EBA欧洲银行-20140611-Fourth-interim-report-on-the-consistency-of-risk-weighted-asset_32页_1mb

Summary of the Fourth Report on the Consistency of Risk Weighted Assets in Residential Mortgage Portfolios

Core Content

This report is the second phase of an analysis on the consistency of risk-weighted assets (RWA) in residential mortgage portfolios across the European Union (EU). It focuses on the role of drill-down variables—specifically Loan-to-Value at Origination (LTVO), Indexed Loan-to-Value (ILTV), Debt-to-Service at Origination (DTSO), Loan-to-Income at Origination (LTIO), and Credit Risk Mitigation (CRMO)—in explaining variations in risk weights (RW) across banks.

The study collects data on exposure at default (EAD) and average RWs by predefined buckets for each variable, and it also includes vintage-based and country-level analyses. The aim is to understand how these variables influence RWs and whether they can explain the differences found in the first phase of the analysis.

Main Findings

1. Use of Drill-Down Variables in Models

• ILTV and LTVO are the most commonly used variables in risk parameter estimation, with ILTV being used in 58% of cases.
• CRMO is also frequently used, especially in LGD models.
• DTSO and LTIO are used less frequently and are typically only included in PD models.
• Some banks use these variables indirectly in credit assessments, even if they are not formal inputs in their models.

2. Variability and Correlation of RWs

• ILTV showed the highest correlation (around 80%) and variability (standard deviation over 25% in about 65% of observations) with RWs.
• The correlation between RWs and the variables is generally strong, but RW sensitivity to these variables is not always a clear explanatory factor of variation.
• EAD distribution across buckets has little impact on RW disparities, suggesting that the level effect is more significant.

3. Influence of Drill-Down Variables on RWs

• ILTV is identified as the most significant variable influencing RW variation.
• The price effect (difference in RW levels from the benchmark) is more significant than the bucket mix effect (distribution of EAD across buckets).
• The level effect (residual RW after controlling for sensitivity) is more important than the sensitivity effect in the EU sample, with about 80% of observations showing a negative sensitivity effect and 70% of banks showing a positive level effect.

4. Country-Specific Considerations

• Country-specific credit risk mitigants (e.g., government guarantees) play a key role in explaining lower RW sensitivity to property values in some countries.
• In France, the Netherlands, and Belgium, the presence of such mitigants can lead to lower RWs for certain exposures, which may explain the observed lower sensitivities to LTV.
• DTSO is more relevant in Italy and Belgium, while LTVO is more significant in the Netherlands.

5. Vintage Analysis

• A strong correlation exists between LTVO and RW levels, suggesting that the origination vintage can influence the risk profile of loans.
• The portfolio composition by vintage also contributes to explaining RW variation, indicating that the mix of loans across different time periods may have a material impact on the overall RW.

6. Banking Group-Level Analysis

• The combination of variables used by banking groups does not consistently explain differences in RWs.
• Drill-down variables used in PD and LGD models do not appear to discriminate between banks in terms of RWs, possibly due to the indirect influence of these variables in credit assessment.

Key Variables and Their Definitions

Limitations and Notes

• The use of standalone variables does not allow for the identification of converging or distorting factors.
• Hypothetical scenarios and multi-profile facilities were not considered, limiting the depth of analysis.
• Predefined buckets may lack granularity, which could affect the accuracy of sensitivity analysis.
• Lack of PD/LGD data prevents direct investigation of how these variables influence specific parameters.
• Country-specific definitions and mitigation techniques can affect the interpretation of results.

Conclusion

• The price effect is the dominant factor in explaining RW variation, especially for ILTV.
• ILTV has the strongest influence on RWs, with a high correlation and variability.
• Country-specific mitigants and portfolio composition by vintage are important but not fully quantified.
• The use of variables in PD/LGD models does not directly impact RWs, but indirect credit filtering may play a role.
• The findings should be read in conjunction with the first phase report published in December 2013.

Annex Overview

• Annex 1: Country-weighted averages for the drill-down variables.
• Annex 2: Methodology details for the top-down analysis.
• Annex 3: Breakdown of the price and bucket mix effects for LTVO, DTSO, and LTIO.
• Annex 4: Average RWs by bucket for ILTV and LTVO.
• Annex 5: Application of the top-down approach at country level.