Corruption and Economic Growth

Corruption and Economic Growth by Min Jung Kim 1 Abstract This study investigates the direct and indirect impact of corruption on economic growth. Recent empirical studies have examined that human capital, investment, and openness serve as the transmission channels in which corruption hampers economic growth. These results, however, are not robust because of endogeneity problems. It is necessary to control not only the time-invariant country heterogeneity but also the interactions between corruption, transmission channels, and economic growth. In order to verify the comprehensive relationships between corruption, growth, and transmission channels, we need to take more elaborate approach that alleviates potential endogeneity problems. This study estimates with panel data, which contains 70 countries economic indicators from period 1980 to 2009. The results show that there are three important transmission channels through which corruption affects economic growth. More specifically, corruption is found to have discouraged investment, accumulation of human capital and trade, thereby lowering economic growth. These findings are robust to various specifications and estimation methods, including OLS estimation, fixed effect estimation, and system GMM estimation. 1. Introduction Bureaucratic corruption has drawn economists attention in terms of its magnitude and consequences for corruption. Specifically, corruption has been found to affect economic growth in several economic literatures. However, there are two contradictory arguments about the effect of corruption on economic growth. One argues that corruption is beneficial to growth, and the other claims that corruption hinders economic growth. Leff(1964), for example, explains that corruption acts as a piece rate for public officials and allows them to serve public service more efficiently. Moreover, entrepreneurs might avoid bureaucratic delay concerning his business by giving a bribe. Therefore, Leff argues that a certain extent of corruption promotes economic growth. However, most of the recent studies have concluded that corruption slows down economic growth. A seminal study, Mauro(1995) verified that corruption impedes economic growth using 58 countries data from 1960 to 1985. Moreover, Mauro found that corruption discourages the investment to GDP ratio. This result suggests that investment is an important channel through which corruption aggravates economic growth. Recently, numerous studies have 1 Department of Economics, Seoul National University. (e-mail : dearmj@snu.ac.kr) 1

2 focused on not only the relation between corruption and growth, but also the transmission effect through which corruption affects economic growth. For example, Pak(2001) verified that corruption hampers accumulation of human capital, investment and political stability, thereby decreasing GDP growth rate. Pellegrini and Gerlagh(2004) investigated that trade policy, human capital, investment, and political stability are important channels through which corruption lowers economic growth. To compare the relative importance of transmission effects through which economic growth is impaired by corruption, analyzing the relationship between corruption, economic growth, and transmission channels is necessary. However, previous studies which verified the transmission effect have a significant drawback that they did not consider the interactions between corruption, economic growth, and other economic variables or so-called growth engines. In other words, either corruption or economic growth engines might be affected by economic growth. Furthermore, growth engines might interact with each other. Therefore, empirical findings which do not take interaction effects into consideration lose robustness. In order to analyze the comprehensive relationship between corruption, economic growth, and growth engines, this study attempts to alleviate potential endogeneity problems through using system GMM estimates. On the basis of an elaborate approach to control endogeneity problems, a robust and significant result can be derived in the end. There are two main purposes for this research. First, the effect of corruption on economic growth would be examined. By controlling for other economic variables in the growth model, a regression result represents a direct effect of corruption on economic growth. Second, transmission effects through which economic growth is deteriorated by corruption would be analyzed. Three growth engines, human capital, investment, and trade, serve as transmission channels. In order to verify the second argument, we need to analyze two relationships. One is the relationship between growth engines and economic growth. If growth engines affect economic growth as transmission channels through which corruption impedes economic growth, then growth engines should have a positive and significant effect on economic growth. Since growth engines are involved in the growth model as control variables, we can identify the significant outcomes in the first argument. The other is the relationship between corruption and growth engines. If corruption hinders economic growth, then corruption should have a negative and significant effect on growth engines. On the basis of two associations, we can verify an indirect effect of corruption on economic growth. To sum up, both the direct and the indirect effects of corruption on economic growth can be indentified in this paper. In addition, it is expected that the direct effect is much smaller than the indirect effect if growth engines are adequate to account for the transmission channels. The paper is organized as follows. The next section introduces the previous studies about the relationship between corruption and economic growth. Section 3 describes model and data. Section 4 presents empirical evidence on the association between corruption, economic growth, and transmission channels. Specifically, sub-

