The Impact of Corruption and Institutional Quality on Trade Protection, Papers of Economics

Download The Impact of Corruption and Institutional Quality on Trade Protection and more Papers Economics in PDF only on Docsity! Corruption and Trade Protection: Evidence from Panel Data Subhayu Bandyopadhyay* & Suryadipta Roy** September 2006 Abstract We complement the existing literature on corruption and trade policy by providing new estimates of the effects of corruption (and institutions) on trade protection. We control for unobserved heterogeneity among countries with properly specified fixed effects, exploiting the time dimension present in the dataset. The issue of endogeneity of corruption with respect to trade policy is addressed. Furthermore, two separate institutional measures are used in the same regression to estimate their comparative impacts on trade policy. The central finding is that corruption and lack of contract enforcement significantly increase trade protection and have negative effects on trade openness. Keywords: Corruption; Trade Policy; Contract Enforcement; Endogenous Tariffs. JEL Codes: F1; F13. ______________________________ *Corresponding author: Associate Professor, Department of Economics, West Virginia University, Morgantown, West Virginia 26505-6025, and, Research Fellow, IZA-Bonn. Email: subhayu.bandyopadhyay@mail.wvu.edu. ** Visiting Assistant Professor, Department of Economics, Lawrence University, Appleton, WI -54911. Email: suryadipta.roy@lawrence.edu 1 1. Introduction The discussion on the effects of corruption on trade policy has its roots in the broad literature on the political economy of trade policy. Given that trade policies often deviate from first best solutions, this strand of the literature has focused on the endogeneity of trade policy. The primary explanation in this regard has been that policy makers do not maximize national welfare. They choose trade policies in response to demands from the special interest groups. Given that trade policy is endogenous, it is natural for capitalist producers in import competing sectors to lobby governments for trade barriers, as has been argued among others by Olson (1965). Grossman and Helpman (1994) developed the “protection for sale” model where they took into account the strategic interactions between the government and the special interest groups. The outcome of this interaction is determined in an equilibrium where the government implements trade policies after taking into account the tradeoffs associated with receiving campaign contributions vis-à- vis reduced consumers’ welfare. Our paper complements this idea and shows that corrupt governments that are more susceptible to lobbying will extend higher levels of trade protection. The level of corruption in an economy can thus serve as a proxy for the amenability of the government to lobby pressures and trade policies are inherently more likely to be protectionist in corrupt countries. Our contention is that the level of trade protection is positively correlated with the level of corruption and that the latter is an important measure of institutional support for special interest groups. This should raise the ability of these groups to successfully lobby for protection. The paper estimates this effect using cross-country regressions over time. We have addressed the endogeneity of 4 (per se), rather than on trade volumes or changes in tariffs. Secondly, we undertake a panel specification using region and time fixed effects, instead of using an ordinary least squares (OLS) specification. Apart from the corruption variable, we have also used a variable describing the quality of institutions – “the risk of contract enforcement”, as a control variable, along with corruption. This measure has been used by Anderson and Marcouiller (2002) within the context of a gravity model, where they find lack of contract enforcement to significantly reduce international trade. They argue that strong institutional support for trade among high income countries lowers transactions costs and consequently allows greater trade between rich countries. Our specification which uses both the corruption and the contract enforcement variable in the same regression allows us to make richer predictions in terms of the importance of the institutional indicators on trade policy. 3. Econometric specification The existing work on the effect of corruption on trade protection (by Lee and Afzar, 2002) bases its results on pooled cross-section estimates and does not account for unobserved (or not included) heterogeneity between countries. In a panel specification, OLS estimates generally tend to be biased and inconsistent due to the correlation of the regressors with the unobserved fixed effects. In order to address this problem, we explicitly allow for region and time fixed effects in our regressions. The region fixed effects would capture the tendency for countries within a region to organize themselves into Free Trade Areas (FTA) or Customs Unions (CU). A fixed effects model with country-specific effects, on the other hand, will not be