Predicting firm bankruptcy using macroeconomic and uncertainty variables: an ensemble machine learning study of the French market

This study investigates whether adding macroeconomic variables and uncertainty indices improves the performance of ensemble bankruptcy prediction models for France's manufacturing and construction sectors. Starting from a baseline using only financial ratios, we estimate separate specifications, each adding one input: macroeconomic variables, the French Economic Policy Uncertainty index, the French Geopolitical Risk index, or a new French Google Trends-based uncertainty index. The results show that incorporating macroeconomic variables significantly improves out-of-sample predictive performance. For the uncertainty measures, each index delivers incremental improvements in accuracy relative to the ratios-only baseline. Notably, the Google Trends-based index yields gains comparable to those from the macroeconomic set, positioning this search engine-based measure as a promising predictor of bankruptcy risk. These insights offer practical value for corporate boards, financial analysts, lenders, and policymakers seeking to strengthen bankruptcy risk assessment during periods of elevated economic and geopolitical uncertainty.