section 4.1 examines the effect of corruption on economic growth. Sub-section 4.2 shows the relationship between corruption and growth engines or so-called transmission channels. Sub-section 4.3 demonstrates the relative importance of the direct effect and the indirect effect of corruption on economic growth. The last section contains concluding remarks. 2. Literature Review Bureaucratic corruption is common across countries, although the scale of corruption is substantially different according to the characteristics of each country. So far, a large number of economists have examined the cause of corruption as well as socio-economic effects of corruption. Furthermore, they have recently been interested in the study of transmission channels through which corruption affects economic growth. There are two competing claims regarding the link between corruption and growth. One argues that corruption is beneficial to growth, and the other denies positive effects of corruption on growth. Leff(1964), for example, focused on positive aspects of corruption because of the two following mechanisms. First, corruption acts as a piece rate for public officials and allow them to serve public service more efficiently. Second, entrepreneurs might avoid bureaucratic delay concerning his business by giving a bribe. Therefore, Leff explained that to some extent corruption might raise economic growth. Most of the studies which deal with the relationship between corruption and economic growth have argued that economic growth rate could drop because of corruption (Gould and Amaro-Reyes, 1983; United Nations, 1989; Klitgaard, 1991; and Shleifer and Vishny, 1993). Mauro (1995), which is widely regarded as a seminal study, engaged in an empirical analysis of the relationship between corruption and growth. Mauro verified that corruption hinders economic growth based on 58 countries cross-sectional data period from 1960 to 1985. To be specific, he found that a one-standard-deviation improvement in corruption is associated with a 0.8 percentage point increase in the annual growth rate of GDP per capita. Moreover, Mauro suggests an additional finding that corruption has a significant negative effect on investment to GDP ratio. It implies that investment is an important channel through which corruption negatively affects economic growth. Tanzi and Davoodi (1997), meanwhile, focus on the association between corruption and public investment and demonstrate that corruption impedes economic growth. When the commissions paid by enterprises to public officials to win investment contracts are tied to the projects costs, an incentive may be created for larger projects rather than established projects for maintaining or operating infrastructure. In other words, higher corruption might be associated with higher public investment, even though such government expenditure cannot contribute to improving infrastructure or productivity. Consequently, corruption slows down economic growth through increasing unnecessary public investment. Gupta et al.(2001) in particular provide evidence that corruption is related to a high level of government spending, especially military expenditure. Since military spending has less effect on 3

4 growth, it can be interpreted that corruption has negative effect on growth through large amounts of public spending such as defense spending. In addition, Ehrlich and Lui (1999) examined the effects of government intervention in private economic activity, political regimes, and corruption on economic growth. With 152 counties panel data from 1960 to 1992, Ehrlich and Lui verified that the higher level of government intervention as well as corruption, the lower economic growth rate. Moreover, both corruption and government size have less impact on growth rate in the more developed countries. In other words, the relationship between government, corruption and growth is nonlinear according to the stage of economic development. Synthesizing the economic analyzes so far, it seems that there is general consensus that corruption negatively affects the overall economy (Jain, 2001). There are further recent empirical researches regarding transmission channels through which corruption affects economic growth. Pak (2001), for example, provides quantitative estimates of the impact of corruption on the growth and importance of the transmission channels with panel data containing 5-year averages from 1960 to 1985. Results suggest that corruption reduces the level of human capital and the share of private investment and causes political instability, thus leading to a slowdown in economic growth. The most important channel through which corruption affects economic growth is political instability, which accounts for about 53% of the total effect. The relative importance of human capital and private investment as transmission channels is about 14.8% and 21.4%, respectively. In other words, the direct effect of corruption on economic growth amounts 11.8% and the rest is attributed to political instability and decreasing level of the human capital and private investment. Pellegrini and Gerlagh (2004) provide another empirical evidence for the relationship between corruption, economic growth, and transmission channels on the basis of similar methodology introduced in Pak s study. Pellegrini and Gerlagh take trade policy into consideration to account for major transmission channel. Based on empirical results, they suggest that the most important factors which are likely to be distorted by the presence of corruption are investment and trade openness. Meon and Sekkat (2005), meanwhile, demonstrate that lower quality of governance leads to larger negative impact of corruption on investment, and thus impedes economic growth. Meon and Sekkat use a weak rule of law, an inefficient government and political violence as indicators of governance quality and verified that corruption slows growth down. Furthermore, these negative impacts are far larger in countries which suffer from a lower level of governance, even when one controls for investment. Therefore, Meon and Sekkat conclude that corruption not only impacts growth through reducing accumulation of capital but also through other channels. In summary, a number of economists have investigated the negative relationship between corruption and economic growth, and pointed out that reduced accumulation of both human and physical capital, trade policies, political instability and lower quality of institution are major transmission channels.

3. Model and Data Description The main purpose of this study is analyzing the comprehensive relationships between corruption, economic growth, and transmission channels through which corruption hinders growth. There are two arguments regarding those associations. First, the hypothesis that corruption impedes economic growth even when one controls for economic engines is investigated on the basis of economic growth model. Second, the hypothesis that corruption affects economic engines negatively is examined on the basis of the previous researches. As verifying these two hypotheses, both direct and indirect effects of corruption on economic growth are estimated. Furthermore, the relative importance of transmission channels through which corruption affects growth is figured out. In neoclassical growth models, such as Solow(1956), Cass(1965), and Koopmans(1965), a country s per capita growth rate tends to be inversely related to its starting level of income per person. In other words, poor countries tend to grow faster than rich countries. Meanwhile, Romer(1990), the pioneer in the field of endogenous economic growth, presents human capital as a key input, which generates new products that underlie technological progress. Therefore, countries with larger initial stocks of human capital achieve a more rapid growth rate. Barro(1991) examined an empirical analysis related to the convergence hypothesis of neoclassical growth model considering endogenous growth model; that not only a county s initial income but also initial stocks of human capital is included in Barro s model. The empirical results in his paper, consequently, conclude that a poor country tends to grow faster than a rich country, but only for a given quantity of human capital. Furthermore, modern growth theory suggests that institutions are important determinants of income growth in an economy. Acemogu et al. (2001, 2002), for example, emphasized the importance of institutions for long-term economic growth, through reversal of fortune in former colonized countries. Specifically, in places where Europeans faced high mortality rates, they were more likely to set up extractive institutions rather than progressive institutions for long-term economic growth. Since institutions persisted to the present, differences in income per capita in each country are attributed to differences in institutions which is either extractive institutions or progressive institutions established by colonizers. Considering above discussions about established economic growth model so far, we need to set up a model for the relationship between corruption and economic growth that take into considerations some factors as follows. First, it is necessary to include some control variables in growth equation, such as human capital, physical capital, policies, and institutions which are emphasized in previous studies of economic growth. In our framework for investigating the role of corruption in economic growth, we introduce corruption as an indicator of institutional failure into a variant of the expanded neoclassical growth model of Barro(1991), estimated by many researchers (Barro 1991; Caselli et al. 1996; Gyimah-Brempong and Traynor 1999; Gyimah-Brempong 2002; Levine and Renelt 1992; Mankiw et al. 1992; Sachs and Warner 1997; and Gyimah-Brempong and Munoz de Camacho 2006). Second, since explanatory variables proposed in economic growth theory affect growth for a 5

