2015年-EBA欧洲银行管理局_EBA_report_on_CCR_benchmarking_2014_58页_2mb

Summary of EBA Report: Counterparty Credit Risk (Internal Model Method and Credit Valuation Adjustment) Benchmarking Exercise

Core Content

This report presents the findings of the EBA's 2014/2015 market risk benchmarking exercise on Counterparty Credit Risk (CCR) and Credit Valuation Adjustment (CVA) risk, based on data from nine EU banks. The exercise is conducted in line with Article 78 of the Capital Requirement Directive (CRD) and aims to improve supervisory consistency and transparency in the use of internal models for calculating capital requirements.

Main Features of the CCR and CVA HPEs

• Objective: To assess the variability in the Internal Model Method (IMM) and CVA Value at Risk (VaR) models across participating banks.
• Focus: On regulatory risk metrics such as Effective Expected Positive Exposure (EEPE) and Stressed EEPE (S-EEPE).
• Data Sources: The EBA used data from EU banks that participated in the Basel Committee's 2014/2015 exercises, collected by National Competent Authorities (NCAs) through the Task Force Supervisory Benchmarking (TFSB).
• Methodology:
  • Banks were asked to submit CVA VaR data using both their own (free) and a standard (fixed) EE profile.
  • The fixed EE profile is based on the median of the submitted EE profiles.
  • Outliers were identified and removed based on statistical criteria (more than 2.33 standard deviations from the median).
  • A qualitative questionnaire was used to gather information on how banks model their exposures and risk factors.

Key Findings

IMM Variability

• IMM Coverage: On average, IMM covers over 65% of total EAD for all banks except one, which has only 30% coverage. The range is from 66% to 95%.
• Netting Agreements: Most of the IMM EAD is covered by netting agreements. The use of netting agreements is over 65% for all banks except one.
• Margining: The ratio of margined to unmargined netting sets is around 50% for most banks. Bilateral margin agreements are most commonly used, with CCP cleared agreements being relatively rare.
• Monte-Carlo Simulations: All banks use Monte-Carlo simulations. The number of scenarios ranges from 1000 to 5000, with one bank using up to 10,000 scenarios.
• Risk Factors and Correlations: Banks assume correlations between risk factors within and across asset classes. For interest rate (IR) and foreign exchange (FX) derivatives, stochastic processes with drift are used to model risk factors.
• Variability by Trade: The variability in EEPE is higher for equity and FX OTC derivatives than for IR derivatives. This is attributed to differences in pricing models and simulation techniques.
• Variability by Netting Set: The variability of EEPE and S-EEPE is also observed across netting sets, with some showing more pronounced deviations from the median.

CVA Variability

• CVA VaR and S-VaR: The variability of CVA VaR and S-VaR is higher when using a free EE profile compared to a fixed one.
• Capital Requirements: Banks' estimations for CVA capital charges lie in a narrower range when the EE profile is fixed and in a wider range when it is free.
• Stress Periods: The choice of stress periods can significantly affect CVA VaR results, highlighting the need for consistent stress testing practices.
• CVA Risk Charge: The CVA risk charge is influenced by the EE profile used. The report includes detailed comparisons between fixed and free EE profiles.
• Outliers and Missing Data: Some banks did not submit data for certain trades or netting sets, either due to lack of supervisory authorization or missing information.

Limitations and Notes

• Sample Size: The sample is relatively small, consisting of nine EU banks, which limits the statistical robustness of the findings.
• Voluntary Submission: The data were submitted on a voluntary basis, which may affect the representativeness of the results.
• Hypothetical Nature: The exercise is based on hypothetical portfolios, so the results reflect potential variability rather than actual risk exposure.
• Non-Risk-Driven Variability: The variability observed may be due to non-risk-based factors such as differences in model assumptions and practices.
• Interpretation of Results: The results should be interpreted with caution due to heterogeneity in alpha factors and trade-dependent add-ons.

Conclusion

The EBA report provides a framework for future benchmarking exercises under Article 78 of the CRD. It highlights the importance of harmonizing internal model approaches and reducing variability across banks. The findings suggest that while there is significant variability in IMM and CVA models, especially for equity and FX OTC derivatives, there is also a need for more consistent practices in stress testing and margining. The report serves as a basis for further regulatory work and supervisory actions aimed at improving the uniformity and reliability of CCR and CVA risk measures.