The Logistic Lasso and Ridge Regression in Predicting Corporate Failure

Abstract The prediction of corporate bankruptcy is a phenomenon of interest to investors, creditors, borrowing firms, and governments alike. Many quantitative methods and distinct variable selection techniques have been employed to develop empirical models for predicting corporate bankruptcy. For the present study the lasso and ridge approaches were undertaken, since they deal well with multicolinearity and display the ideal properties to minimize the numerical instability that may occur due to overfitting. The models were employed to a dataset of 2032 non-bankrupt firms and 401 bankrupt firms belonging to the hospitality industry, over the period 2010-2012. The results showed that the lasso and ridge models tend to favor the category of the dependent variable that appears with heavier weight in the training set, when compared to the stepwise methods implemented in SPSS.