long time, it is desirable to employ long-term data for estimation. In our study, therefore, the data contains 70 countries panel data from period 1980 to 2009. To reduce short run fluctuations and concentrate on the long run relationships between corruption, economic growth, and other variables, we worked with 5-year averages, that is 6 periods; 1980-1984, 1985-1989, 1990-1994, 1995-2000, 2001-2004, 2005-2009 (Deininger and Squire, 1996; Li et al., 2000; Paldam, 2002; Mendez and Sepulveda, 2006). Associations between corruption and economic growth as well as corruption and transmission channels are examined by Eq.(1) and Eq.(2), respectively...(1)..(2) In equation (1), the dependent variable Growth denotes the growth rate of real GDP per capita. In addition, the key independent variable Corruption denotes the measure of the corruption level obtained from the International Country Risk Guide (ICRG) index. The ICRG index of corruption is published annually from Political Risk Services Inc., a private firm. The ICRG corruption index is intended to assess the degree of corruption prevailing in a certain country and is based on a survey among foreign investors. Corruption index indicates whether high government officials are likely to demand special payments and illegal payments are generally expected throughout lower levels of government in the form of bribes connected with import and export licenses, exchange controls, tax assessment, police protection, or loans. All indices are on a scale from 0(worst, most corrupt) to 6(best, least corrupt). 2 Meanwhile, a quadratic term of corruption is introduced to analyze the marginal effect of corruption on dependent variables. Engines of economic growth, T, which are major transmission channels as well, are composed of human capital, investment growth rate to GDP ratio, and trade volume to GDP ratio. The secondary school enrollment ratio is used for the proxy of the human capital stock. Since it takes time for human capital to affect income growth, it might be rational to consider one period lagged value of human capital as an economic growth engine. In this regard, the transmission effect of human capital on the association between corruption and economic growth while controlling for the impact of initial value of human capital in Barro s growth model can be investigated. 3 Meanwhile, regarding the effect of trade policies on economic growth, 2 See Knack and Keefer(1995), for further details. 3 In other words, human capital is instructed for an engine of economic growth, T, as well as a 6

we introduce the more real variable, which is actual exports of manufactured goods plus actual imports of goods and services to GDP ratio rather than openness index itself. In economic literature, trade share in GDP is significantly and positively correlated with growth (Frankel and Romer, 1999; Yanikkaya, 2003; Moskalyk, 2008). In the context of economic growth, however, natural resource abundance might contribute to exports, though it could damage to long-run economic growth (Sachs and Warner, 1995; Leite and Weidmann, 1999). Therefore, we exclude exports in natural resource and focus on exports in manufactured goods. 4 Control variables, X in equation (1), are composed of initial level of real GDP per capita, initial stock of human capital, and the average rate of population growth, which are introduced in the standard growth model of Barro(1991). More specifically, the secondary school enrollment ratio in 1980 is used for the proxy of the initial stock of human capital. 5 Democratic accountability is included to check the robustness of estimations. Measure of democratic accountability level comes from International Country Risk Guide (ICRG) index. The ICRG index of democracy is measured by the degree of government response to its people. The less responsive it is, the more likely it is that the government will fall, peacefully in a democratic society, but possibly violently in a non-democratic one. All indices are on a scale from 0(worst, least democratic accountability) to 6(best, most democratic accountability). Control variables, Z in equation (2), are composed of the level of real GDP per capita in PPP$, the average rate of population growth, and institution index. Institution index which comes from the Polity IV dataset measures the level of constraint on the executives who are social elites. This measure ranges from 1 to 7. The higher institution index, the greater the constraints imposed on politicians and politically powerful elites. 4. Empirical Analysis 4.1. Effect of Corruption on Economic Growth This section empirically analyzes the relationship between corruption, economic growth and other economic variables, namely growth engines. The paper makes an attempt to present various specifications as economic growth engines, and institutional factor are included sequentially from Table 1 to Table 3. More specifically, Table 1 represents a basic growth model as Barro(1991) examined. Table 2 introduces two growth engines, investment and trade, in addition to Table 1. Table 3 adds an institutional factor, democratic accountability, to Table 2. Moreover, control variable, X. 4 Fuel export, ores and metal export, agricultural export, and food export are excluded as natural resource. 5 In a fixed effect model which considers a time-invariant heterogeneity across individuals and provides a consistent estimate, time constant variables such as initial value are removed. Therefore, one period lagged value is adopted for initial value in fixed effect estimation. 7

