Uncover the predictive structure of healthcare efficiency applying a bootstrapped data envelopment analysis

One of the main problems in efficiency analysis is to determinate the environmental variables that have an impact on the production process. This paper shows that applying bootstrap to data envelopment analysis (DEA) before performing classification and regression trees (CART) increase the quality of the results. In particular, employing data on the Italian Health System, the paper highlights that bias corrected DEA allows to individuate variables affecting health efficiency which would remain undiscovered when the traditional DEA model is applied.

► We propose a method to identify environmental variables that affect efficiency.
► We use bootstrap-DEA and CART analysis in a two stage approach.
► The accuracy of the variable selection thought CART increase applying bootstrap-DEA.
► Italian health efficiency is influenced by the decisions of local governments.
► Italian health efficiency is influenced by the patient mobility.