Data Envelopment Analysis as Nonparametric Least Squares Regression

Data Envelopment Analysis (DEA), a nonparametric mathematical programming approach to productive efficiency analysis, envelops all observed data. In this paper we show that DEA can be interpreted as nonparametric least squares regression subject to shape constraints on frontier and sign constraints on residuals, and that classic parametric programming model is a constrained special case of DEA. Applying these insights, we present a nonparametric variant of the corrected ordinary least squares (COLS) method. We show that this new method, which we term corrected concave nonparametric least squares (C2NLS) is consistent and asymptotically unbiased. The linkages established in this paper contribute to further integration of the econometric and linear programming approaches