three different estimates, OLS estimates, fixed effect estimates, and system GMM estimates, are presented, and thus it would provide robust results. Controlling for the other determinants of growth, corruption is negatively associated with economic growth at the 5 % level of significance in all models. 6 In addition, a quadratic term of corruption is positively related with economic growth. It implies that corruption hinders growth rate of real GDP per capita, but the marginal effect of corruption on economic growth is diminishing. Meanwhile, both Table 1 and Table 2 show that lagged value of human capital affects economic growth positively in fixed effect estimates. The positive relationship between lagged value of human capital and GDP growth rate explains that the larger initial value of human capital, the more growth in economy as Barrow argued. Table 2 and Table 3 report the other economic growth engines, which are investment and trade. It is verified that investment has significant positive effect on the economic growth rate in system GMM estimates and fixed effect estimates, as well as OLS estimates. More specifically, 1% increase in investment growth rate positively contributes to economic growth rate by about 0.17% point. On the other hand, the ratio of trade to GDP has a significant and positive relationship with economic growth in both fixed effect model and system GMM model. In short, open economy vitalizes the national economy and thus promotes economic growth. The results, reported in the 5 th and 6 th row of Table 3, indicate that both initial level of GDP per capita and growth rate of population are negatively and significantly associated with economic growth. Controlling for human capital as well as other determinants of growth, the negative relationships between the initial level of economy and current economic growth can be a supportive evidence for the convergence hypothesis proposed by neoclassic school. Meanwhile, democratic accountability, which might affect economy as an institutional factor, has a positive impact on economic growth at 5 % level of significance in fixed effect estimates. 6 Corruption index from ICRG scales from 0, which means the most corrupt to 6, which means the least corrupt. Therefore, the higher scale of corruption index implies the lower level of corruption. In other words, a positive relationship between corruption and economic growth in regression results indicates that the lower level of corruption encourages economic growth. 8

Pooled OLS Fixed Effect System GMM Lagged GDP growth 0.457(11.39)*** 0.212 (4.47)*** 0.268 (1.93)* Corruption 0.007 (2.42)** 0.008 (2.18)** 0.023 (2.14)** Corruption2-0.001 (2.17)** -0.002 (2.82)*** -0.003 (1.88)* Lagged GDP level -0.004 (3.51)*** -0.037 (8.27)*** -0.003 (0.66) Population -0.653 (5.78)*** -0.985 (5.49)*** -1.006 (1.99)* Lagged human capital 0.006 (0.83) 0.022 (2.92)*** 0.013 (0.66) No. of obs. 394 402 402 No. of groups 70 70 R2 0.3915 0.2804 F-test [0.000]*** [0.000]*** Wald test [0.000]*** Hansen test [0.191] AR(2) test [0.732] Tab. 1 : The Effect of Corruption on Economic Growth (Model 1) Note : Dependent variable is growth rate of real GDP per capita (1980-2009, 5 year Average). t statistics are in parentheses and p-values are in brackets. The superscripts *, **, and *** following the t statistics and p-values represent a 10%, 5%, and less than1% significant level, respectively. Pooled OLS Fixed Effect System GMM Lagged GDP growth 0.488 (13.1)*** 0.294 (6.59)*** 0.398 (3.80)*** Corruption 0.007 (2.39)** 0.009 (2.66)*** 0.020 (2.13)** Corruption2-0.001 (2.03)** -0.001 (2.63)*** -0.002 (1.94)* Lagged GDP level -0.004 (4.03)** -0.034 (7.63)*** -0.005 (2.05)** Population -0.715 (7.10)** -0.947 (5.30)*** -0.844 (3.54)*** Lagged human capital 0.005 (0.83) 0.014 (2.07)** 0.002 (0.20) Investment 0.203 (8.02)*** 0.179 (7.73)*** 0.168 (2.60)** Trade 0.003 (1.48) 0.015 (2.15)** 0.012 (2.06)** No. of obs. 376 384 384 No. of groups 70 70 R2 0.5211 0.3987 F-test [0.000]*** [0.000]*** Wald test [0.000]*** Hansen test [0.396] AR(2) test [0.776] Tab. 2 : The Effect of Corruption on Economic Growth (Model 2) Note : Dependent variable is growth rate of real GDP per capita (1980-2009, 5 year Average). t statistics are in parentheses and p-values are in brackets. The superscripts *, **, and *** following the t statistics and p-values represent a 10%, 5%, and less than1% 9

