Bias in Transport Efficiency Estimates Caused by Misspecified DEA Models

One of the assumptions of Data Envelopment Analysis (DEA) in transportation studies is the substitutability/transformation of inputs/outputs. In this paper we examine three transport modes that frequently have employed DEA to measure technical efficiency. We ascertain whether input substitutions and output transformations are present in the transport data, and the impact on transport DEA scores when substitutions or transformations are not present. We then propose methods for correcting substitution and transformation specification errors. Future transport DEA studies should test for substitution and transformation in their data, to avoid biased DEA scores and spurious second-stage regression results.