between economic freedom and corruption with this summary index score, because the weight of the suppression effect of the size of government cannot be assigned to the impact of the summary index score on a dependent variable. All other areas show high correlations in Table 1. The signs always point in the expected direction. Therefore, it seems to be adequate to use the various area-indicators to test the relationship between economic freedom and corruption. 3. Pattern of influence of economic freedom on corruption Some theoretical approaches to corruption imply that the type of corruption varies between societies because of their different histories and traditions (Scott, 1972). Nepotism as a typical form of corruption may be likely to arise in societies where particularistic obligations are pervasive. Those obligations are associated with pre-capitalistic, traditional societies (Lipset and Lenz, 2000). Tight family structures and mutual help systems are marks of Eastern countries (Weede, 2000) implying a high level of corruption, since tight networks make it difficult to resist pressure for favourable treatment (especially, if a norm exists that grants unrestricted priority to relatives or friends). With a few exceptions (such as Estonia), Transparency International rates most former communist countries and most African countries as being highly corrupt. In the past, the majority of these countries were characterised by family structures, hierarchical (religious) cultures and party particularism. Most of these countries are also poor (see Appendix A). In recent centuries, the societies of most rich countries have moved towards a market economy. They have abandoned many ideas of traditionalism, encouraged social mobility and aligned decision processes to criteria such as rationality and effectiveness. One of the very apparent indications of the changing society and economy during the development process of rich countries is the declining importance of the family for production processes (Ben-Porath, 1980). Probably, the rich countries have reduced the likelihood of traditional types of corruption in this manner. Nevertheless, they have also created new forms of corruption. The need to finance political campaigns is a source of corruption that would never occur in poor, autocratic systems (Rose-Ackerman, 1999, p. 142). It seems almost impossible to assemble all the various factors of a society that foster or impede corruption (Tanzi, 2000). But if a link between economic freedom and corruption is assumed, the question arises as to whether improvements in economic freedom always affects corruption in the same way regardless of whether the country is poor or rich. We attempted to answer this question by regressing corruption on the seven areas of economic freedom (see Table 2). Model 1 in Table 2 shows the results when the full sample is used. When GDP per capita is also included as a control variable, the areas of economic freedom have only a weak influence on corruption. Areas I and V reveal

610 P. Graeff, G. Mehlkop / European Journal of Political Economy 19 (2003) 605 620 Table 2 Results of OLS- and Backward-Regression (dependent variable: mean of CPI 1998 2000) OLS regression OLS regression Backward regression Model 1 Model 2a Model 2b Model 3a Model 3b all countries poor countries rich countries poor countries rich countries Constant 3.104 3.502 2.979 2.682 0.599 Area I 0.352* 0.079 0.518** 0.542** 2.765 0.469 3.113 3.957 Area II 0.101 0.132 0.088 0.810 0.773 0.542 Area III 0.021 0.141 0.028 0.266 1.847 0.214 Area IVa 0.086 0.150* 0.018 0.148** 1.510 2.509 0.181 3.100 Area V 0.264* 0.204 0.661* 0.560** 2.308 2.001 2.702 2.776 Area VI 0.026 0.206 0.306s 0.157 1.182 1.296 Area VII 0.274 0.280* 0.514 0.257** 0.710** 1.957 2.298 1.751 3.114 4.011 GDP p.c. 0.0001** 4.909 Adj. R 2 0.750 0.322 0.743 0.265 0.758 N 68 29 39 29 39 Reset-Test significant not significant not significant not significant not significant For Backward Regression only significant results are listed. First entry in every cell except the last three rows: unstandardized regression coefficient, second entry: t value; *p < 0.05; **p < 0.01, two-tailed test. A line reports that the variable is not significant at the 5% level. significant coefficients, whereas the sign of Area I is counter-intuitive but was found in earlier studies as well (e.g. Paldam, 2001). 6 To separate poor and rich countries (models 2a, 2b, 3a and 3b in Table 2), we used the OECD classifications into low-income countries, lower middle-income countries, upper middle-income countries and high-income countries. We computed the mean of the GNP per capita values that separated lower middle income and upper middle income countries for the years 1990 to 1995. This value is almost equal to the median of the sample. Macroeconomic data should always be treated with caution. Therefore, we checked all models for outliers and used the Ramsey Reset Test of functional form. Since it is our aim to discover those aspects that strongly influence economic freedom, we eliminated insignificant variables using the backward elimination technique. When due to the theoretical arguments mentioned above the sample is divided into rich and poor countries, the only area that is significant for corruption in poor and rich countries is Area VII (models 3a and 3b in Table 2). Otherwise, the significant variables are 6 The significant Reset Test in model 1 points to a problem. Since this model is not used later on we do not go into further investigations on the problem. For a discussion of this problem refer to Paldam (1999, p. 13).

