Abstract
Credit rating agencies’ (CRA) increasingly important role in financial markets and the real economy
has resulted in new regulatory requirements with respect to backtesting of their credit rating models.
The European Union and the European Securities and Market Authority (ESMA) have developed a
comprehensive set of statutory laws, including three basic attributes of a rating system that need to be
regularly monitored by credit rating agencies operating in the EU: Descriptiveness, calibration quality
and historical robustness. ESMA requires that CRAs establish internal review functions responsible
for validating the credit rating process. This paper provides a set of methodologies and statistical tools,
which when properly implemented enables a credit rating agency to become compliant with ESMA’s
requirements from a quantitative validation perspective. The methodologies and statistical tools are
implemented on two case studies which use credit rating data gathered from a Nordic CRA (NCR)
and a Norwegian savings and loan bank. The credit rating model employed by the bank displayed
both descriptiveness, satisfactory calibration quality and good historical robustness. The NCR data
were insufficient to draw any conclusions with regard to the credit rating process.