Dictatorship is the predominant political system in many developing countries. However, different dictators act quite differently: a good dictator implements growth-enhancing economic policies, e.g. investment in public education and infrastructure, whereas a bad dictator expropriates wealth of her citizens for her own consumption. The present paper provides a theoretical model by deriving underlying determinants of dictatorial behavior. We assume that the engine of economic growth is private investment. It can increase the productivity of individuals who invest, as well as the aggregate technological level. A good dictator encourages this investment in order to expropriate more. However, the cost of this encouragement is that the ensuing higher growth rate will induce earlier democratization. In this paper we will illustrate the trade-off between economic benefits from a growth-enhancing policy in the short run and the shorter life-time of the dictator in the long run. Furthermore, we will find that the higher the return from private investments is the less likely the dictator will be a good one. Contrary to McGuire and Olson (1996) we find that a long life-time does not always induce positive incentives among dictators.