Andrei Drives Bentley, Boris Drives Lada: Why Corrupt States Have Clean Agencies.

Abstract

In societies where corruption is rife, why do bureaucrats in some agencies extract more than enough to get rich, while others do not? This dissertation explains why corruption varies across agencies within the same state, when the factors that explain variation across states (e.g. political competition) are weak or absent. In other words, why in corrupt states aren't all agencies equally corrupt?

My theory explains how these puzzling cases arise from the interaction between bureaucrats in a hierarchy. They must balance risky collusion necessary to grow the available rents, against competition to take a larger share of these spoils. Characteristics of agencies well-known in the bureaucratic politics literature change the terms of this trade-off, most importantly the degree of information asymmetry between bureaucrats authorizing corruption, and those carrying it out. The greater this asymmetry, the more difficult it is to maintain the collusion necessary to succeed at corruption. I test this and other implications using a novel dataset on Russian public procurement covering five million purchases by 110,000 agencies from 2011-2015.