Computation and inference in semiparametric efficient estimation

This paper examines computational issues in semiparametric efficient estimation of stochastic panel distance functions. Our basic model is an extension of Hausman and Taylor (1981), Park and Simar (1995), Park et al. (1996), and allows for a subset of the regressors to be correlated with the random effects. We motivate issues of computation and inference by analyzing the productive efficiency of U.S. banks.