Modeling Blank Data Entries in Data Envelopment Analysis

We show how Data Envelopment Analysis (DEA) can handle missing data. When blank data entries are coded by appropriate dummy values, the DEA model automatically excludes the missing data from the analysis. We extend this result to weight-restricted DEA models by presenting a simple modification to the usual weight restrictions, which automatically relaxes the weight restriction in case of missing data. Our approach is illustrated by a case study, describing an application to international sustainable development indices.