significant level, respectively. Pooled OLS Fixed Effect System GMM Lagged GDP growth 0.487 (13.05)*** 0.279 (6.23)*** 0.409 (3.81)*** Corruption 0.007 (2.28)** 0.008 (2.32)** 0.021 (2.21)** Corruption2-0.001 (1.99)** -0.001 (2.53)** -0.003 (1.89)* Lagged GDP level -0.004 (4.03)*** -0.035 (7.75)*** -0.005 (2.25)** Population -0.702 (6.65)*** -0.899 (5.03)*** -0.877 (3.42)*** Lagged human capital 0.005 (0.72) 0.011 (1.54) 0.004 (0.51) Investment 0.201 (7.87)*** 0.173 (7.47)*** 0.168 (2.74)*** Trade 0.004 (1.51) 0.014 (2.01)** 0.012 (2.15)** Democracy 0.000 (0.40) 0.002 (2.37)** -0.001 (0.67) No. of obs. 376 384 384 No. of groups 70 70 R2 0.5200 0.4096 F-test [0.000]*** [0.000]*** Wald test [0.000]*** Hansen test [0.441] AR(2) test [0.800] Tab. 3 : The Effect of Corruption on Economic Growth (Model 3) Note : Dependent variable is growth rate of real GDP per capita (1980-2009, 5 year Average). t statistics are in parentheses and p-values are in brackets. The superscripts *, **, and *** following the t statistics and p-values represent a 10%, 5%, and less than1% significant level, respectively. 4.2. Effect of Corruption on Growth Engines As government officials pursue their vested rights or strive to secure higher positions, they have an incentive to compete over the privilege through corruption. Therefore, bureaucratic corruption is regarded as a rent seeking activity, which causes social inefficiency and hinders economic growth in the long run. In the context of rent seeking, corruption might distort the allocation of human capital or investment and the direction of trade policy, and finally slow down economic growth. Ehrlich and Lui(1999) argue that investment in political capital, so-called rent seeking, consumes economic resources that could otherwise be used for production or investment in human capital. A recent empirical study finds that corruption lowers investment, accumulation of human capital, or trade openness (Mauro, 1995; Pak, 2001; Dollar and Kraay, 2003; Gyimah-Brempong and Munoz de Camacho, 2006). This section empirically analyzes the indirect effect of corruption on economic growth. More specifically, it is verified that corruption has negative effects on human capital, investment, and trade, which are economic growth engines as well, 10

and hence impedes economic growth. Tables 4, 5 and 6 present empirical results of the association between corruption and human capital, investment, and trade, respectively. Every model is analyzed on the basis of OLS estimates, fixed effect estimates, and system GMM estimates. The results reported on the second row of Table 4 show that corruption is negatively and significantly associated with human capital. In addition, a quadratic term of corruption is positively related to human capital. These results imply that corruption hampers accumulation of human capital, but the marginal effect of corruption on human capital is diminishing. Meanwhile, there is a strong positive relationship between the level of GDP per capita and human capital at 1% of significance. Therefore, increase in human capital is not only a determinant of economic growth as an economic growth engine, but also a consequence of economic development. Institution index, in addition, has positive effect on human capital in system GMM estimates and fixed effect estimates, as well as OLS estimates. Specifically, accumulation of human capital increases as more constraints on executives are imposed and hence the property rights for civilians are effectively protected. Trade, on the other hand, is positively associated with human capital in OLS estimates, while there is no significant relationship between trade and human capital in both fixed effect estimates and system GMM estimates. Therefore, it implies that trade has no significant effect on human capital as we controlled for endogenous problems caused by unobservable time-invariant heterogeneity or reverse causality in which human capital affects trade. Pooled OLS Fixed Effect System GMM Corruption -0.011 (0.39) 0.066 (2.02)** 0.528 (3.07)*** Corruption2 0.002 (0.57) -0.012 (2.61)*** -0.080 (3.30)*** GDP level 0.190 (15.37)*** 0.202 (5.38)*** 0.396 (3.39)*** Population -6.851 (6.96)*** -9.115 (6.12)*** 1.885 (0.37) Investment 0.054 (0.22) 0.066 (0.33) -0.673 (1.12) Trade 0.086 (3.17)*** 0.103 (1.62) -0.120 (0.76) Institution index 0.017 (3.08)*** 0.023 (3.51)*** 0.087 (2.87)*** No. of obs. 347 347 347 No. of groups 71 71 R2 0.7691 0.4462 F-test [0.000]*** [0.000]*** Wald test [0.000]*** Hansen test [0.805] AR(2) test [0.105] Tab. 4 : The Effect of Corruption on Growth Engine : Human capital Note : Dependent variable is human capital (1980-2009, 5 year average). t statistics are in parentheses and p-values are in brackets. The superscripts *, **, and *** following the t statistics and p-values represent a 10%, 5%, and less than1% significant level, respectively. 11

Table 5 shows a negative relationship between corruption and investment growth rate in system GMM estimates and fixed effect estimates, as well as OLS estimates. It is verified that a quadratic form of corruption positively affects investment. Therefore, we can find that corruption lowers investment growth rate, but the marginal effect of corruption on investment is diminishing. The result, reported on the eighth row of Table 5, indicates that institution index is positively and significantly related to investment growth rate except for the system GMM estimates. Therefore, there is weak evidence that well-developed institution induces growth of investment when we control for endogenous problems. Pooled OLS Fixed Effect System GMM Corruption 0.018 (2.79)*** 0.019 (1.95)* 0.067 (3.04)*** Corruption2-0.002 (2.50)** -0.002 (1.62) -0.008 (2.81)*** GDP level -0.012 (3.36)*** -0.003 (0.23) -0.017 (1.55) Population 0.135 (0.59) 0.941 (1.97)** -0.631 (1.06) Human capital 0.003 (0.22) 0.006 (0.33) 0.009 (0.37) Trade 0.001 (0.23) 0.027 (1.37) 0.014 (0.92) Institution index 0.003 (2.87)*** 0.003 (1.74)* 0.000 (0.04) No. of obs. 347 347 347 No. of groups 71 71 R2 0.0952 0.0643 F-test [0.000]*** [0.011]** Wald test [0.016]** Hansen test [0.319] AR(2) test [0.658] Tab. 5 : The Effect of Corruption on Growth Engine : Investment Note : Dependent variable is investment to GDP ratio (1980-2009, 5 year average). t statistics are in parentheses and p-values are in brackets. The superscripts *, **, and *** following the t statistics and p-values represent a 10%, 5%, and less than1% significant level, respectively. A negative association between corruption and trade is investigated and the result is presented in Table 6. In addition, a quadratic form of corruption is positively related to trade. These results imply that corruption discourages trade, but the marginal effect of corruption on trade is diminishing. From Table 4 to Table 6, it is verified that corruption has negative effect on human capital, investment growth rate, and trade, respectively. An implication of these results is that corruption is likely to lower growth engines, and hence indirectly impedes economic growth in the long run. Human capital, on the other hand, has a positive effect on trade in OLS estimates, while there is no significant relationship between human capital and trade in both fixed effect estimates and system GMM estimates. Similar to the result of 12

