Copula-based Stochastic Frontier Model with Autocorrelated Inefficiency

Abstract

The paper considers the modeling and estimation of the stochastic frontier
model where the error components are assumed to be correlated and the
inefficiency error is assumed to be autocorrelated. The multivariate FarlieGumble-Morgenstern (FGM) and normal copula are used to capture both
the contemporaneous and the temporal dependence between, and among, the
noise and the inefficiency components. The intractable multiple integrals that
appear in the likelihood function of the model are evaluated using the Halton
sequence based Monte Carlo (MC) simulation technique. The consistency
and the asymptotic efficiency of the resulting simulated maximum likelihood
(SML) estimators of the present model parameters are established. Finally,
the application of model using the SML method to the real life US airline
data shows significant noise-inefficiency dependence and temporal dependence
of inefficiency.