Semiparametric Approaches to Stochastic Panel Frontiers with Applications in the Banking Industry.

This article introduces new modeling and estimation methods designed to help mitigate problems of endogeneity and misspecification. The authors use an output distance function to model the technology of a multioutput firm. They also use results from B. U. Park and L. Simar (1995) and B. U. Park, R. C. Sickles, and L. Simar (1998) on semiparametric efficient estimation of panel models to estimate multioutput stochastic distance frontiers for which the distribution of the effects and a subgroup of regressors is not specified. Furthermore, the authors introduce a new semiparametric method that makes minimal assumptions on the functional form of inputs in the distance function.