Table 4, the significance of the relationship between trade and human capital is lost as we controlled for endogenous problems caused by unobservable time-invariant heterogeneity or reverse causality in which trade affects human capital. There is a positive link between institution index and trade in both OLS estimates and fixed effect estimates, whereas system GMM estimates is not. Therefore, there is weak evidence that well-developed institution encourage trade when we controlled for endogenous problems. Pooled OLS Fixed Effect System GMM Corruption 0.074 (1.27) 0.041 (1.31) 0.162 (1.75)* Corruption2-0.011 (1.51) -0.011 (2.49)** -0.033 (2.49)** GDP level 0.007 (0.23) 0.184 (5.13)*** 0.083 (0.84) Population -0.334 (0.16) -1.814 (1.20) -3.226 (0.93) Human capital 0.336 (3.17)*** 0.093 (1.62) 0.144 (0.82) Investment 0.111 (0.23) 0.262 (1.37) 0.054 (0.16) Institution index -0.019 (1.77)* 0.013 (2.12)** -0.001 (0.05) No. of obs. 347 347 347 No. of groups 71 71 R2 0.0876 0.3409 F-test [0.000]*** [0.000]*** Wald test [0.005]*** Hansen test [0.299] AR(2) test [0.549] Tab. 6 : The Effect of Corruption on Growth Engine : Trade Note : Dependent variable is trade to GDP ratio (1980-2009, 5 year average). t statistics are in parentheses and p-values are in brackets. The superscripts *, **, and *** following the t statistics and p-values represent a 10%, 5%, and less than1% significant level, respectively. 4.3. Transmission Effect of Corruption on Economic Growth By analyzing the relationship between corruption and economic growth as well as corruption and economic growth engines, both direct effect and indirect effect of corruption on economic growth can be identified. On the basis of the outcomes so far, the magnitude of the direct effect and the indirect effect can be calculated as follows; 13

(3) A comprehensive relationship between corruption, growth engines, and economic growth can be analyzed by taken each growth engine into consideration. Based upon the outcome of estimation which is involved economic growth engines (i.e. human capital, investment, and trade) altogether, we can examine relative magnitude of indirect effect. More specifically, if GDP growth rate depends on economic growth engines, which in turn depends on the level of corruption, then the indirect effect of corruption on GDP growth rate can be expressed as part C in Eq.(3). Table 3 shows that all growth engines are considered and the democratic accountability index is included as well. Therefore, we can quantify both the direct effect and the indirect effect on the basis of those outcomes. 7 To quantify the role of economic growth engines as the transmission effects, part C in Eq(3) is used. For example, multiply the coefficient of corruption in Table 4 by that of human capital in Table 3. The outcome indicates the magnitude of the human capital s channel effect through which corruption hinders economic growth. In other words, an increase of 1% in corruption index decreases 0.27% point of GDP growth rate due to the human capital channel. This way, we can quantify the magnitudes of the other transmission effects, namely investment channel effect and trade channel effect. Specifically, 1% increase in the level of corruption hampers the rate of investment growth, and thus declines 0.02% point of GDP growth rate in the long term. In addition, about 0.11% drops of the GDP growth rate in the corruption and growth linkage is due to the trade channel. In sum up, it is found that the total magnitude of the transmission effects is about 0.4 %. In other words, an increase of 1% in the level of corruption reduces 0.4% point of GDP growth rate indirectly. On the other hand, the results, shown on the third and fourth rows of Table 3, explain the direct effect of corruption on economic growth. In short, an increase of 1% in corruption index directly slows down 0.03% point of the growth rate of GDP per capita. 8 It is a direct effect of corruption on economic growth after controlling 7 The results, reported on the fourth column of Table 3, are more robust than others in terms of the model specification and the methodology. Therefore, it is desirable to quantify the direct effect and the indirect effect based upon the results on the fourth column of Table 3. 8 Growth rate of GDP per capita, as a dependent variable of Table 3, is expressed by log formation, whereas corruption index is expressed by scale. Therefore, we can calculate growth elasticity of corruption by multiplying the estimated coefficient by the average level of corruption. In addition, it is desirable to consider the effect of the quadratic term of 14

