Developing a statistical model to assess the credit risk of financial institutions is a formidable challenge mainly because:
- Financial institutions tend to be highly heterogeneous from a credit risk point of view
- The sector exhibits low default frequency
- The sector’s default frequency is volatile over time
Therefore it is common belief that the assessment of credit risk for such companies can only be conducted using an expert-judgement framework, such as that employed by rating analysts, or a scorecard methodology that is usually inspired by and/or benchmarked with actual ratings.
Expert-judgement approaches are usually very successful in quantifying counterparty credit risk, but suffer from inherent operational limitations. Ratings tend to cover only a limited number of financial institutions; scorecards require a significant amount of time and resources for the generation of a single assessment and each counterparty needs to be assessed individually.
A statistical model, that combines the advantages of an expert-judgement approach driven by ratings with an automated engine, was not available up to now, but is highly desirable in order to:
- Assess the credit risk of financial institutions, expanding the universe of scored companies beyond what is normally covered by rating analysts and
- Accelerate and scale the credit assessment process
At S&P Capital IQ, we have managed to bridge this gap by developing a cutting-edge statistical model that is trained on Standard & Poor’s ratings and uses company financials, macroeconomic and industry-specific factors to generate a letter-grade credit score, for public and private banks and insurance companies, globally. Scores are reported in lower-case, to avoid confusion with actual Standard & Poor’s ratings, as they represent a purely statistical view of the credit strength of a financial institution.