Issues and Solutions in the Statistical Analysis of the Occurrence of Bankruptcy of Small and Medium Enterprises

After Basel II, Credit Scoring techniques have assumed increasing importance and active research is taking place in this field. The availability of data and their accuracy always play an important role and became crucial when it must be decided which technique, among the several available alternatives, should be used. Our research interests focus on predicting the probability of default of Small and Medium Enterprises mainly using balance sheet indicators. In this paper, we present a review of statistical techniques, software solutions (R and SAS) and problems that may arise with retrospective data collection or data that also include the time at which a firm changed its status (time to the first event). Topics include solutions available when the event object of study is rare (unbalanced data), how to treat time-varying covariates, and comparing and testing models