P. Graeff, G. Mehlkop / European Journal of Political Economy 19 (2003) 605 620 611 different for rich and poor countries. Only Area IVa is additionally significant in the poor countries sub-sample. Corruption in rich countries is influenced by Areas I, V and VII. If corruption is perceived as a consequence of economic freedom, in rich countries it seems to be inversely associated to the size of government (Area I). Bigger government results in lower levels of corruption. We shall refer to this unexpected result later on. Here, we would like to suggest that this association does not reflect a simple effect of income. The coefficient of Area I does not become insignificant or, at least, does not decrease considerably when GDP per capita is inserted as a control variable in the equations (see Section 4). Area V measures the legal structure of a country. It stresses the aspects of the security of private ownership rights, the risk of contract repudiation by the government and the supportiveness of legal institutions for the principle of the rule of law. The possibility to confiscate private property or to repudiate contracts gives public officials the opportunity to abuse their public position for personal gain. Improving the legal structure leads to less corruption. Therefore, the coefficient of this aspect of economic freedom shows a positive sign. Furthermore, according to transaction cost theory, a weak legal structure (which is used in an arbitrary manner) generates opportunities for corruption as well. If market forces are used to allocate capital instead of political or governmental considerations (Area VII), corruption decreases. This result is valid both for rich and for poor countries. For poor countries Area IVa is significant, too. It measures how easy it is to conduct business in foreign currencies. Notice that the sign of the coefficient of Area IVa is negative, which implies that the freedom to own foreign currencies leads to corruption. For corrupt deals, this aspect of economic freedom reduces transaction costs, especially when the participants of the illegal bargain belong to foreign countries. The patterns influencing corruption seem to be different (except for the freedom of exchange in capital and financial markets) for poor and rich countries. The remaining areas appear to be unimportant in explaining corruption. Area II (The Structure of the Economy/ Use of Markets), Monetary Policy/Price Stability (Area III) and Freedom to Trade with Foreign Countries (Area VI) are never significant, neither for poor nor for rich countries. Summing up, we conclude that the impact of economic freedom on corruption depends (with the exception of the freedom of exchange in capital and financial markets Area VII) on which areas of economic freedom are taken. 7 4. Additional test for stability To test the stability of our results, we also insert control variables that could be regarded as determinants of corruption in a macroeconomic analysis (see Mauro, 1995; Treisman, 7 Since unexpected sign effects could occur (like in the case of Area I), we used a two-tailed tailed test and the usual 5% criterion. The patterns would have differed slightly if variables at the 10% level had been used. Usually, different measures of corruption reveal high inter-correlation (much higher than 0.90). Therefore, the results and the patterns do not change if a different corruption index (Kaufman et al., 1999a, 1999b) is used instead of the CPI. There is one exception: Area V becomes significant in the poor countries sub-sample when the Kaufman Kraay Zoido Lobatón index is used.

