Measuring the technical efficiency trough maximun likelihood models: a comparison between alternative specifications applied to a university

Abstract

This paper presents an application of efficiency measurement stochastic techniques to a University Department scientific production. In the first part, we follow Lovell´s (1982) model to carry out a static analysis with the variable averages in the selected period. The model sensibility to the chosen adjusted production function is analysed considering the CES, Cobb-Douglas and Translog functions. Afterwards, the efficiency estimation for each one of the Departments in consideration is carried out by means of the efficiency term conditional expectancy on the global error. The second part studies the dynamic estimate, following Batesse and Coelli (1992, 1995), through different functional forms. In this case, different assumptions of efficiency evolution, for fixed and variable effects, are considered and estimated again for each production unit (Department) and period. To be able to consider non-lineal models and split residuals, maximum likelihood estimate is applied. Lastly, the efficiency indexes homogeneity is studied in both dynamic and static models.