Bankruptcy Prediction: Evidence from Vietnam

Abstract

This paper examines the best method to predict corporate bankruptcy in Vietnam. Using an unique sample of 300 Vietnamese firms from 2017 to 2019, we find that the intelligent approaches significantly outperform the classical logistic regression. In particular, the random forest method produces the best overall forecasts in all three evaluation criteria, namely the accuracy ratios, Type I and Type II errors. Moreover, using the adaptive lasso technique to select significant candidate variables prior to the estimation considerably improves the default forecasts. Finally, we find that earning power and financial leverage ratios contain the most significant information in bankruptcy prediction in Vietnam.