EBA欧洲银行-20140611-Fourth-interim-report-on-the-consistency-of-risk-weighted-asset_31页_1mb

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

Core Content

This report focuses on the drill-down analysis of residential mortgage portfolios to assess the consistency of risk-weighted assets (RWA) across EU banks. It aims to understand whether differences in RWA can be explained by the use of specific variables that describe the risk profile of the portfolios.

The analysis examines five key variables:

• Loan-to-Value at Origination (LTVO)
• Indexed Loan-to-Value (ILTV)
• Debt-to-Service at Origination (DTSO)
• Loan-to-Income at Origination (LTIO)
• Credit Risk Mitigation (CRMO)

The study uses data from 43 banks in the EBA sample, collected in December 2012, and includes both qualitative and quantitative analyses to explore the impact of these variables on RWA.

Main Points

1. Use of Variables in Risk Models

• ILTV is the most commonly used variable in PD/LGD estimations (58% of banks).
• LTVO and CRMO are the second and third most frequently used variables.
• LTIO is the least used, and DTSO is used less frequently, typically only in PD models.
• Some banks use variables indirectly in credit assessments, not as formal inputs in their models.

2. Variable Definitions and Variability

• Definitions vary significantly across banks and countries.
• LTVO has a core definition with numerator as loan disbursements and denominator as property market value, but many variants exist, such as including or excluding prior liens, charges, and other collateral.
• ILTV includes indexation methods (quarterly, semi-annually, or yearly) and stressed value calculations.
• DTSO and LTIO have high variability in data availability and definition, with many banks only recently starting to collect this information.

3. Correlation and Variability of RWA

• A strong correlation (above 80%) was found between RWA and all variables, with ILTV showing the highest correlation (around 65% of observations).
• ILTV also has the highest variability (standard deviation relative to the mean >25% in around 65% of observations).
• ILTV is identified as the most significant variable influencing RWA variation.

4. Price and Bucket Mix Effects

• The price effect is the main driver of RWA variation across the EU sample.
• The bucket mix effect has a minor impact on the average RWA at the portfolio level.
• The level effect is more important than the sensitivity effect, with both effects varying significantly between banks and countries.
• In about 80% of the observations, the sensitivity effect is negative, while 70% of banks have a positive level effect.

5. Country-Specific Patterns

• Country-specific credit risk mitigants (e.g., government guarantees, NHG in the Netherlands, 'Credit Logement' in France) can reduce RWA.
• These mitigants may explain the lower sensitivity to LTV observed in some countries.
• Country-level analysis can provide more specific insights, but the report focuses on EU-wide trends.

6. Vintage Analysis

• A close link exists between LTVO levels, RWA, and portfolio composition by vintage.
• Vintage appears to influence how RWA varies across the portfolio.

7. Limitations

• The standalone use of variables does not clarify whether differences in RWA are due to converging or distorting factors.
• The absence of a complementary approach (e.g., hypothetical facilities) limits the understanding of credit policy impacts.
• Predefined buckets across countries may lack granularity.
• No data on PD/LGD parameters was available, limiting the ability to determine the direct influence of variables on specific risk parameters.

8. Findings

• The direct use of drill-down variables in PD and LGD models does not significantly discriminate between banks in terms of RWA.
• ILTV is the most significant variable in explaining RWA variation.
• The price effect dominates over the bucket mix effect in influencing RWA.
• Country-specific credit risk mitigants can affect RWA, particularly in France, the Netherlands, and Belgium.

Key Information

• The EU benchmark RWA by variable is provided, showing varying risk weights across different buckets.
• Figure 5 illustrates the correlation and variability of RWA by variables.
• Figure 9 shows a clear correlation between the level and sensitivity effects of ILTV.
• Figure 15 highlights the deviation of average drill-down variables from the European average.
• Annex 1 provides country-weighted averages by variable.
• Annex 2 details the methodology of the top-down analysis.
• Annex 3 and Annex 4 offer breakdowns of the price and bucket mix effects for different variables.
• Annex 5 illustrates the country-level top-down analysis.

Conclusion

The report concludes that while ILTV is the most significant variable in explaining RWA variation, the price effect plays a dominant role in the EU sample. Country-specific credit risk mitigants can significantly affect RWA, but the use of variables in PD/LGD models does not clearly differentiate banks in terms of RWA. The lack of detailed data and consistent definitions across banks and countries limits the depth of the analysis.