612 P. Graeff, G. Mehlkop / European Journal of Political Economy 19 (2003) 605 620 2000; Paldam, 2001). These determinants could be classified as having economical and cultural or social impacts. In poor countries with high levels of illiteracy, many people have little understanding of governmental operations (Rose-Ackerman, 1999). To control for this, we include the primary school enrolment rate. For such people, it is often also not clear as to what they should expect from a legitimate government. In such a situation, corruption is more likely because people suppose that they ought to present gifts as gratitude for favourable decisions (Pasuk and Sungsidh, 1994, Mauss, 1966). This way, corruption is less a matter of bargain, it is more a matter of culture and social exchange. Because countries classified as being highly corrupt under-invest in education (Mauro, 1998) and neglect the creation of human capital, we also include the secondary school enrolment rate. Often, poor countries also have rapid population growth and therefore we take up average population growth. For the state, this creates problems in providing social and health services and in creating an adequate infrastructure. Investment in capital-intensive infrastructure or health facilities often provides lucrative opportunities for corruption (Mauro, 1997). To proxy for this, we include investment as share of GDP. Prevalent corruption should be (negatively) linked to the level of the economic development of a country (reflected by GDP per capita) (Treisman, 2000), its rate of economic growth of GDP and the level of democracy. Finally, we considered the inequality of income distribution proxied by the Gini coefficient as a possible determinant for corruption. High-income inequality may correspond to perceptions of unfair state operations and foster feelings of injustice which could make the incidence of corruption more likely (Smelser, 1971). In order to test the stability of the patterns for poor and rich countries, we inserted these control variables one after another and ran backward regressions. For the group of the poor countries, none of the variables in the equations that were significant before (Area IVa and Area VII) change sign or qualitatively affect the value of their coefficients when the control variables are added to the equations, with one exception. Area II becomes significant and Area VII loses significance if primary school enrolment rates are considered in the model. The pattern of influencing variables of economic freedom on corruption seems to be exceptionally stable for rich countries, too. Although population growth, growth of GDP and primary school enrolment rate are significant in their influence on corruption, Areas I, V and VII remain as the only significant variables of economic freedom. There is only one exception: Area V becomes insignificant when secondary school enrolment rates are added to the model as a control variable. 8 8 It is obvious that the patterns may depend on the criterion to distinguish poor and rich countries. To test the stability of the patterns we changed the composition of the sub-samples. In the World Development Reports of the World Bank, the countries are ranked as rich and poor countries depending on their most recent per capita Gross National Product. To test the validity of our results, we used the classification of values of 1990 (benchmark: US$2450, see World Bank, 1992) and also the classification of 1995 (US$3100, see World Bank, 1997) instead of the mean of the years from 1990 1995 that had been used in the previous calculations. We estimated the equations with backward regression once again and used the same control variables taken up in previous regressions. The sample size differs slightly but the patterns for rich and the poor countries remains stable. In order to save space, the detailed results are not reported here, but are available on request.

