Analysis of dependence between the random components of a stochastic production function for the purpose of technical efficiency estimation

In elaboration of the stochastic frontier methodology we offer an approach to test a statistic hypothesis about independence of random components of a stochastic production function for the purpose of estimation of technical efficiency. We describe the dependence between the error components by a copula. For parameters estimation in the econometric model in case of dependent error components the analytical expressions for log-likelihood function and its derivatives are given. The results of an experimental hypothesis test based on simulated data with dependent error components are also provided. We use two approaches for the parameters estimation: statistical package Stata 10.0 under an assumption of independence of the error components and created in MS Excel macro which gives the possibility to analyze models with dependent error components. It is shown that using non-tested assumption of independence of the random components of a stochastic production function may lead to wrong results in estimation of the technical efficiency.