able to properly identify the 5 estimates for the most important corruption variable, given little variability of the data within groups. For example, Dutt and Mitra (2005) note this problem. Moreover, such an approach also uses up large degrees of freedom and results in high multicollinearity between the country-specific effects and some of the right-hand side variables, thereby making the interpretation difficult. Use of region-specific effects also allows us to employ time constant variables as instruments for corruption and contract enforcement, something that would not have been possible with country-fixed effects. Time specific effects address the issue of regressor endogeneity due to correlation of the right-hand side variables with the error terms in other periods. They also capture any pattern that the countries exhibit as a group over the years, e.g., whether all countries tend to become more protectionist or less protectionist (especially since our data cover the period of signing the Uruguay Round negotiations and the launching of the WTO). Our main econometric specification is of the form: ititittjit eXCorruptiony +′++++= βθγλα (1) where, ity denotes the level of trade protection in country i at time period t, α is the common intercept term, sj −λ are the region-specific effects, and st −γ are the time- specific effects common to all countries. itX -s are the set of control variables in all the equations. Our first point of departure is the introduction of jλ and tγ in equation (1). The central variable of interest is the corruption variable, our proxy for the susceptibility of the government to lobbying pressures. We also perform robustness checks to test the validity of our hypotheses. In a dynamic context, the level of corruption may be endogenous with respect to trade policy. An improvement in institutions in a country can lead to a reduction in bureaucracy which in turn might lead a further lowering of 6 corruption. Furthermore, as Rodrik (2000) has argued, trade reforms not only lead to a change in import prices but also result in institutional reforms. Hausman tests that were conducted suggest endogeneity of the corruption variable. Therefore, we performed a two-stage least squares estimation, where we instrument the corruption variable by a dummy variable indicating whether the country was a British colony. The British colonial heritage dummy has been found to be associated with significantly lower corruption in cross-country regressions and has been previously discussed as an instrument for corruption in the existing literature (La Porta et al, 1999; Treisman 2000, Acemoglu et al. 2001). We find this variable to be highly significant in all first-stage regressions for corruption in the presence of other control variables and hence used it as an instrument for corruption in the 2SLS regressions. This is our second departure from the previous literature. 4. Data The key independent variable in this study is the International Country Risk Guide’s (ICRG) popular index for corruption in government. According to Knack and Keefer (1995), who used the variable to explain investment and economic growth, lower scores for this variable indicate greater likelihood for government officials to demand special payments and/or bribes connected with import and export licenses, exchange controls, tax assessment, policy protection and loans. This comes fairly close to capturing bureaucratic corruption as we have conceptualized it here. The other measure of institutional efficacy that we have used in this study is the repudiation of contracts by government that indicates the risk of a contract being annulled by the government due to 9 Spain, etc., to a high of around 53% for India. The standard deviation is greater for taxes on international trade with the maximum going up to almost 75% for Uganda. The greatest standard deviation across countries is displayed by variations in the level of trade-GDP ratio with a low of around 6% for Ghana to a high of over 400% for Singapore. In general, advanced countries tend to have low levels of protection as is apparent in the second table. Import duty, export duty, and international trade tax are all negatively correlated with the level of real GDP while the trade-GDP ratio is positively correlated with real GDP. Levels of trade protection, measured by the two indicators are positively correlated with the level of corruption and the lack of contract enforcement. On the other hand, the level of openness determined by the trade-GDP ratio is negatively correlated with the levels of corruption and contract enforcement. Both the institutional variables are negatively correlated with the level of real GDP, government expenditure, and current account deficit. The level of trade openness is positively correlated with real GDP, government expenditure, and the current account balance. Table 2 presents the OLS estimates of the effect of corruption on trade protection (with and without control variables). While corruption appears with the correct sign in all the regressions, the variable loses significance when control variables are introduced in the regression on import duty. The variable is positive and significant for trade tax, while it is negative and significant in the case of trade-GDP ratio. The OLS regressions also show the low-income countries to have higher levels of trade protection in all the regressions. The middle-income countries are found to have higher levels of import duties and lower levels of trade-GDP ratio in comparison to the high-income countries (the excluded dummy variable in our regressions). 