Semiparametric approaches to stochastic panel frontiers with applications in the banking industry

This article introduces new modeling and estimation methods designed to hel p mitigate problems of endogeneity and misspecification. We use an output d istance function to model the technology of a multioutput firm. We also use results from Park and Simar and Park, Sickles, and Simar on semiparametric efficient estimation of panel models to estimate multioutput stochastic di stance frontiers, for which the distribution of the effects and a subgroup of regressors is not specified. Furthermore, we introduce a new semiparamet ric method that makes minimal assumptions on the functional form of inputs in the distance function.