Bank of Italy paper finds social and governance indicators modestly improve ICAS default prediction, with strongest gains for micro and small firms

The Bank of Italy has published a paper examining whether social and corporate governance indicators can improve default-risk assessment for Italian non-financial corporations within its In-house Credit Assessment System, or ICAS. The study finds that social and governance indicators improve default prediction, but only moderately overall. Both sets of indicators are statistically significant for micro and small firms, while for medium-sized and large firms only social indicators remain statistically significant. The analysis uses social indicators derived from Italian National Social Security Institute data and governance indicators drawn from a structured database on Italian firms' ownership and governance links. When these measures were added to the baseline statistical ICAS model, the augmented model's out-of-sample discriminatory power improved by 0.7 to 0.8 percentage points in AuROC for micro firms, 0.5 to 0.9 points for small firms, 0.6 to 0.8 points for medium-sized firms, and 0.4 points for large firms. The paper identifies workforce-related measures such as layoff rates, employee turnover, fixed-term contract rates and wages as the most useful social predictors, while governance variables had predictive value mainly for smaller firms and were not significant in the final integrated model for medium-sized and large firms. The paper also notes that governance-related factors for large firms will continue to be considered through ICAS's expert assessment by credit analysts. It adds that, in the coming years, statistical ratings from models such as S-ICAS will be used directly in the Eurosystem Credit Assessment Framework for micro, small and medium-sized firms, and that any integration of social and governance indicators into that framework would require further estimation and validation.