10 Given that OLS estimates are biased and inconsistent in panel regressions, we use the fixed effects specification given in equation (1) in order to control for individual heterogeneity in estimation. The results for the fixed region-and-time-effects are presented in Table 3. In case of the import duty, corruption is still not significant while it is significant in case of the trade tax and the trade-GDP ratio. For the international trade tax, a one standard-deviation increase in corruption is associated with a 0.096 standard deviation increase in the tax. Trade tax is also negatively correlated with real GDP, government expenditure, current account balance, and population. In the regression on trade openness, a one standard-deviation increase in the level of corruption is associated with a 0.23 standard deviation reduction in the trade-GDP ratio. However, real GDP, government expenditure, and the level of current account balance are not significant with respect to the level of openness. The population variable is significant and negative, showing that an increase in the population size reduces the level of openness. As discussed above, there might be an issue of endogeneity between the level of corruption and trade protection in a cross-country context. Keeping this in mind, Hausman tests were conducted that rejected the null hypotheses of exogeneity of the corruption variable. We took care of this endogeneity by instrumenting for corruption in a two-stage least squares regression where a dummy variable indicating the past colonial origins of the country was used as the instrument. This variable was also found to be negative and highly significant in a first-stage regression for corruption in the presence of other control variables. The colony dummy was partly constructed from Demirguc-Kunt and Levine (2001) for the majority of the countries in our dataset. Data on colonial origins for the rest of the countries were obtained from the Wikipedia website 11 (http://en.wikipedia.org/wiki/British_Empire). The results for the two-stage least squares regressions are reported in Table 4. Corruption is found to be highly significant in all the regressions for import duty, international trade tax, and the trade-GDP ratio. Moreover, the effects of corruption on trade protection are also substantially larger after correcting for endogeneity. A one standard deviation increase in corruption is associated with 1.23 and 0.5 standard deviation increases in import duty and international trade tax, respectively, while the trade-GDP ratio is reduced by almost 4%. Both real GDP and government expenditure, however, change signs in case of the instrumental variable regression. Both the variables are positively correlated with import duty, negatively correlated with the level of openness measured by the trade-GDP ratio, and are not significant in case of taxes on international trade. The change in the signs can be attributed to the correlation of the instrument variable used in the regression with the other explanatory variables used in the regression, namely real GDP, government expenditure, current account balance, and population. As Anderson and Marcoullier (2002) have argued, lack of contract enforcement adds to the transactions cost between North-South trade and significantly reduces the trade volume. In order to estimate the effect of contract enforcement on trade policy and trade protection, we re-estimated equation (1) using both the measures of corruption and contract enforcement, along with other control variables. Given the endogeneity issue affecting both these indicators, it might be advisable to instrument both the variables. In order to do this, we used an index for linguistic fractionalization developed by Alesina et al. (2003) for about 190 countries along with the colony dummy variable. Indices based 14 References Acemoglu, Daron, Simon Johnson, and James. A. Robinson, 2001, “The Colonial Origins of Comparative Development: An Empirical Investigation,” American Economic Review, 91 (5), 1369-1401. Ades, Alberto, and Rafael Di Tella, 1999, “Rents, Competition and Corruption,” American Economic Review, 89(4), 982-994. Alesina, Alberto, Arnaud Devleeschauwer, William Easterly, Sergio Kurlat, and Romain Wacziarg, 2003, “Fractionalization,” Journal of Economic Growth, 8(2), 155-94. Anderson, James E. and Douglas Marcouiller, 2002, “Insecurity and the Pattern of Trade: An Empirical Investigation.” Review of Economics and Statistics, 84(2), 342-52. Bandyopadhyay, Subhayu, Sajal Lahiri, Suryadipta Roy, 2006, “CU or FTA? The Role of Political Asymmetries”, Working paper series, West Virginia University. Bardhan, Pranab, 1997, “Corruption and Development: A Review of Issues,” Journal of Economic Literature, 35(3), 1320-46. Bhagwati, Jagdish N., 1982, “Directly Unproductive, Profit-seeking (DUP) Activities,” Journal of Political Economy, 90(5), 988-1002. Demirguc-Kunt, Asli, and Ross Levine, 2001, eds. “Financial Structure and Economic Growth: A Cross-Country Comparison of Banks, Markets, and Development,” Cambridge and London: MIT Press. Dutt, Pushan, 2005, “Does Protection Beget Corruption?” INSEAD Working Paper Series. 