The Prediction of Default with Outliers: Robust Logistic Regression

Abstract

This paper suggests a Robust Logit method, which extends the conventional logit model by taking outliers into account, to implement forecast of defaulted firms. We employ five validation tests to assess the in-sample and out-of-sample forecast performances, respectively. With respect to in-sample forecasts, our Robust Logit method is substantially superior to the logit method when employing all validation tools. With respect to the out-of-sample forecasts, the superiority of Robust Logit is less pronounced.