The Default Risk of Firms Examined with Smooth Support Vector Machines

In the era of Basel II a powerful tool for bankruptcy prognosis isvital for banks. The tool must be precise but also easily adaptable tothe bank's objections regarding the relation of false acceptances (TypeI error) and false rejections (Type II error). We explore the suitabil-ity of Smooth Support Vector Machines (SSVM), and investigate howimportant factors such as selection of appropriate accounting ratios(predictors), length of training period and structure of the trainingsample in°uence the precision of prediction. Furthermore we showthat oversampling can be employed to gear the tradeoff between errortypes. Finally, we illustrate graphically how different variants of SSVMcan be used jointly to support the decision task of loan officers.