To match, not to match, how to match: Estimating the farm-level impact of the CAP-first pillar reform (or: How to Apply Treatment-Effect Econometrics when the Real World is;a Mess)

This paper aims at evaluating the impact of the 2003/2005 CAP reform on farm production choices. The outcome of "market orientation" is measured by considering both the short-term production choices and the long-term investment decisions. The Treatment Effect (TE) is estimated through alternative approaches due to the difficulties encountered in finding appropriate counterfactuals. Different versions of the Propensity Score Matching (PSM) estimators, the Difference-In-Difference (DID) estimate, and alternative multiple/continuous TEs estimates, based on the Generalized Propensity Score (GPS), are performed, their statistical robustness assessed and results compared. Results show that the 2003/2005 reform of the first pillar of the CAP actually had an impact more in (re)orienting short-term farm production choices then investment decisions and this effect is significantly more evident for farms with a limited contribution of the CAP on their own Gross Production Value.