Predicting technical effciency in stochastic production frontier models in the presence of misspecification: a Monte-Carlo analysis

Abstract

This paper provides a theoretical explanation for the sensitivity of technical efficiency measures to the choice of functional specification in stochastic production frontier models. It is shown that inappropriate functional specifications translate into a misspecification in the conditional mean of the stochastic frontier regression model. This misspecification, in turn, results in estimates of technical efficiency, confidence intervals and production elasticities being biased, even asymptotically. Monte-Carlo simulations reveal that the severity of the bias depends on the functional specification and the percentage contribution of the variance of technical inefficiency to the total variance of the composed errors.

Estimating technical efficiency of Australian dairy farms using alternative frontier methodologies

“…It could be argued that comparisons, when the DGP is unknown, are uninteresting because parametric stochastic frontiers and DEA simply incorporate different assumptions regarding the underlying DGP. By contrast, Monte Carlo studies such as Gong and Sickles (1992), Giannakas et al (2003) and Sickles (2005) can cast light on the performance of different methods under alternative DGPs. Research aimed at identifying the correct DGP and, therefore, the correct choice of method is obviously valuable.…”

Annual and cropping season environmental production conditions effects on smallholder technical efficiency in sub‐Saharan Africa: Evidence from Ethiopia

“…Additionally, employing the wrong functional form will distort the efficiency scores (Gong & Sickles, ). Giannakas, Tran, and Tzouvelekas (, b) also argue that the choice of functional form does not only affect the predicted scores but also leads to different conclusions regarding the sources of inefficiencies. In contrast, Guermat and Hadri () assert that model misspecification is not a major problem in SFA if the researcher is only interested in the mean technical efficiency.…”

Institutions and Economic Development

“…Secondly , Coelli, et al (1998) realise that when using the stochastic frontier approach, the specification of the functional form of the production function matters for the results. Monte Carlo simulation results from Giannakas, Tran, and Tzouvelekas (2003) indicate that the bias in the mean efficiency measures from stochastic frontier methods due to misspecification of functional form is sizeable. It can suggest a high level of inefficiency (10-30%) of output for the most efficient producers.…”