Calculating Corruption

Why do some authoritarian regimes exhibit high levels of corruption, while others produce very little? In this study, I show how corruption is used as a signal of performance and loyalty in autocratic regimes. I find that elites in non-democratic regimes reduce corruption in the face of political competitiveness. I test this theory using extensive micro-level data on the public's experiences with bribery in contemporary Russia. This data set is comprised of over 180,000 responses to public opinion surveys from 2001-2016 in Russia's subnational units. Identification of the causal effect of political competition on corruption is achieved with the use of an exogenously-determined electoral calendar--I show how the scheduled end of a term in office is an exogenous positive shock to political competition for authoritarian leaders in Russian regions, a shock that decreases experienced bribery by over 13% in those years. A wide array of alternative measures including novel search engine data and crime statistics support my conclusions. I also show that governors' tenuous hold on their positions--all the more tenuous when in their final years of a term in office--can be bolstered by additional resources that may be at their disposal. By showing how shocks to political competition drive governors to reduce corruption levels for fear of losing their jobs, but also that those shocks have varying effects for different governors, I illustrate the power of a dissatisfied public and authoritarian formal rules to shape behavior in non-democratic regimes. I also examine the linking assumption between public dissatisfaction and corruption experiences. These findings have implications for our understanding of autocratic politics, corruption, and studies of Russia. I show that corruption in authoritarian regimes is not a byproduct of authoritarianism, nor is it merely a result of low capacity--it is also a means of rule and control for autocrats. Modern authoritarian rulers are more discriminating in their application of petty corruption than is commonly understood. Finally, I employ and extend multilevel regression and poststratification (MRP) to generate descriptive estimates of corruption as experienced by the public with much greater accuracy and precision than has been possible previously.