Financial ratios as indicators in bankruptcy prediction: A comparative analysis of statistical and machine learning models

This paper investigates the optimal approach for predicting corporate bankruptcy risk within the context of Vietnam based on financial ratios. As a unique dataset of listed Vietnamese firms from 2010 to 2021 is employed, we confirm that machine learning models for bankruptcy prediction significantly surpass the traditional logistic regression. In addition, our dataset is divided into two subsets for training and testing models with proportions of 75% and 25%, respectively. The results demonstrate that the XGBoost and Random Forest techniques are superior to K-Nearest Neighbor and Logistic Regression in forecasting failure in both periods. Notably, our paper reveals that the predictive performance was slightly decreased compared to the two periods, and the forecasting after one year is higher than two years ahead.

本文针对越南市场情境下基于财务比率的企业破产风险预测最优方法展开研究。本文通过采用2010至2021年越南上市公司的独有数据集，证实用于企业破产风险预测的机器学习模型性能显著优于传统逻辑回归（Logistic Regression）模型。此外，本数据集按75%与25%的比例划分为训练子集与测试子集，分别用于模型的训练与测试。研究结果表明，在两个预测周期内，XGBoost与随机森林（Random Forest）算法的破产预测性能均优于K近邻（K-Nearest Neighbor）与逻辑回归模型。值得注意的是，本文发现模型的预测性能相较于两个预测时段略有下降，且提前一年的预测精度高于提前两年的预测精度。