Bankruptcy prediction for SMEs using relational data

Ellen Tobback, Julie Moeyersoms, Marija Stankova and David Martens

Working Papers from University of Antwerp, Faculty of Business and Economics

Abstract: Bankruptcy prediction has been a popular and challenging research area for decades. Most prediction models are built using traditional data such as financial gures, stock market data and firm specific variables. We complement such dense data with ne-grained data by including information on the company's directors and managers in the prediction models. This information is used to build a network between Belgian enterprises, where two companies are related if they share or have shared a director or high-level manager. We start from two possibly related assumptions: (i) if a company is linked to many (or only) bankrupt firms, it will have a higher probability of becoming bankrupt and (ii) the management has an inuence on the performance of the company and incompetent or fraudulent managers can lead a company into bankruptcy. The weighted-vote relational neighbour (wvRN) classier is applied on the created network and transforms the relationships between companies in bankruptcy prediction scores, thereby assuming that a company is more likely to file for bankruptcy if one of the related companies in its network has failed. The more related companies have failed, the higher the predicted probability of bankruptcy. The relational model is then benchmarked against a base model that contains only structured data such as financial ratios. Finally, an ensemble model is built that combines the relational model's output scores with the structured data. We find that this ensemble model outperforms the base model when detecting the riskiest firms, especially when predicting two-years ahead.

Pages: 17 pages
Date: 2016-04
New Economics Papers: this item is included in nep-ent, nep-rmg and nep-sbm
References: View references in EconPapers View complete reference list from CitEc
Citations: View citations in EconPapers (2)

Downloads: (external link)
https://repository.uantwerpen.be/docman/irua/5d7fdf/131871.pdf (application/pdf)

Related works:
This item may be available elsewhere in EconPapers: Search for items with the same title.

Export reference: BibTeX RIS (EndNote, ProCite, RefMan) HTML/Text

Persistent link: https://EconPapers.repec.org/RePEc:ant:wpaper:2016004

Access Statistics for this paper

More papers in Working Papers from University of Antwerp, Faculty of Business and Economics Contact information at EDIRC.
Bibliographic data for series maintained by Joeri Nys ().