Estimating a mixture of Stochastic Production Frontier: Technology gap and technical efficiency measures

Abstract

This study aims to assess the ignorance impact of the technological heterogeneity on the technical efficiency measurement, using stochastic production frontier. We proposed a methodology based on a switching regression model in order to take account of the technology differences in the frontier specification. Indeed, we extend the switching model developed by Caudill (2003) for the panel case in two ways.First, we incorporate technical inefficiency in the model. Then, we consider that each firm has a specific probability to use one of the two technologies, in the industry. Next, we construct two technology homogenous groups. Finally, we estimate two production frontiers for each group of technologically homogeneous firms in order to estimate „net‟ technical inefficiency scores. This methodology was applied to a panel of French textile firms (1993-1997). The empirical results show that the technical efficiency is very low, 45% on average, under the homogeneous technology hypothesis.However, the average technical efficiency is about 80% with the extended switching regression model. Thus, if we ignore the technology heterogeneity, we underestimate the technical efficiency of approximately 30% on average