Statistical Matching in agricultural economics: how to integrate different farm data sources

This study approaches the challenge of integrating different farm data sources by exploiting all the information that researchers already have at disposal. Indeed, due to privacy claim constraints, time-demanding data releasing procedures, etc. farm elementary data are often either partially available or difficult-to-find. In addition, collecting new data can be both highly expensive and time-demanding. A suitable solution is then to aggregate different data sources which are easily and/or already available. This work attempts to deal with both farm data shortage and the characteristics of the observational studies research context. Indeed, we integrate farm data resorting to the non-parametric micro SM hot deck techniques, which guarantee the preservation of the real observed data and avoid the misspecification bias generated by the parametric framework. This is remarkable in the observational studies research context (e.g. for the counterfactual analysis in agricultural economics) where, in order to assess a policy impact, we have to deal with just observed data. We apply the hot deck techniques to integrate FADN (Farm Accountancy Data Network) data and an ad hoc project survey referred to the Emilia-Romagna (Italy) farm population. We also propose a validation strategy in order to assess the matching goodness. Acknowledgement :