You are here
Analysis of the Predictors of Default for Portuguese Firms
Ana I. Lacerda
Russ A. Moro
C14 - Semiparametric and Nonparametric Methods
C45 - Neural Networks and Related Topics
G33 - Bankruptcy; Liquidation
The paper presents an insolvency risk analysis of Portuguese companies with three techniques: logistic regression, discriminant analysis and support vector machines (SVM). It identifies the most critical predictors of default based on the accounting, employee and debt concentration data. A comparison of the three methods reveals a superiority of SVM. Non-financial information such as employee data and the debt concentration index appear to be strong predictors of default.