P. Graeff, G. Mehlkop / European Journal of Political Economy 19 (2003) 605 620 613 Table 3 Extreme Bound Analysis for poor countries (dependent variable: mean of CPI 1998 2000) Variable Coefficient Significance Adj. R 2 N Additional variables Robust/ fragile Area VIa high: 0.606 3.261 0.231 35 population growth, robust growth of GDP base: 0.581 3.323 0.274 36 low: 0.500 2.348 0.196 29 secondary school enrolment Area VII high: 0.666 3.679 0.272 36 population growth robust base: 0.622 3.560 0.274 36 low: 0.551 2.394 0.120 30 growth of GDP, primary school enrolment The column Coefficient refers to the standardized coefficient. In case the base regression had the highest (lowest) coefficient the cells in the row high ( low ) are empty. In the next step, we use Extreme Bound Analysis (EBA) to test the strength of the pattern after inserting more than one control variable into one equation (see Leamer, 1983, 1985; Levine and Renelt, 1992; De Haan and Siermann, 1998). The idea of the EBA is to select a number of control variables that have been shown to have an impact on corruption and to insert any possible linear combination of up to three control variables additionally to the variable(s) of interest into the equation. The relationship between the areas of economic freedom and corruption is labelled as being robust if, independent of the choice of control variables, the standardised coefficient of the variable of interest remains statistically significant and does not change. According to Sala-i-Martin (1997), the test applied in the EBA is too strong any coefficient changes if enough regressions with different variables are run. For this reason, we decided to take only four control variables into account: growth of population, growth of GDP, primary school enrolment rates and secondary school enrolment rates. These are the Table 4 Extreme bound analysis for rich countries (dependent variable: mean of CPI 1998 2000) Variable Coefficient Significance Adj. R 2 N Additional variables Robust/ fragile Area I high: 0.410 3.953 0.764 31 growth of GDP, robust primary school enrolment base: 0.330 3.957 0.758 39 low: Area V high: 0.370 3.050 0.846 30 population growth, fragile secondary school enrolment base: 0.343 2.776 0.758 39 low: 0.200 1.506 0.761 32 primary school enrolment Area VII high: 0.643 4.482 0.783 31 primary school enrolment, robust secondary school enrolment base: 0.482 4.011 0.758 39 low: See Table 3.

614 P. Graeff, G. Mehlkop / European Journal of Political Economy 19 (2003) 605 620 control variables with the biggest impact on corruption in our sample neither the Gini coefficient, GDP per capita, the investment rate nor the level of democracy has a significant effect at all. 9 The results of the EBA are reported in Tables 3 and 4. Freedom to Use Alternative Currencies (for poor countries), Size of Government (for rich countries) and Freedom of Exchange in Capital and Financial Markets (for rich and poor countries) have been identified as predictors of corruption. These variables survive even the strong Extreme Bound Analysis (see Tables 3 and 4). There is a relationship between the property rights and corruption in rich countries but this relationship is not robust. If the school enrolment rates are inserted into the equation in addition, the effect of the property rights is not far removed from zero. 5. Discussion From a broader point of view, this paper deals with the impact of state regulation on the incidence of corruption. 10 By and large, the Index of Economic Freedom measures forms of regulations in order to describe economic freedom as the absence of regulation. In practice, such regulation means little in many (mainly poor) countries because the state lacks enforcement capacity. Economic freedom may arise from the ineffectiveness of the state to monitor compliance with rules (Bratton, 1989). Putting the meaning of freedom this way, corruption could work as an informal buffer which mediates particularistic and universalistic interests (Smelser, 1971). Up to now, there is no theory that is able to explain the relation between economic freedom and corruption (Chafuen and Guzmán, 2000). Nevertheless, it is obvious that in modern economies many restrictions of economic freedom provide opportunities for corruption. The authority of public officials to grant import or export permits (Shleifer and Vishny, 1993) or to impose import and export duties or to induce taxes are examples. We found the restrictions of capital and financial markets (Area VII) to have the biggest impact on corruption. Especially when political leaders or senior public officials are involved in corruption (Scott, 1972), the absence of financial restraints plays a major role. Economic freedom deals with a country s link to the broader world and this link could be beneficial or obstructive for illegal actors. If money could be hidden abroad or is easily taken across borders, it is much easier to launder money from corrupt 9 Only those variables which have proved to have an influence on the areas were used in our EBA. Since there is no criterion to preselect variables for an EBA by this we limited the calculation effort (Sala-I-Martin, 1997). 10 With this point of view we claim state regulation being causal for the incidence of corruption. But state regulation might result from corruption deals, too. There are only a small number of public officials with the power and the opportunity to create prescriptions or regulations at their corruption partners behest. The vast majority of corrupt public officials use prescriptions or regulations that already exist and construe those in favour of their partners in corruption. Therefore, it is more appropriate to investigate the influence of state regulation on corruption than the opposite direction of causality.

