Multivariate ordinal regression models for enhanced credit risk modeling

  • Vana Gür, Laura (PI - Project head)
  • Hirk, Rainer (Researcher)

Project Details

Financing body

Oesterreichische Nationalbank (Jubiläumsfonds)


When financial intermediaries assess the credit risk of their counterparties they either rely on models based on a historical data base of actual defaults, or on third-party information, i.e., credit ratings provided by credit rating agencies. The modeling of default data and credit rating data is typically performed separately, even in cases where both are employed in the internal processes. However, recent regulatory frameworks as
well as scientific studies have called for a combined approach to credit risk modeling, where all available information sources are integrated into the modeling framework. Also under IFRS 9 banks are required to consider all reasonable and supportable information, including forward-looking measures such as credit ratings. Statistical models tailored to the characteristics of credit risk data together with efficient software implementations must be developed in order to address the pressing need of such an enhanced approach.
Effective start/end date1/10/2031/03/22