Do Criminal Politicians Reduce Corruption? Evidence from India

This paper relates unique data on criminal records of local politicians in India to corruption, crime and poverty. Using a regression discontinuity design, whereby individuals living in districts where a criminal politician barely won are compared to individuals living in districts where a criminal politician barely lost, this paper shows that criminal politicians reduce bribe-taking behavior of law and order officials by 34 percent. One possible explanation for this result is that when interests of politicians and those of interest groups converge, criminal politicians' control over bureaucrats acts as a substitute for bribes from these interest groups. This is not to say that criminal politicians should be elected to eradicate corruption, but rather that corruption is underestimated if only measured by bribe-taking without taking into account political control: as less bribes need to be paid, criminal offences, similar to those mostly committed by criminal politicians, increase by 25 percent. Moreover, the urban headcount ratio, the welfare of those not connected with politicians, increases by 22 percent