P. Graeff, G. Mehlkop / European Journal of Political Economy 19 (2003) 605 620 615 deals or other criminal acts. According to our findings (Area IVa), this occurs mainly in poor countries. Therefore, it might be somewhat problematic to state that all aspects of economic freedom are deterrents to corruption. When lack of economic freedom increases the transaction costs of illegal bargains, corruption could become less likely. Unfortunately, increasing the transaction costs for illegal actors is usually an obstacle for legal actors as well. When the incidence of corruption is considered for poor and rich countries separately, general differences could be detected which are approximated by the variables used to test the stability of our results. In poor countries, many people have low levels of education. There is often a huge gap between the rich and the poor. The growth of GDP is small or negative while the growth of the population is large. In rich countries, the paradox emerges that countries with larger governments usually have lower corruption. We doubt that Size of Government is a useful element in the Index of Economic Freedom being linked to corruption. Government size is regarded as being an indicator of fiscal regulation (Tanzi, 2000) but it also includes many aspects which could hardly be perceived as being regulatory (e.g. loans of public officials). Furthermore, the concept of government size needs defining in order to separate the amount of public spending which is useful for an economy to prosper from the amount of public spending which does harm to the economy, something which Gwartney et al. (2000) seem to have forgotten in constructing their index. Different authors suggest different separation values (Tanzi and Schuhknecht, 2001; Gwartney et al., 1998). If one is to explain corruption through economic freedom, it seems more appropriate to focus on how public servants work. As Tanzi (1998, p. 10) points out: Rather, the way the state operates and carries out its function is far more important than the size of public sector activity. It is influenced by many factors. A crucial determinant might be the incentive to engage in illegal activities or the ease in conducting illegal business. For poor countries, the ease of winding up affairs severely affects the likelihood of corruption (Area IVa). This is not true for rich countries where corruption is more affected by the legal structure in a country (Area V) and, therefore, by the way the state supports the rule of law. The empirical findings lead to two suggestions: first, that there are regulations which decrease corruption by increasing the transaction costs of corruption. Second, that big governments in rich countries are not equivalent to a high level of corruption. Therefore, one has not only to investigate the relationship between the size of government and corruption but also the legal and institutional capacity of monitoring and sanctioning illegal bargaining. Acknowledgements We thank Erich Weede, Jakob De Haan, Ekkart Zimmermann, the referees and the participants of the SOM workshop on economic freedom in Groningen, November 2001, for their comments and suggestions.

