This is a technical note on the estimation of Stochastic Frontiers, based on the specifications of Battese and Coelli (1992) and Cornwell and Schmidt (1990). We focus on the functional form that one should use to estimate technical efficiency, cost and profit of the Chilean banking system. Three functional forms are specified: Fourier Flexible, Translog and Cobb Douglas. The LR and Fisher tests show that the frontier of the Chilean banking sector should be estimated by a Fourier Flexible. The results also show that Translog and Cobb-Douglas could underestimate bank efficiency. Moreover, the tests reject the hypothesis of efficiency persistence, that is to say efficiency does not remain constant over time. Finally, Hausman's test do not reveal significant differences between the models of Stochastic Frontiers and Fixed Effect. This means that the distributional assumptions and that of absence of correlation between the exogenous variables and efficiency imposed by Stochastic Frontier are not rejected.
Vergara, M. (2006). Technical note for estimating stochastic frontiers: an application to Chilean banking. Estudios De Administración, 13(2), 47–66. https://doi.org/10.5354/0719-0816.2006.56453
