Using nonparametric frontiers for technical efficiency evaluation: the case of Portuguese public hospitals

Abstract

Objective: Portuguese public hospitals have exhibited an improvement of resources management in the past few years. However, only few analyses over their performance have been done. This study aims at find technical efficiency levels of Portuguese public hospitals using robust methods. Methods: This study has considered two different nonparametric alternatives for efficiency assessment. These methods are based on the Data Envelopment Analysis and on the Order-a frameworks. Four main variable categories were used: total expenses, risk-adjusted treated patients, delivered healthcare services’ quality, and the epidemiologic and demographic conditions in which hospitals operate. A sample of 27 Portuguese public hospitals and hospital centres was used. Data regards the year of 2016. Results: Inefficient hospitals exhibit average technical inefficiency levels of about 10%, which corresponds to the waste of nearly €617 million. Conclusions: Inefficiency levels have a strong dependence on model features. Hence, performance evaluation should account for the existence of potential outliers, the number of variables, the existence of external factor affecting the hospitals’ production process, as well as the characteristics of the achieved frontiers. An important aspect is the assumption of convexity of the frontier. A better solution for hospital performance assessment is, then, the Order-a method.