Do existing corruption levels matter in controlling corruption?

We examine the determinants of corruption using recent cross-sectional data for nearly one hundred countries. While the causes of corruption have drawn economists' interest in recent years, our main contribution is to examine the corruption determinants throughout the conditional distribution of corruption across nations. Are there different causes of corruption in highly corrupt nations compared to the least corrupt countries? For instance, we examine whether greater democracy and more economic freedom consistently reduce corruption among the most and the least corrupt. Our results for the significant determinants support some findings in the literature, but also provide new conclusions. In many cases, quantile regression estimates are quite different from those from OLS regressions. Among the most corrupt nations, larger governments and greater economic freedom do not appear to reduce corruption, but greater democracy seems to alleviate it. Our results suggest that some current corruption control policies may be reconsidered, especially among the most corrupt and least corrupt nations.