616 P. Graeff, G. Mehlkop / European Journal of Political Economy 19 (2003) 605 620 Appendix A. CPI and GDP p.c. by country Poor countries Rich countries Country CPI GDP p.c. Country CPI GDP p.c. Albania 2.30 708.48 Argentina 3.17 7896.74 Bulgaria 3.23 1508.29 Australia 8.57 19265.02 Cameroon 1.63 654.99 Austria 7.60 28377.67 China 3.33 507.07 Belgium 5.60 26624.29 Colombia 2.77 1975.25 Botswana 6.07 3143.83 Costa Rica 5.37 2563.56 Brazil 4.00 4250.04 Ecuador 2.43 1541.45 Canada 9.20 19125.25 Egypt 3.10 996.13 Chile 7.03 3805.14 El Salvador 3.87 1557.63 Croatia 3.20 3799.64 Guatemala 3.15 1428.54 Czech Republic 4.57 4934.09 Honduras 1.75 703.81 Denmark 9.93 33652.12 India 2.87 338.79 Estonia 5.70 3564.72 Indonesia 1.80 960.09 Finland 9.80 24507.55 Jamaica 3.80 1642.02 France 6.67 26095.07 Jordan 4.57 1482.42 Greece 4.90 10764.81 Kenya 2.20 337.80 Hungary 5.13 4402.23 Lithuania 3.95 2271.24 Ireland 7.70 16531.67 Malawi 4.10 152.76 Israel 6.83 14748.50 Morocco 4.17 1318.59 Italy 4.63 18618.88 Namibia 5.33 2113.00 Japan 6.07 40769.74 Nicaragua 3.05 443.71 Latvia 3.17 3064.13 Nigeria 1.57 260.43 Malaysia 5.07 3897.44 Pakistan 2.45 483.62 Mexico 3.33 3286.30 Paraguay 1.75 1823.65 Netherlands 8.97 25274.87 Peru 4.47 2297.68 New Zealand 9.40 16009.16 Philippines 3.23 1053.18 Norway 9.00 32206.49 Romania 3.07 1405.00 Poland 4.30 2935.26 Senegal 3.40 550.30 Portugal 6.53 10364.71 Tanzania 2.10 158.19 Russia 2.30 2819.53 Thailand 3.13 2525.59 Slovak Republic 3.70 3311.03 Tunisia 5.07 1990.61 Slovenia 5.75 9219.72 Uganda 2.37 281.78 South Africa 5.07 3452.93 Ukraine 2.30 2302.43 South Korea 4.00 9258.03 Zambia 3.47 420.18 Spain 6.57 14067.01 Zimbabwe 3.77 675.53 Sweden 9.43 25904.73 Switzerland 8.80 44266.71 Turkey 3.60 2734.69 United Kingdom 8.67 18412.03 United States 7.60 26373.24 Uruguay 4.35 5489.84 Venezuela 2.53 3542.24

P. Graeff, G. Mehlkop / European Journal of Political Economy 19 (2003) 605 620 617 Appendix B. List of variables Name Variable Source CPI 98 2000 Mean of corruption perception index 1998 2000 Transparency International 2000 Area I Size of government: consumption, transfers, and subsidies 1996/7 Gwartney et al. (2000) (A) General government consumption expenditures as a percent of total consumption (B) Transfers and subsidies as a percent of GDP Area II Structure of the economy and use of markets 1996/7 Gwartney et al. (A) Government enterprises and investment as a (2000) share of the economy (B) Price control: extent to which businesses are free to set their own prices (C) Top marginal tax rate (D) The use of conscription to obtain military personnel Area III Monetary policy and price stability 1996/7 Gwartney et al. (A) Average annual growth rate of money supply (2000) during the last 5 years minus the growth rate of real GDP during the last 10 years (B) Standard deviation of the annual inflation rate during the last 5 years (C) Annual inflation rate during the most recent year Area IVa Freedom to use alternative currencies 1996/7 Gwartney et al. (A) Freedom of citizens to own foreign currency (2000) bank accounts domestically and abroad Note: We excluded Area IVb from the regressions (see footnote 3). Area V Legal structure and property rights 1996/7 Gwartney et al. (A) Legal security of private ownership rights (2000) (B) Viability of contracts (C) Rule of law: legal institutions supportive of the principles of rule of law and access to a nondiscriminatory judiciary Area VI International exchange: freedom to trade with foreigners 1996/7 Gwartney et al. (2000) (A) Taxes on international trade (i) Revenue from taxes on international trade as a percent of exports plus imports (ii) Mean tariff rate (iii) Standard deviation of tariff rates (B) Non-tariff regulatory trade barriers (i) Percent of international trade covered by non-tariff trade restraints (ii) Actual size of trade sector compared to the expected size (continued on next page)

