Accounting Choices in Data Envelopment Analysis

Abstract Due to the increasing availability of large-scale digitalized databases containing summarised financial accounting measures, a growing number of DEA models are using these variables. Where only accounting measures are used, we term these "FinDEA" models. Accounting measures are subject to accounting choices regarding recognition and measurement, which means that they are not equivalent to underlying physical measures used in conventional DEA. This paper investigates the impact of accounting choices on FinDEA results when using alternative accounting measures of capital. Using both simulated and real-world data, we find that accounting choices impact FinDEA results, with the magnitude influenced by heterogeneity in accounting choices and sample sizes. Our results suggest that the variation in accounting choices needs to be considered when designing DEA models using accounting measures.