Using the World Bank database “Distortions to agricultural incentives,” this paper analyzes the impact that agricultural (dis)incentives have on food security for a wide sample of countries over the 1990–2010 period. We adopt a continuous treatment approach applying generalized propensity score matching to reduce potential biases stemming from differences in observed country characteristics. The results provide strong evidence of self-selection and heterogeneous food security impacts at different levels of policy intensity. Estimates of the dose-response functions show that both discrimination against agriculture and large support for it lead to poor performance in the availability, access, and utilization dimensions of food security.