This paper analyzes which institutional features contribute to poverty reduction when countries democratize. For this, theories and data are used that distinguish between different areas of democratic institutions - namely electoral, liberal, participatory, deliberative, and egalitarian democratic institutions. The data analysis uses semi-parametric treatment effects estimates to estimate average treatment effects of democratization on poverty reduction and estimates this separately for countries with relatively weaker and stronger institutions at the time of democratization. The estimations reveal no clear pattern of a specific, predominant area of institutions while others are remain less important. In each area of institutions, some of its features contribute to poverty reduction and some do not. Especially, the hypothesis that stronger institutions lead to poverty reduction cannot confirmed. Stronger institutions only lead to significant poverty reduction for the institutions of executive and legislative being elected, and judicial constraints on the executive. For most other significant effects, weaker institutions lead to poverty reduction. This is likely due to triggered changes in institutions by democratization that only occur when institutions are still relatively weak, rather than due to growth or inequality changes. When the effect of democratization on poverty reduction is significant, it is meaningful in size (ranging from around 12 to 25 percent during the first five years after democratization, depending on the specific mid- and low-level institution). The pattern is the clearest for deliberative, participatory, and egalitarian institutions and less clear for electoral and liberal democratic institutions. This illustrates that institutions which capture how responsive policymaking is to its constituencies are more important for poverty reduction than institutions that capture formal aspects of democracy.