The effect of the decentralization degree on corruption : a new interpretation

This work contributes to empirical studies on decentralization and corruption by trying to resolve the uncertainty that the literature so far has shown. It also gives reasons supporting the ‘best’ decentralization structure which a country can adopt to discourage corrupt behaviour, and suggests an intermediate degree of decentralization. The trade-off between the moral hazard and the adverse selection aspect of the principal-agent framework, that emerges in this literature, can be better captured by a non-linear specification (e.g. cubic, as the more general non-linear model); neither very small nor very high degrees of decentralization are appropriate against corruption, but an intermediate one. Being monitored by the voters, local politicians, in a intermediate decentralized setting, have an incentive to perform in the voters interest and, being local resources they manage not very much, they have little incentive to appropriate part of such resources for personal use.