618 P. Graeff, G. Mehlkop / European Journal of Political Economy 19 (2003) 605 620 Appendix B (continued) Name Variable Source Area VII Freedom of exchange in capital and financial markets 1996/7 Gwartney et al. (2000) (A) Ownership of banks: percent of deposits held in privately owned banks (B) Extension of credit: percent of credit extended to private sector (C) Interest rate controls and regulations that lead to negative interest rates (D) Restrictions on the freedom of citizens to engage in capital transactions with foreigners Economic freedom Economic freedom 1996/7, summary rating Gwartney et al. (2000) GDP p.c. Arithmetic mean of gross domestic product per capita 1990 1997 constant market prices 1995 World Development Indicators 1997 CD-ROM Gini Gini-coefficient of distribution of income World Bank (2000, pp. 282 3) Duration of Political System Logarithm of the duration of a political system before 1998. Since the duration of two political systems was shorter than 1 year, to calculate the logarithm of the values their score was fixed to 1. Average annual growth of population percent 1990 1997 Average annual growth rate of gross domestic product percent 1990 1997 Gross domestic investment as percent of gross domestic product, arithmetic mean of 1990 1995 Percentage of age group enrolled in primary education 1990 Percentage of age group enrolled in secondary education 1990 Polity 98 database Population avg. ann. growth 1990 97 World Bank (1998, pp. 194 5) Growth of GDP World Bank 1990 97 (1998, pp. 210 1) Investment 1990 95 World Development Reports (1990 1995) Primary school World Bank enrolment (1993, pp. 294 5) Secondary school World Bank enrolment (1993, pp. 294 5 Level of democracy Level of democracy 1995 Polity 98 database References Ades, A., Di Tella, R., 1997. The new economics of corruption: a survey and some new results. In: Heywood, P. (Ed.), Political Corruption. Blackwell, Oxford, pp. 80 100. Becker, G.S., Becker, G.N., 1997. The Economics of Life. McGraw-Hill, New York. Ben-Porath, Y., 1980. The F-connection: families, friends, and firms and the organization of exchange. Population and Development Review 6, 1 30. Bratton, M., 1989. The politics of government NGO relations in Africa. World Development 17, 569 587. Buchanan, J.M., 1975. The Limits of Liberty, Between Anarchy and Leviathan. The University of Chicago Press, Chicago. Chafuen, A.A., Guzmán, E., 2000. Economic freedom and corruption. In: O Driscoll, G.P., Holmes, K.R., Kirkpatrick, M. (Eds.), Index of Economic Freedom. The Heritage Foundation, Washington, pp. 51 63. De Haan, J., Siermann, C.L.J., 1998. Further evidence on the relationship between economic freedom and economic growth. Public Choice 95, 363 380. De Haan, J., Sturm, J.E., 2000. On the relationship between economic freedom and economic growth. European Journal of Political Economy 16, 215 241.

P. Graeff, G. Mehlkop / European Journal of Political Economy 19 (2003) 605 620 619 Gwartney, J., Lawson, R., Block, W., 1996. Economic Freedom of the World 1975 1995. The Fraser Institute, Vancouver. Gwartney, J., Lawson, R., Holcombe, R., 1998. The scope of government and the wealth of nations. Cato Journal 182, 163 190. Gwartney, J., Lawson, R., Samida, D., 2000. Economic Freedom of the World 2000 Annual Report. The Fraser Institute, Vancouver. Jones, E.L., 1981. The European Miracle. Cambridge Univ. Press, Cambridge, UK. Kaufman, D., Kraay, A., Zoido-Lobatón, P., 1999a. Aggregating governance indicators. Policy Research working paper 2195. The World Bank, Washington, DC. Kaufman, D., Kraay, A., Zoido-Lobatón, P., 1999b. Governance matters. Policy Research working paper 2196. The World Bank, Washington, DC. Lambsdorff, J., 1999. The Transparency International Corruption Perception Index 1999 Framework Document. Transparency International, Berlin. Lancaster, T.D., Montinola, G.R., 2001. Comparative political corruption: issues of operationalization and measurement. Studies in Comparative International Development 36, 3 28. Leamer, E.E., 1983. Let s take the con out of econometrics. American Economic Review 73, 31 43. Leamer, E.E., 1985. Sensitivity