Competition and Public School Efficiency in Sweden – An Empirical Evaluation of Second Stage Regression Results for Different Models of Nondiscretionary Inputs in Data Envelopment Analysis (DEA)

The focus in this study is on how efficiency in public education is affected by competition from private schools. The Swedish educational system is used, since the Swedish large scale voucher program implies that private and public schools compete on similar terms. In 2002 approximately 5% of the Swedish children attended private schools, and the share is rapidly increasing. Public school efficiency is estimated using Data Envelopment Analysis (DEA). Modelling education is difficult since educational production is not only dependent on factors under control of the school management, but also on the students' socio-economic backgrounds. A number of approaches have been proposed concerning how to model this in a DEA setting. In this study, four different approaches are used and compared. Special focus is put on a second stage regression, where the efficiency estimates are regressed on competition and other explanatory variables. We can not show that the share of children attending private schools is related to public school performance.