DEMOCRACY AND CORRUPTION: SOME NEW EVIDENCE Jacob Musila Athabasca University, Canada jacobm@athabascau.ca corruption. ABSTRACT This study explores the relationship between democracy and corruption. Using data on African countries, I find the relationship between democracy and corruption measures to be statistically significant and nonlinear. The coefficients of the nonlinear regressions suggest that corruption is slightly lower in autocracies than in partial democracies; and once past a threshold, corruption is substantially lower in full or near-full democracies. Besides the democracy variables, I find legal systems, religious traditions, and ethnic fractionalization to be important determinants of cross-country variations in corruption rates. OVERVIEW OF PRESENTATION the problem/issue literature analytical framework empirical analysis results threshold democracy conclusion THE PROBLEM OF STUDY Gary Becker observed that the former Soviet bloc countries have similar long-run growth rates as those in developed Western economies. Ehrlich and Lui (JPE, 1999) claimed that autocratic regimes could achieve growth rates equal or higher than decentralized democracies because corruption is much constrained in the autocracies. Ehrlich and Lui’s observation, together with the empirical analyses that find a positive correlation between democracy and corruption, suggests that the relation between democracy and corruption is nonlinear. Nobody has successfully test this possibility. EVIDENCE 1 Corporate Ethics and Democracy, 2004 70.0 60.0 Corporate Ethics Index 50.0 40.0 y = 0.0095x2 - 0.7768x + 42.651 R2 = 0.3253 30.0 20.0 10.0 0.0 0.0 10.0 20.0 30.0 40.0 50.0 60.0 Political & Civil Rights Index 70.0 80.0 90.0 100.0 EVIDENCE 2 Public Sector Ethics and Democracy, 2004 60.0 50.0 Public Sector Ethics Index 2 y = 0.0075x - 0.4717x + 24.435 2 R = 0.4619 40.0 30.0 20.0 10.0 0.0 0.0 10.0 20.0 30.0 40.0 50.0 60.0 Political & Civil Rights Index 70.0 80.0 90.0 100.0 EVIDENCE 3 Corporate Goverance and Democracy, 2004 90.0 80.0 Coporate Goverance Index 70.0 y = 0.0078x2 - 0.5241x + 42.646 R2 = 0.261 60.0 50.0 40.0 30.0 20.0 10.0 0.0 0.0 10.0 20.0 30.0 40.0 50.0 60.0 Political and Civil Rights Index 70.0 80.0 90.0 100.0 EVIDENCE 4 Corruption Perception, 2004 7.0 6.0 CPI Score 5.0 y = 0.0008x2 - 0.0489x + 2.7782 R2 = 0.725 4.0 3.0 2.0 1.0 0.0 0.0 10.0 20.0 30.0 40.0 50.0 60.0 Political & Civil Rights Index 70.0 80.0 90.0 100.0 LITERATURE Linear relationship: Sung, Hung En (Crime, Law & Social Change, 2004) Treisman, Daniel (Journal of Pub. Economics, 2000) Nonlinear relationship: Montinola and Jackman (BJPS, 2002, p.166) Corruption = - 14.855 + 2.235 log (GDP/POP88) - 1.816 OPEC member + 0.043 log (Govt. size88) + 0.203 democracy88 + 0.080 (democracy88)2 Estimator = OLS; adjusted R2 = 0.76 ANALYTICAL FRAMEWORK Corruption function: Φ = b(y) – c(x). (1) Φ is non-decreasing in b(y) and non-increasing in c(x). If b(y') ≥ b(y) or c(x') c(x), then for rational individuals Φ(b(y'), c(x')) ≥ Φ(b(y), c(x)). (2) EMPIRICAL ANALYSIS A well known determinant of corruption is the risk of getting caught and punished. According to public choice literature, this risk depends on: (a) Effectiveness of the legal system (b) Religious traditions (c) Historical/colonial traditions (d) Ethnic divisions (e) Democracy EMPIRICAL MODEL Corruption = a0 + a1 former British colony + a2 common law system + a3 politically relevant ethnic groups + a4 Christian affiliation + a5 democracy +a6 democracy-squared + e (3) where the corruption variable takes on corporate ethics index, public sector ethics index, corporate governance index, and corruption perception index; ai’s are coefficients to be