Using panel data analysis to estimate confidence intervals for the DEA efficiency of individual decision making units

This article demonstrates a methodology using panel data analysis to estimate confidence intervals for the data envelopment analysis efficiency of individual decision making units (DMUs), and the statistical significance of trends in individual DMU efficiency. The procedure accounts for stochastic variations of the inputs and outputs of the target DMU as well as stochastic variations of the inputs and outputs of its efficient benchmark peers. The procedure is demonstrated using 9 years of data from 34 Canadian paratransit agencies.