Bankruptcy prediction of privately held SMEs using feature selection methods

We test alternative feature selection methods for bankruptcy prediction and illustrate their superiority versus popular models used in the literature. We apply these methods to a comprehensive dataset of more than 1.8 million financial statements covering the entire universe of privately held Norwegian SMEs in 2006–2020. We find that input variables chosen by an embedded least absolute shrinkage and selection operator (LASSO) method yield the best in-sample fit, out-of-sample performance, and stability, robust across different time periods and estimation techniques. In a real-world simulation of a competitive credit market, even small differences in model performance translate into large differences in bank profitability, with LASSO outperforming all alternatives. Finally, contrasting bankruptcy predictors for SMEs with those for larger firms reveals economically meaningful differences consistent with theory: leverage and liquidity dominate for SMEs while profitability matters more for larger firms, reflecting SMEs’ higher refinancing risk and limited access to external financing. Predictors tailored specifically to SMEs yield superior prediction performance and higher bank profitability than those derived from larger firms.