?
Specification of a stochastic production function model in the extended class of stochastic frontier models
Copulas are being successfully applied for derivation of estimates in the models related to stochastic production functions. They can be used for handling of panel data, for the analysis of models with multiple outputs and for improvement of estimates in classical models. The research proposes an algorithm for specification of extended class of models for stochastic production functions where a possible dependence between the error components is assumed. To describe this dependence we consider two functions: normal copula and Frank copula. Simulated data are used to prove the necessity to take into account potential dependence between the error components and to illustrate the importance of considering several types of copulas for different problems related to estimation of technical efficiency. In addition we analyze an influence of the choice of copula type on estimates of main parameters in the model and propose possible problems where classical models for stochastic production function can be applied.