Predicting bankruptcy in European e-commerce sector

In the current competitive and uncertain e-commerce environment, businesses have the need to predict in advance their likelihood of falling into bankruptcy. The central focus of this paper is to statistically model through different approaches the bankruptcy probability of e-commerce companies in Europe. The authors examine the econometric techniques twostep cluster, logistic regression, discriminant analysis, data mining tree, and roc curves to correctly classify these companies into bankrupt and not bankrupt. The paper finds evidences about the current credit underwriting inexperience among several financial institutions. The classification approaches included in this paper may be applied in real working practice whether by credit underwriters or by business decision makers. The research was developed using financial and accounting information available in the Bureau Van Dijk database. The paper suggests further analytical developments in the field of predictive bankruptcies, and recommends improvements on the credit evaluation scorecards such as the inclusion of advanced online metrics to increase the accuracy of the creditworthiness evaluation of an e-commerce company.