transmission effects through which corruption impedes growth engines (i.e. human capital, investment, trade) and hence decreases GDP growth rate. Consequently, the total effect of corruption on economic growth is about 0.43 %, since it is decomposed into the direct effect and the indirect effect, which are 0.03% and 0.4% respectively. It implies that the majority of the negative impact of corruption on GDP growth is attributed to the weak economic growth engines. In other words, corruption itself impedes GDP growth rate slightly. Indirect effect of corruption, however, considerably lowers economic growth, because it possible undermines the foundation of entire economy. A comprehensive result is demonstrated in Figure 1 as following. Indirect Effect : 0.40% Economic Engines 0.27% Human Capital Corruption 0.02% Investment Economic Growth 0.11% Trade Direct Effect : 0.03% Total Effect : 0.43% Fig. 1 : Direct effect and indirect effect of corruption on economic growth 5. Conclusions This study closely scrutinizes the relationship between corruption and economic growth using panel data which contains 70 countries from period 1980 to 2009. corruption to quantify the effect of corruption on economic growth. 15

Since we analyzed a comprehensive relationship between corruption, transmission channels, and economic growth, there are considerable potential endogeneity problems. In this study, unobserved time-invariant country heterogeneity is taken into consideration on the basis of fixed effect estimates. In addition, the interactions between corruption, transmission channels, and economic growth are controlled on the basis of system GMM estimates. By taking such elaborate approaches which alleviate potential endogeneity problems, robust results are derived. There are two main findings for this research. First, corruption slows economic growth down directly. More specifically, an increase of 1% in the level of corruption directly decreases GDP growth rate by 0.03% point. This direct effect implies net effect in which corruption affects GDP growth rate after controlling various determinant variables on growth model. Second, corruption indirectly hinders economic growth through transmission channels. Three growth engines, human capital, investment, and trade, serve as the transmission channels. The results show that corruption discourages investment, accumulation of human capital and trade, thereby hampering economic growth. More specifically, 1% improvement in the level of corruption is associated with a 0.27% point increase in the growth rate of GDP per capita through encouraging accumulation of human capital. Investment channel effect and trade channel effect which explain the indirect effect of corruption on economic growth are 0.02% point and 0.11% point, respectively. Therefore, total indirect effect in which corruption slows down economic growth through transmission channels is about 0.4% point. To sum up, we can quantify total effect in which an increase of 1% in corruption index affects GDP growth rate is about 0.43% point. This total effect is decomposed into direct effect and indirect effect and the magnitude of each effect is about 0.03% and 0.4%, respectively. It implies that the majority of the negative impact of corruption on GDP growth is attributed to the shrink of the economic growth engines. In other words, corruption itself impedes GDP growth rate slightly. Indirect effect of corruption, however, considerably lowers economic growth, because it possible undermines the foundation of entire economy. Appendix Variable Description Source GDP growth rate Growth rate of real GDP per capita (log form) WDI Corruption Corruption index (scale : 0~6) ICRG Population growth rate Growth rate of population (log form) WDI 16

Human capital Secondary school enrolment to gross ratio WDI Investment growth rate Growth rate of investment to GDP ratio WDI Trade Trade volume to GDP ratio WDI Democracy Democratic accountability (scale : 0~6) ICRG Institution Constraint on executives index (scale : 1~7) Polity IV Tab. 7 : Source and description of regression variables Note : Trade volume is manufactured exports plus goods and service imports. Variable Obs. Mean Std. Dev. Min Max GDP growth rate 842 0.01927 0.02621-0.11885 0.10781 Corruption 456 3.28561 1.46136 0 6 Population growth rate 961 0.01480 0.01492-0.01641 0.18436 Human capital 1004 0.57329 0.35611 0.00350 1.54568 Investment growth rate 764 0.00293 0.03958-0.18018 0.18071 Trade 858 0.51062 0.37825 0.04813 3.97358 Democracy 456 4.10134 1.56994 0 6 Institution 832 4.59591 2.27035 0 7 Tab. 8 : Descriptive statistics of regression variables (1) (2) (3) (4) (5) (6) (7) (8) (1) 1.0000 (2) 0.1091 1.0000 (3) -0.3725-0.4504 1.0000 (4) 0.2067 0.6050-0.6828 1.0000 (5) 0.2521-0.0367 0.0778-0.1228 1.0000 (6) 0.1512 0.1214-0.1578 0.2665-0.0318 1.0000 (7) 0.2144 0.6540-0.6236 0.6965 0.0093 0.1470 1.0000 (8) 0.1322 0.4980-0.5843 0.5680 0.0669 0.0694 0.7613 1.0000 Tab. 9 : Correlation matrix of regression variables Note : (1) Growth rate of real GDP per capita, (2) Corruption index, (3) Growth rate of population, (4) Human capital stock, (5) Growth rate of investment to GDP ratio, (6) Trade to GDP ratio, (7) Democratic accountability index, (8) Institutional index(constraint of executives). 17