analysis would help. American Economic Review 75, 313 398. Levine, R., Renelt, D., 1992. A sensitivity analysis of cross-country growth regressions. American Economic Review 82, 942 963. Lipset, S.M., Lenz, G.S., 2000. Corruption, culture, and markets. In: Harrison, L.E., Huntington, S.P. (Eds.), Culture Matters. Basic Books, New York, pp. 112 125. Mauro, P., 1995. Corruption and growth. Quarterly Journal of Economics 110, 681 712. Mauro, P., 1997. Why Worry About Corruption? International Monetary Fund, Washington, DC. Mauro, P., 1998. Corruption and the composition of government expenditure. Journal of Public Economics 69, 263 279. Mauss, M., 1966. The Gift. Routledge and Kegan Paul, London. Messick, R.E. (Ed.), 1996. World Survey of Economic Freedom 1995 1996. A Freedom House Study. Freedom House and Transaction Publishers, New Brunswick. North, D.C., Thomas, R.P., 1973. The Rise of the Western World: A New Economic History. Cambridge Univ. Press, Cambridge. Olson, M., 1996. Big bills left on the sidewalk. Journal of Economic Perspectives 10, 3 14. O Driscoll, G.P., Holmes, K.R., Kirkpatrick, M. (Eds.), 2000. 2000 Index of Economic Freedom. The Heritage Foundation, Washington. Paldam, M., 2001. Corruption and religion. Adding to the economic model? Kyklos 54, 383 414. Paldam, M., 2002. The big pattern of corruption, economics, culture and the seesaw dynamics. European Journal of Political Economy 18, 215 240. Pasuk, P., Sungsidh, P., 1994. Corruption and Democracy in Thailand. The Political Economy Centre, Faculty of Economics, Bangkok. Rose-Ackerman, S., 1999. Corruption and Government. Cambridge Univ. Press, Cambridge, UK. Scott, J.C., 1972. Comparative Political Corruption. Prentice Hall, Englewood Cliffs. Sala-i-Martin, X., 1997. I just ran four million regressions. Working paper, Department of Economics, Columbia University. Schumpeter, J.A., 1942. Capitalism, Socialism, and Democracy. Harper and Brothers, New York. Shleifer, A., Vishny, R.W., 1993. Corruption. Quarterly Journal of Economics 108, 599 617. Smelser, N.J., 1971. Stability, instability and the analysis of political corruption. In: Barber, B., Inkeles, A. (Eds.), Stability and Social Change. Little Brown, Boston, pp. 7 29. Smith, A., 1976. An Inquiry into the Nature and Causes of the Wealth of Nations. Oxford Univ. Press, Oxford. Tanzi, V., 1998. Corruption around the world: causes, consequences, scope, and cures. Working paper number 98/ 63. International Monetary Fund, Washington, DC. Tanzi, V., 2000. Policies, Institutions, and the Dark Side of Economics. Edward Elgar, Cheltenham. Tanzi, V., Schuhknecht, L., 2001. Public Spending in the 20th Century. Cambridge Univ. Press, New York. Treisman, D., 2000. The causes of corruption: a cross-national study. Journal of Public Economics 76, 399 457.

620 P. Graeff, G. Mehlkop / European Journal of Political Economy 19 (2003) 605 620 Weede, E., 2000. Asien und der Westen. Politische und kulturelle Determinanten der wirtschaftlichen Entwicklung. Nomos, Baden-Baden. World Bank, 1990. World Development Report 1990. Oxford Univ. Press, New York. World Bank, 1991. World Development Report 1991. Oxford Univ. Press, New York. World Bank, 1992. World Development Report 1992. Oxford Univ. Press, New York. World Bank, 1993. World Development Report 1993. Oxford Univ. Press, New York. World Bank, 1994. World Development Report 1994. Oxford Univ. Press, New York. World Bank, 1995. World Development Report 1995. Oxford Univ. Press, New York. World Bank, 1997. World Development Report 1997. Oxford Univ. Press, New York. World Bank, 1998. World Development Report 1998/1999. Oxford Univ. Press, New York. World Bank, 2000. World Development Report 2000/2001. Oxford Univ. Press, New York.