Title: The panel stochastic frontier model with endogenous inputs and correlated random components
Authors: Hung-pin Lai - National Chung Cheng University (Taiwan) [presenting]
Subal Kumbhakar - State University of New York at Binghamton (United States)
Abstract: The four-component stochastic frontier (4CSF) panel model, where the inputs are endogenous and correlated with the composite error, is considered. We assume that the correlation of inputs can be with one or more of the inefficiency components, or with all four random components in the production function. Furthermore, we allow correlation between the time-invariant and time-varying components. This correlation can arise in various ways. For example, the correlation can arise due to dependence between (i) the long- and short-run inefficiency components, (ii) firm-effects with short-run inefficiency, (iii) firm-effects and the noise term. We propose a three-step procedure to estimate the model parameters. In the first step, we use either within or difference transformation to eliminate the time invariant endogenous components. We use a previous approach to generate the instruments and obtain unbiased and consistent estimator the parameters in the frontier part, except the intercept. In the second step we use the maximum likelihood procedure to estimate the parameters associated with the distributions of the time-varying random components. In the third step, we estimate the intercept and the remaining parameters. We propose using copula approach to model the dependence between the time-varying and time-invariant components.