Evaluating the Effect of Operational Conditions and Practices on Warehouse Performance

Analyzing warehouse performance across different environments is critical to improving overall productivity and reducing costs. Although two-stage DEA estimators have been shown to be statistically consistent, the finite sample bias of DEA in the first stage carries over to the secondstage regression, which causes bias in the estimated coefficients of the contextual variables. The bias is particularly severe when the contextual variables are correlated with inputs. To address this shortcoming, we apply insights from Johnson and Kuosmanen (2010), who demonstrate that DEA can be formulated as a constrained special case of the Convex Nonparametric Least Squares (CNLS) regression to develop a new semiparametric one-stage estimator. The new model is applied to a set of warehouses to illustrate its performance.