A dynamic stochastic frontier production model with time-varying efficiency

In this paper technical efficiency is introduced via the intercept that evolve over time as a AR(1) process in a stochastic frontier (SF) framework in a panel data framework. Following are the distinguishing features of the model. First, the model is dynamic in nature. Second, it can separate technical inefficiency from fixed firm-specific effects which are not part of inefficiency. Third, the model allows estimation of technical change separate from change in technical efficiency. It is proposed that the ML method be used estimate the parameters of the model. Finally, expressions are derived to calculate/predict technical inefficiency (efficiency).