18 References Acemoglu, D., Johnson, S., and Robinson, James A., (2001), The Colonial Origins of Comparative Development : An Empirical Investigation, The American Economic Review, vol. 91, no. 5, pp. 1369-1401., (2002), Reversal of Fortune : Geography and Institutions in the Making of the Modern World Income Distribution, The Quarterly Journal of Economics, vol. 117, no. 4, pp. 1231-1294. Barro, Robert J., (1991), Economic Growth in Cross Section of Countries, The Quarterly Journal of Economics, vol. 106, no. 2, pp. 407-443. Caselli, F., Esquivel, G., Lefort, F., (1996), Reopening the Convergence Debate : a new look at cross-country growth empirics., Journal of Economic Growth, vol. 1, pp. 363-389. Cass, D., (1965), Optimum Growth in the Aggregative Model of Capital Accumulation, The Review of Economic Studies, vol. 32, no. 3, pp. 233-240. Dollar, D., and Kraay, A., (2003), Institutions, Trade, and Growth, Journal of Monetary Economics, vol. 50, pp. 133-162. Deininger, K., and Squire, L., (1996), A new data set measuring income inequality, World Bank Economic Review, vol. 10. Ehrlich, Isaac, and Lui, Fransis., (1999), Bureaucratic Corruption and Endogenous Economic Growth, Journal of Political Economy, vol. 107, no. 6. Frankel, Jeffrey A. and D. Romer, (1999), Does Trade Cause Growth? American Economic Review, Vol. 89, No. 3, pp. 379-99. Gould, David J., and Amaro-Reyes, Jose A., (1983), The Effects of Corruption on Administrative Performance, World Bank Staff Working Paper, no. 580. Gupta, Sanjeev, de Mello Luiz and Sharan, Raju, (2001), Corruption and Military Spending, European Journal of Political Economy, vol. 17, pp. 749-777. Gyimah-Brempong, K., (2002), Corruption, Economic Growth, and Income Inequality, Economic Governance, vol. 3, pp. 183-209. Gyimah-Brempong, K, and Traynor, T., (1999), Political Instability, Investment and Economic Growth in Sub-Saharan Africa, Journal of Africa Economy, vol. 8, pp. 52-86. Jain, Arvind K., (2001), Corruption : A Review, Journal of Economic Surveys, vol. 15, no. 1. Klitgaard, R., (1991), Gifts and Bribes, in Richard Zeckhauser, ed., Strategy and Choice, Cambridge, MA. Knack, S., Keefer, P., (1995), Institutions and Economic Performance : Cross-Country Tests Using Alternative Institutional Measures, Economics and Politics, Vol. 3, pp. 207-227. Koopmans, Tjalling C., (1965), On the Concept of Optimal Economic Growth, The Econometric Approach of Development Planning, Pontif. Acad. Sc. Scripta Varia. Vol. 28, pp. 225-300. Leff, N., (1964), Economic Development through Bureaucratic Corruption, American Behavioral Scientist, pp. 8-14. Leite, C. and J. Weidmann, (1999), Does Mother Nature Corrupt? Natural Resources, Corruption, and Economic Growth, IMF Working Paper. Levine, R. and Renelt, D., (1992), A Sensitivity Analysis of Cross-Country Growth Regressions, American Economic Review, vol. 84, pp. 942-963. Li, H., Xu, C. L., and Zou, H., (2000), Corruption, income distribution and growth, Economics and Politics, vol. 12, pp. 155-182. Mankiw, G., Romer, D., Weil, D. N., (1992) A Contribution to the Empirics of Economic

Growth, Quarterly Journal of Economy, vol. 107, pp. 407-437. Mauro, P., (1995), Corruption and Growth, The Quarterly Journal of Economies, vol. 110, no. 3, pp. 681-712. Mendez, Fabio, and Sepulveda, F., (2006), Corruption, Growth and Political regimes : Cross country evidence, European Journal of Political Economy, vol. 22. Meon, P. G., and Sekkat, K., (2005), Does Corruption Grease or Sand the Wheels of Growth?, Public Choice, vol. 122, no. 1/2, pp. 69-97. Moskalyk, Roman, (2008), Impact of Trade Openness and Technology Transfers on Growth: Panel Data Investigation for Developing Countries, Economics Education and Research Consortium Working Paper. Pak, Hung Mo, (2001), Corruption and Economic Growth, Journal of Comparative Economics, vol. 29, no. 1, pp. 66-79. Paldam, M., (2002), The Cross-Country Pattern of Corruption : Economics, Culture and the Seesaw Dynamics European Journal of Political Economy, vol. 18. Pellegrini, Lorenzo and Gerlagh, Reyer, (2004), Corruption s Effect on Growth and Its Transmission Channels, Kyklos, vol. 57. Romer, Paul M., (1990), Endogenous Technological Change, Journal of Political Economy, vol. 98, no. S5, pp. S71-S102. Sachs, J. and A. Warner, (1995), Natural Resource Abundance and Economic Growth, NBER Working Paper 5398 (Cambridge, MA : National Brueau of Economic Research). Sachs, J., and Warner, A., (1997), Fundamental Sources of Long-Run Growth, American Economic Review, vol. 8, pp. 184-188. Shleifer, A., and Vishny, Robert W., (1993), Corruption, The Quarterly Journal of Economics, vol. 108, no. 3, pp. 599-617. Solow, Robert M., (1956), A Contribution to the Theory of Economic Growth, The Quarterly Journal of Economics, vol. 70, no. 1, pp. 65-94. Tanzi, V., and Davoodi, H., (1997), Corruption, Public Investment, and Growth, IMF Working Paper. United Nations, (1989), Corruption in Government, New York : United Nations. Yanikkaya, Halit, (2003), Trade Openness and Economic Growth: A Cross-Country Empirical Investigation. Journal of Development Economics, Vol. 72, pp. 57 89. 19