While empirical studies which analyze large cross section country data find that corruption lowers investment and thereby economic growth, this result cannot be established for certain subsamples of countries. We argue that one reason for these mixed findings may be that a country's corruption and growth rates are tightly linked as variables of a dynamic process which can have several equilibria or have different sets of equilibria. In order to understand the circumstances in which a country

more »

... verges towards a certain equilibrium, we model the individual decisions to invest and corrupt as an evolutionary game. In this model the quality of government institutions is an endogenous variable, depending on the corruption rate, the population income, and the type of institutions; the quality of institutions itself then determines the future incentives to corrupt. The comprehension of these feedback effects allows us to study the role of the type of institutions for the dynamics of corruption. We present the equilibria for different types of institutions and discuss the resulting dynamics. The results suggest that cross country studies may significantly underestimate the impact of corruption on growth for certain countries. Depending on how the quality of institutions depends on corruption and income, corruption can either lower growth, suppress it entirely, or be positively correlated with growth in some special situations.