The objective of an Early Warning system is to anticipate the signs of deterioration of the most fragile loans: all exposures that do not yet present objective evidence of deterioration, but which are close to a significant increase in risk, must enter the Early Warning scope.
NON-TRADITIONAL DATA: CONTRIBUTION TO EARLY WARNING
- The state-of-the-art of Early Warning models in banks uses so-called "traditional" information sets, e.g., quick and easy data sources with a low impact in terms of costs and resources (e.g., customer behavior)
- Over the years, CRIF has gained significant experience in evaluating the contribution to Early Warning systems of "non-traditional" data sources, in particular Credit Bureau data and Current Account data in its most granular form (individual transactions), appropriately categorized
- "Non-traditional" data sources allow the already good performance of models based only on "traditional" data to be improved. Overall, there was an increase of up to 10 points (GINI Index) thanks to alternative data sources