The Use of Panel Quantile Regression for Efficiency Measurement: Insights from Monte Carlo Simulations
Audrey Laporte, Adrian Rohit Dass
In panel stochastic frontier models, the Fixed Effects (FE) approach produces biased technical efficiency scores when time-invariant variables are important in the production process, and the Random Effects (RE) approach imposes distributional assumptions about the inefficiency. Moreover, technical efficiency scores obtained from these models are biased when the sample contains a large number of firms near the efficient frontier. We propose the use of quantile regression (QR) with a Correlated Random Effects (CRE) specification as an alternative to these approaches. Using Monte Carlo simulations, we show that CRE QR can overcome the limitations of FE and RE stochastic frontier models.
technical efficiency, quantile regression, panel data, stochastic frontier analysis