estimated; and e is the error term (i.e., white noise). DATA Variable Corporate ethics (cei) Public sector ethics (pei) Corporate governance (cgi) Corruption perception (cpi) Former British colony (fbc) Common law system (cls) Christian affiliation Politically rel. ethnic groups Democracy source Kaufman (GCR, 2004) ,, ,, Transparency International Treisman (J. Pub E, 2000) ,, CIA World Factbook Posner (AJPS, 2004) freedomhouse.org value 0<cei<100 0<cpi<10 1 if fbc; 0 otherwise 1 if cls; 0 otherwise % of Christians in pop 0 = auth; 100= demo RESULTS: Table 1, OLS Estimates (most exogenous variables only) Dependent variables Explanatory variables Constant Former British colony Common law system Politic rel. ethnic groups Christian religion affil. Adjusted R-Square Number of observations Corporate ethics Public sector ethics Corporate governance Corruption perception 24.249*** (4.076) 0.111 (0.017) 12.156 (1.583) -16.078 (-1.647) 0.174 (1.695) 23.031*** (3.414) 3.845 (0.507) 6.303 (0.724) -23.882** (-2.154) 0.134 (1.155) 27.668*** (3.689) 8.893 (1.055) 11.571 (1.195) -13.059 (-1.060) 0.136 (1.055) 3.460*** (10.077 -0.554 (-1.121) 1.150* (2.030) -2.932*** (-5.151) 0.008 (1.343) 0.483 19 0.294 19 0.483 19 0.569 28 RESULTS: Table 2, OLS Estimates (democracy introduced nonlinearly) Dependent variables Explanatory variables Constant Corporate ethics 51.204*** (4.196) Former British colony -8.580* (-1.778) Common law system 23.530*** (4.038) Politic rel. ethnic groups -5.921 (-0.819) Christian religion affiliation 0.022 (0.279) Democracy -1.363*** (-3.239) Democracy-square 0.014*** (3.732) Adjusted R-square 0.774 Number of observations 19 Public sector ethics Corporate governance Corruption perception 34.983* (2.164) -4.491 (-0.702) 15.323* (1.985) -10.968 (-1.145) 0.009 (0.087) -0.894 (-1.605) 0.011* (2.105) 0.579 19 47.175** (2.418) 0.248 (0.032) 21.893** (2.350) -1.277 (-0.110) -0.004 (-0.034) -1.139 (-1.693) 0.013* (2.062) 0.626 19 3.476*** (4.658) -0.323 (-0.705) 0.596 (1.058) -1.821*** (-3.596) 0.008* (1.893) -0.047* (-1.971) 0.001*** (3.027) 0.772 18 COMBINED EFFECTS OF DEMOCRACY VARIABLES The coefficients of the combined democracy variables computed à la Ralph and Srivastava (The American Statistician, 1979) were positive and statistically significant at the 5% level or better. The t-ratios for combined democracy variables in the regressions are as follows: Corporate ethics Public sector ethics Corporate governance Corruption perception Democracy and Democracy-square 4.464 3.512 2.909 4.579 THRESHOLD DEMOCRACY Democracy (%) Corporate ethics index Public sector ethics index Corporate governance index Corruption perception index 48.7 40.6 43.8 23.5 corresponding index 26 16 35 2.1 CONCLUSION This study empirically tests for nonlinearity between democracy and the various corruption measures for African countries. The results show that the correlation between democracy and corruption is statistically significant and the relationship is of first-degree polynomial. The estimates suggest that authoritarian countries are slightly less prone to corruption than countries at intermediate levels of democracy, and, that beyond the threshold level of democracy, more democratic countries are less prone to corruption. The threshold levels of democracy range between 23.5% and 48.7% for African countries depending on the measure of corruption. The estimates also suggest that countries with large Christian followers are perceived to be less corrupt, countries with high ethnic divisions are perceived to be more corrupt, and countries with common law legal system have higher corporate and public sector ethics as well as higher corporate governance standards.