Overcoming Data Limitations in Nonparametric Benchmarking: Applying PCA-DEA to Natural Gas Transmission

Abstract

This paper provides an empirical demonstration for a practical approach of efficiency evaluation against the background of limited data availability in some regulated industries. Here, traditional DEA may result in a lack of discriminatory power when high numbers of variables but only limited observations are available. We apply PCA-DEA for radial efficiency measurement to US natural gas transmission companies in 2007. This allows us to reduce dimensions of the optimization problem while maintaining most of the variation in the original data. Our results suggest that the PCA-DEA methodology reduces the probability of over-estimation of the individual firm-specific performance. It also allows for a large number of original variables without substantially reducing the discriminatory power of the model.