The study focuses on testing the hypothesis that the subsidy system of the Common Agricultural Reform in 1992 (CAP’92) drove to changes in farm efficiency towards the thereby claimed objectives. With sequential applications of semiparametric methods we succeed to identify the impact of the direct payments on environmental adaptation, productivity and efficiency before and after CAP’92 without restrictive model specifications. We find that the claimed objectives of the EU subvention policy were met only partly, but that the CAP’92 was, however, a step forward. Our case study uses large Spanish data sets of animal orientated farms. This paper applies non parametric methods for policy evaluation at firm level. The study focuses on testing the hypothesis that the subsidy system of the Common Agricultural Reform in 1992 (CAP’92) drove to changes in farm efficiency towards the thereby claimed objectives. We concentrate here on animal oriented farms, in particular cattle, pig, sheep and goat farms. The correct quantification of efficiency and productivity differentials due to CAP’92 is crucial for such a policy analysis as different models can easily lead to different conclusions. Using non parametric methods we do not need to specify the production function of the farms. With sequential applications of semiparametric methods we succeed to identify the impact of the direct payments on environmental adaptation, productivity and efficiency before and after CAP’92 without restrictive model specifications. We find that the claimed objectives of the EU subvention policy were met only partly, but that the CAP’92 was – at least partly – indeed a step forward in that sense. Our case study uses large Spanish data sets of animal orientated farms. This is justified, among other reasons, by the relevance of these farms for Mediterranean forest and grazing land preservation in Spain.