An empirical analysis of private SMEs' insolvency in Thailand using machine learning

Abstract

We investigate the insolvency of private Thai SMEs in the manufacturing, trading, and service sectors from 2017 to 2021. We model insolvency as a function of industry-relative financial ratios, firm characteristics, and local economic conditions using Least Absolute Shrinkage and Selection Operator (LASSO) logistic regression. The analysis shows the significant influence of financial ratios on the probability of insolvency for all sectors, particularly inventory turnover, accounts payable turnover, assets to equity, and debt to assets ratio. The service sector shows a unique positive effect of working capital to total assets on insolvency risk, implying that firms with high current assets or very low current liabilities are more prone to insolvency. Medium-sized firms, those registered as juristic ordinary partnerships, owned by foreigners, and located in less competitive areas are less likely to face insolvency.