The Bank of Italy published a research paper on the “expert assessment” overlay in its in-house credit assessment system for Italian non-financial firms (ICAS), setting out recent methodological enhancements and estimating how much analysts’ judgement contributes to rating quality beyond the statistical model. The study describes how two analysts review and adjust a model-based rating, including through new elements such as climate-related risk integration and sector analysis. Using more than 25,000 analyst assessments from 2016 to 2022, including corporate default events, it finds the expert assessment materially improves the predictive and discriminatory power of ICAS ratings versus model-only outputs, increasing the area under the receiver operating characteristic curve (AUROC) by around 2 percentage points. It also reports that the expert overlay helps preserve ICAS performance during periods of macroeconomic stress.