EBA欧洲银行-B2-V.-de-Bruyckere-Discussion_12页_1mb

EBA Summary: Stress Testing the Credit Risk of Mortgage Loans

Core Content

This document presents a methodology for calculating the Loss Given Default (LGD) of a bank's mortgage portfolio under a stressed housing market scenario. The approach is based on the distribution of Loan-to-Value (LTV) ratios within the portfolio and derives a closed-form solution for the mean portfolio LGD. The model is designed to assist risk managers, rating agencies, and regulators in benchmarking analyses and assessing credit risk under adverse conditions.

Main Points

• Methodology for Stress Testing: The study introduces a method to estimate LGD under stressed scenarios, considering the LTV distribution of the mortgage portfolio.
• Assumptions:
  • The LTV ratio is assumed to follow a Beta distribution.
  • The model assumes a fixed recovery rate (RR) in the stress test scenario.
• Use of Beta Distribution:
  • The Beta distribution is used to model LTV ratios due to its flexibility in representing a wide range of distributions.
  • A truncated Beta distribution may be more realistic in certain contexts.
• Benchmarking Rule-of-Thumb:
  • The paper provides a "rule-of-thumb" for benchmarking, which can be used by risk managers and regulators to assess the consistency of LGD estimates.
• Regulatory Context:
  • The EBA has a mandate to develop regulatory technical standards (RTS) for the assessment of mortgage lending value under the Capital Requirements Regulation (CRR).
  • Article 164(6) CRR allows competent authorities to set higher minimum LGD values based on specific conditions.
  • Article 181(3)(a) CRR requires RTS for specifying the nature, severity, and duration of an economic downturn.

Key Information

LTV Distribution and LGD Calculation

• The methodology assumes a Beta distribution for LTV ratios.
• It derives a closed-form solution for the mean portfolio LGD, which can be used for benchmarking purposes.
• The model is applicable to a stress test scenario where housing prices are expected to drop.

Empirical Results and Limitations

• The study includes data from European countries (AT, BE, CZ, DE, DK, ES, FI, FR, IE, IT, LU, NL, NO, PL, PT, SE, SK, UK) based on EBA data collection (December 2012).
• The table presents the minimum, maximum, and EAD-weighted average values for various drill-down variables, including:
  • LTVO: Loan-to-Value (with prior liens)
  • ILTV: Internal Loan-to-Value
  • DTSO: Debt-to-Value (with prior liens)
  • LTIO: Loan-to-Value (with internal models)
  • CRMO: Credit Risk Mitigant Over Mortgages
• The note clarifies that country-weighted averages are based on bucket medians, with the upper bound used for the lowest bucket. For exposures above the latest bucket, specific thresholds are applied for different variables.

Recovery Rate and Valuation Methods

• The valuation method (Market Value or Mortgage Lending Value) affects the recovery rate (RR).
• If the property is valued based on MLV, the RR is typically higher, whereas if based on MV, RR varies with market conditions.
• Germany has a strong regulatory framework for MLV, as per Pfandbriefregulation.

Use for Benchmarking Analysis

• The document highlights that the model should be used cautiously for benchmarking analysis due to variations in the definition of LTV across European banks.
• Different "value" concepts are considered, such as market value, purchase price, and internal models, which may influence the calculation of LTV and, consequently, LGD.

Suggestions for Improvement

• Empirical Validation:
  • The model should be validated against empirical data to assess its accuracy.
  • Comparison with loan-level results is recommended.
• Bayesian Approach:
  • Given the lack of data in IRB modelling, the model could be adapted into a Bayesian variant to incorporate prior information from data pooling.
• Model Extensions:
  • The model could be extended to include other variables such as Recovery Rate (RR), Loan-to-Income (LTI), Debt-to-Service-to-Value (DSTI), etc.
• National Policies:
  • The effect of national macro-prudential policies on LGD is suggested for further analysis.
  • Examples include:
    • UK: Limits the number of mortgage loans at or above 4.5 times LTI to no more than 15% of total mortgage loans.
    • Ireland: Limits new owner-occupied and buy-to-let (BTL) lending, with restrictions on LTVs above 80% and LTIs above 3.5x.

Conclusion

The document provides a theoretical framework for stress testing mortgage portfolios by considering the distribution of LTV ratios and deriving a closed-form solution for LGD. It emphasizes the importance of understanding valuation methods and their impact on recovery rates, and suggests empirical validation and potential model enhancements to improve accuracy and applicability.