15 Dutt, Pushan, and Devashish Mitra, 2005, “Political Ideology and Endogenous Trade Policy: An Empirical Investigation.” The Review of Economics and Statistics, 87(1), 59-72. Easterly, William, and Sergio Rebelo, 1993, Marginal Income Tax Rates and Economic Growth in Developing Countries,” European Economic Review, 37(2-3), 409-17. Easterly, William, and Ross Levine, 1997, “Africa's Growth Tragedy: Policies and Ethnic Divisions,” Quarterly Journal of Economics, 112(4), 1203-50. Gatti, Roberta, 2004, “Explaining Corruption: Are Open Countries Less Corrupt?” Journal of International Development, 16(6), 851-61. Grossman, Gene M. and Elhanan Helpman, 1994, “Protection for Sale,” American Economic Review, 84(4), 833-50. Knack, Stephen, and Philip Keefer, 1995, “Institutions and Economic Performance: Cross-Country Tests Using Alternative Institutional Measures,” Economics and Politics, 7(3), 207-27. Krueger, Anne O., 1974, “The Political Economy of the Rent-Seeking Society,” American Economic Review, 64(3), 291-303. La Porta, R., F. Lopez-de-Silanes, Andrei Shleifer and Robert Vishny, 1999, “The Quality of Government,” The Journal of Law, Economics, & Organization, 15(1), 222-279. Lee, Young, and Omar Azfar, 2002, “Does Corruption Delay Trade Reform?” Working Paper, IRIS Center, University of Maryland. Mauro, Paolo, 1995, “Corruption and growth.” Quarterly Journal of Economics, 110(3), 681–712. 16 Olson, Mancur, 1965, “The Logic of Collective Action,” Harvard University Press. Rodrik, Dani, 2000, “Trade Policy Reform as Institutional Reform.” Mimeo. Rodrik, Dani, 1998, “Why Do More Open Economies Have Bigger Governments?” Journal of Political Economy, 106(5), 997-1032. Sachs, Jeffrey D. and Andrew M. Warner, 1995, “Economic Reform and the Process of Global Integration,” Brookings Papers on Economic Activity, 0(1), 1-95. Treisman, Daniel, 2000, “The Causes of Corruption: A Cross-National Study,” Journal of Public Economics, 76(3), 399-457. Wei, Shang-Jin, 2000, “Natural Openness and Good Government,” National Bureau of Economic Research, Inc, NBER Working Papers, # 7765. 19 Table 2: OLS regression Import duty Trade tax Trade-GDP ratio Import duty Trade tax Trade-GDP ratio Corruption .117*** (0.224) .205*** (.348) -.155*** (1.599) .049 (.23) .116*** (.329) -.117** (1.504) Real GDP -.534*** (.031) -.392*** (.041) .002 (.215) -.449*** (.041) -.344*** (.049) -.163** (.327) Government expenditure .064** (.04) -.072** (.067) .037 (.346) .071*** (.041) -.075** (.072) .012 (.395) Current account balance .051** (.032) -.112*** (.049) .114** (.323) .105*** (.145) -.032 (.051) .099** (.351) Population (‘000,000) 0.127** (.0004) -.023 (.0002) -.172*** (.0009) .091* (.0003) -.072*** (.0002) -.155*** (.0008) Low income .288*** (1.075) .325*** (1.641) -.262*** (6.495) Mid income .089* (.874) .025 (1.283) -.223*** (7.296) No. of observations 1056 1056 1056 1056 1056 1056 R-square 0.36 0.39 0.08 .40 .47 .10 *- significant at 10% level; **- significant at 5% level; ***- significant at 1% level; standardized beta coefficients in each cell; robust standard errors in parentheses. 20 Table 3: Fixed effects model with time-specific & comprehensive region-specific effects Import duty Trade tax Trade-GDP ratio Corruption .009 (.235) .096*** (.34) -.231*** (1.691) Real GDP -.213*** (.028) -.145*** (.044) -.017 (.381) Government expenditure -.052* (.044) -.234*** (.07) .045 (.277) Current account balance .016 (.022) -.167*** (.043) .054 (.257) Population (‘000,000) .102*** (.0004) -.109*** (.0002) -.227*** (.00001) No. of years 16 16 16 Average # of countries each year 66 66 66 Overall R-square 0.48 0.58 0.26 *- significant at 10% level; **- significant at 5% level; ***- significant at 1% level; standardized beta coefficients in each cell; robust standard errors in parentheses. 21 Table 4: Fixed effects IV model with time-specific and comprehensive region-specific effects Import duty Trade tax Trade-GDP ratio Corruption 1.232*** (2.065) .514** (2.049) -3.588*** (27.499) Real GDP .369** (.148) .054 (.146) -1.618*** (1.929) Government expenditure .274** (.146) -.123 (.15) -.842*** (1.944) Current account balance .001 (.059) -.172 (.046) .095 (.659) Population (‘000,000) .158** (.0004) -.09*** (.0002) -.382*** (.00003) No. of years 16 16 16 Average # of countries each year 66 66 66 Overall R-square 0.22 0.52 0.06 *- significant at 10% level; **- significant at 5% level; ***- significant at 1% level; standardized beta coefficients in each cell; robust standard errors in parentheses.