Corporate bankruptcy risk prediction in Vietnam: An application with LASSO method

Abstract

Corporate bankruptcy risk prediction has important implications to the corporate owners, lenders,
investors and regulators in their supervision, decision makings, which provides early warning indicators to
the firm’s financial strength. Several statistical and machine-learning based models have been developed to
predict the corporate bankruptcy risks, however, the performance of these models largely depends on the
arguably choice of the predictors. In this study, we examine the potentials of the popular variable selection
method, namely LASSO (Least Absolute Shrinkage and Selection Operator) to improve the predicting ability
of the corporate bankruptcy risks in Vietnam. Using data sample from 284 Vietnamese companies in period
2017- 2019, our study shows that the use of the LASSO technique to ex-ante select suitable predictors
significantly improve the forecasting power of the prediction models, especially for the machine-learning
based models in correctly identifying bankrupted firms in the testing sample