Review and Implementation of Credit Risk Models of the Financial Sector Assessment Program (FSAP)

IMF Working Papers, May 1, 2006

Source details

Canonical URL

Review and Implementation of Credit Risk Models of the Financial Sector Assessment Program (FSAP)

Other formats

• Markdown version
• Structured JSON version
• Bundle manifest

Bibliographic details

• Authors: Kexue Liu, Jean Salvati, Renzo G Avesani, Alin T Mirestean
• Published: May 1, 2006
• Series: IMF Working Papers
• DOI: https://doi.org/10.5089/9781451863949.001

Summary and purpose

• Presents the basic Credit Risk+ model and proposes modifications.
• Positions the model as a benchmark for credit risk evaluations in stress-testing financial sector assessments.
• Outlines a progression from simple to fully-fledged model specifications to illustrate methodological development and applicability.

Model specifications and methodological progression

• Begins with the setting and basic definitions common to all model specifications used in the paper.
• Simple model:
  • Based on Bernoulli-distributed default events.
  • Assumes known default probabilities.
• Full Credit Risk+ implementation:
  • Based on the Poisson approximation.
  • Incorporates uncertain default probabilities determined by mutually independent risk factors.
• Extension:
  • Presents a Credit Risk+ specification with correlated risk factors as in Giese (2003).

Illustrations and empirical application

• Uses a specific portfolio of obligors to illustrate:
  • Characteristics of different models.
  • Results obtained from the Bernoulli-based model, the Poisson-approximation Credit Risk+ implementation, and the correlated risk-factor extension.

Key findings and implications for FSAP stress-testing

• Credit Risk+ and its variants can serve as useful benchmarks in stress-testing exercises for the financial sector.
• Modeling choices (Bernoulli vs. Poisson approximation; independent vs. correlated risk factors) materially affect loss distribution characteristics and stress-test outcomes.
• Introducing uncertainty in default probabilities and allowing for factor correlation (per Giese (2003)) provides richer risk dynamics relevant for assessment exercises.

Content in this bundle

• 8. Model 4: Correlated Sectors, Inter-Sector Covariance = 0.2
  • 8. Model 4: Correlated Sectors, Inter-Sector Covariance = 0.2 (Markdown version)
  • 8. Model 4: Correlated Sectors, Inter-Sector Covariance = 0.2 (PDF)