Inverted VEA for Worst-Practice Benchmarking: With an Application to Distress Prediction of European Banks

Giannis Karagiannis, Stavros Kourtzidis, Panagiotis Ravanos

Research output: Contribution to conferencePaperpeer-review


In this paper we introduce managerial preferences in the assessment of worst-practices of Decision Making Units (DMUs), by means of Value Efficiency Analysis (VEA). Our model involves the selection of a DMU residing on the Data Envelopment Analysis (DEA) worst-practice frontier, that has the least desirable input/output structure by view of a Decision Maker (DM). The method then assesses all DMUs based on the worst favorable sets of input/output multipliers for the chosen DMU. The results of the associated linear program, referred to as Inverted VEA, are higher or equal to the respective Inverted DEA scores. Higher (lower) differences between Inverted DEA and Inverted VEA scores highlight an input/output structure that is farther (closer) to the least desirable one. This aids central management to identify DMUs which should be marked for closer monitoring and inspection or put through a restructuring process. We illustrate the usefulness of the method by applying it to assess the relative financial distress of 33 major European banks that were evaluated by the European Banking Authority in the 2018 stress test. The results provide fruitful insights to regulatory authorities regarding the prioritization of their on-site financial inspections, thus aiding towards the better management of scarce time and resources.
Original languageEnglish
Publication statusPublished - Jun 2021
Event12th North American Productivity Workshop - Online, Miami, United States
Duration: 7 Jun 202111 Jun 2021


Conference12th North American Productivity Workshop
Abbreviated titleNAPW XII Virtual
Country/TerritoryUnited States
Internet address


Dive into the research topics of 'Inverted VEA for Worst-Practice Benchmarking: With an Application to Distress Prediction of European Banks'. Together they form a unique fingerprint.

Cite this