The aim of this paper is to assess the causal impact of trade policy distortions on food security. The added value of this work is twofold: i) its use of a non-parametric matching technique with continuous treatment, namely the Generalised Propensity Score (GPS) to address the self selection bias; ii) its analysis of heterogeneity in treatment (by commodities) as well as in outcome (i.e. dierent dimensions of food security). The results of our estimates clearly show that trade policy distortions are, overall, signicantly correlated with the various dimensions of food security analysed. Both discrimination against agriculture and 'excessive' support lead to poor performances in all dimensions of food security (availability, access, utilisation and stability).