Identifying the Determinants of Corruption: An Application of Instrumental Variable Bayesian Model Averaging

Previous studies in the corruption literature have introduced numerous variables as the determinants of corruption. This articles aims to evaluate the robustness of potential determinants of corruption by addressing the model uncertainty and endogeneitry. The results derived from an instrumental variable Bayesian model averaging analysis indicate that based on the data of 123 countries, rule of law, with a posterior inclusion probability (PIP) of 1 and posterior mean of 0.662 has the most important role in keeping corruption under control among 36 explanatory variables. Government effectiveness, with a PIP of 0.964 and posterior mean of 0.358 is another significant variable in curbing corruption. Also, with a PIP of 0.965 and posterior mean of -0.194 the Asia dummy variable tells that corruption is a serious problem in the Asia region. Further, confining the analysis to 95 developing countries reveals that rule of law with a PIP of 0.999 and posterior mean of 0.684 is the most critical variable in the fight against corruption.