Data envelopment analysis (DEA) is compared to stochastic production function estimation (SPFE) in a noisy setting. The statistic of interest is the average efficiency estimator. Monte-Carlo simulations show that the mean squared error of the DEA-estimator even for considerable noise remains below the MSE of the SPFE analogue. A bootstrapping approach is designed to get some first-step statistical underpinning of this DEA average efficiency estimator. The coverage of the bootstrapping approximation to the distribution of this estimator is shown to be fairly good.
|Publication status||Published - 1994|