An out-of-sample evaluation framework for DEA with application in bankruptcy prediction

Abstract

Nowadays, data envelopment analysis (DEA) is a well-established non-parametric methodology for performance evaluation and benchmarking. DEA has witnessed a widespread use in many application areas since the publication of the seminal paper by Charnes, Cooper and Rhodes in 1978. However, to the best of our knowledge, no published work formally addressed out-of-sample evaluation in DEA. In this paper, we fill this gap by proposing a framework for the out-of-sample evaluation of decision making units. We tested the performance of the proposed framework in risk assessment and bankruptcy prediction of companies listed on the London Stock Exchange. Numerical results demonstrate that the proposed out-of-sample evaluation framework for DEA is capable of delivering an outstanding performance and thus opens a new avenue for research and applications in risk modelling and analysis using DEA as a non-parametric frontier-based classifier and makes DEA a real contender in industry applications in banking and investment.

Multi-criteria ranking of corporate distress prediction models: empirical evaluation and methodological contributions

“…According to the Basel Committee on Banking Supervision (BCBS), default in credit risk refers to a failure of a borrower or counterparty to meet its obligations in accordance with agreed terms (Basel Committee on Banking Supervision, 2000, p. 1). Corporate credit studies have used the same parametric and non-parametric frameworks to predict business failure events such as credit default (e.g., Beaver, 1996), bankruptcy (e.g., Barboza et al, 2017;Ouenniche and Tone, 2017;Liang et al, 2016, Kim et al, 2016, Bauer and Agarwal, 2014Tinoco and Wilson, 2013;Zhou, 2013;Hillegeist et al, 2004;Shumway, 2001;Wilson and Sharda, 1994;Ohlson, 1980), financial distress (e.g., Altman et al, 2017;Li et al, 2014Li et al, , 2017Sun et al, 2017;Zhou et al, 2015;Geng and Chen, 2015;Campbell et al, 2008;Bandyopadhyay, 2006), insolvency (e.g., Callejón et al, 2013;Jackson and Wood, 2013), and loan default (e.g., Jiang et al, 2017;Kou et al, 2014;Bhimani and Gulamhussen, 2013). Amongst the above-mentioned failure events, bankruptcy and distress events have been the subject of many prediction studies.…”

“…From a statistical point of view, a failure prediction model (FPM) is a typical classification problem, which uses the selected features; say accounting, market, and macroeconomic-based information, to classify firms into distress or non-distress categories or classes. During the last decades, numerous studies have employed different types of techniques from statistics, operational research (e.g., Li et al, 2017;Ouenniche and Tone, 2017;Avkiran and Cai, 2014;Li et al, 2014;Premachandra et al, 2011;Yeh et al, 2010;Premachandra et al, 2009), and artificial intelligence (e.g., Chen et al, 2016;Fethi and Pasiouras, 2010;Bahrammirzaee, 2010;Charalambous et al, 2000) fields to design new failure prediction models. Initial studies on failure prediction use statistical techniques such as univariate discriminant analysis (e.g., Beaver, 1966Beaver, , 1968, and multivariate discriminant analysis (e.g., Altman, 1968Altman, , 1973Altman, , 1983 as classification techniques.…”

A comparative analysis of two-stage distress prediction models

“…The less vulnerable classifers to the underlying statistical assumptions are the ones from the field of artificially intelligent and expert systems (AIES) such as recursively partitioned decision trees (e.g., Frydman, Altman, & Kao, 1985), case-based reasoning models (e.g., H. Li & Sun, 2009, neural networks (e.g., Du Jardin & Séverin, 2012; Kim & Kang, 2010), rough set theory (e.g., McKee & Lensberg, 2002;Yeh et al, 2010), genetic programming (e.g., Back, Laitinen, Sere, & Wezel, 1995;Alfaro-Cid, Sharman, & Esparcia-Alcazar, 2007;Etemadi, Anvary Rostamy, & Dehkordi, 2009), as well as the ones from field of operations research (OR), such as multi-criteria decision making analysis (MCDA) (e.g., Zopounidis & Doumpos, 2002) and Data Envelopment Analysis (DEA) (e.g., Sueyoshi & Goto, 2009;Sueyoshi, Goto, & Omi, 2010;Z. Li, Crook, & Andreeva, 2014, Ouenniche & Tone, 2017 for a detailed classification of failure prediction models, the reader is referred to Balcaen and Ooghe (2006), Aziz and Dar (2006), Bellovary, Giacomino, & Akers (2017), Baharammirzaee (2010), Abdou and Pointon (2011), and Chen, Ribeiro, & Chen (2016).…”

Development of a Stable Corporate Bankruptcy Classification Model: Evidence From Taiwan

“…Based on a sample of 22 companies, he shows how his proposed method improves forecasting accuracy of corporate bankruptcy. Ouenniche and Tone (2017) develop a framework based on data envelopment analysis (DEA) for risk assessment and bankruptcy prediction of companies listed on the London Stock Exchange, and conclude that DEA is a valuable tool for bankruptcy evaluation and benchmarking.…”