This paper studies the effect of political regime transitions on public policy using a dataset on global agricultural distortions over 50 years (including data from 74 developing and developed countries over the period 1955-2005). We employ both difference-in-differences regressions and semi-parametric matching methods, exploiting the time series and cross-sectional variation in the data. Our semi-parametric estimates show that parametric methods might underestimate the effect of democracy on public policy. In addition, we find that the effect is asymmetric: agricultural protection increases after a country's transition to democracy of about 9% points, but there is no effect when the political regime shifts from democracy to autocracy. Overall, the evidence supports the redistributive nature of democratic institutions toward the majority and, therefore, it is consistent with the median voter model of political behaviour.