EBA欧洲银行-20131217-Report-on-variability-of-Market-RWA_103页_2mb

Summary of the Report on Variability of Risk Weighted Assets for Market Risk Portfolios

Core Content

This report presents the findings of the 2013 Hypothetical Portfolio Exercise (HPE), conducted by the European Banking Authority (EBA) in parallel with the Basel Committee's SIGTB. The exercise aimed to assess the variability in market Risk Weighted Assets (RWA) generated by banks' internal models across different asset classes and risk metrics.

The report includes 35 general portfolios (28 individual and 7 aggregated) and 7 correlation trading portfolios, capitalised under VaR, SVaR, and IRC models. The variability in RWA is measured using the Coefficient of Variation (standard deviation divided by average value). The results indicate that the variability is highest for the APR model (81%), followed by IRC (77%), SVaR (44%), and VaR (33%).

Main Features of the 2013 HPE

• Participating Banks: 17 banks from 9 countries participated in the SIGTB exercise, with the EBA collecting data from 13 banks across 8 EU jurisdictions.
• Portfolio Types: The exercise included portfolios in Equity, Interest Rates, Foreign Exchange (FX), Commodities, and Credit.
• Data Collection: Banks provided Initial Market Valuation (IMV) data for all portfolios as of 10 May 2013, with the exercise running from 3 to 14 June (10 working days).
• P&L Data: Banks using Historical Simulation (HS) for VaR were required to provide one-year P&L vectors to perform alternative VaR calculations (VaR Alt).
• Data Cleaning: Outlier values were excluded from the analysis, and some portfolios were not provided due to model limitations or internal restrictions.

Key Results

Variability in VaR

• The variability in VaR is 33%, which is lower than SVaR (44%) and IRC (77%).
• Applying a homogenised VaR metric reduces variability by around 30% for individual portfolios and 50% for aggregated portfolios.
• Variability in reported VaR for HS banks decreases by 9% for individual portfolios and 14% for aggregated portfolios.
• However, in 4 portfolios (11.4%), variability increases for VaR Alt, and in 10 portfolios (28.6%), it increases for VaR Comp.

P&L Complementary Analysis

• The P&L volatility, rather than correlation, is the main driver of variability in VaR capital outcomes for HS banks.
• There is a lack of consensus on how banks model certain risk factors, such as basis risk in CDS and bonds, or index components.

IRC and SVaR

• IRC variability is 77%, significantly higher than VaR, especially for bespoke portfolios.
• SVaR variability is 44% on average, but the lack of a normalised stressed period makes variability analysis less reliable.
• For corporate risk portfolios, variability is lower (38%), comparable to regulatory VaR (33%).

Internal Model for Correlation Activities (APR)

• APR variability is 81%, higher than IRC and other metrics.
• The standardised APR calculation shows an even higher dispersion of 133%, highlighting the impact of correlation models.

Aggregated Portfolios: Diversification Benefit (DB)

• Larger aggregated portfolios show greater DB than smaller ones.
• Alt VaR has lower dispersion than regulatory VaR.
• SVaR DB is generally higher than VaR DB but not significantly so.
• IRC DB has a median level of 41-42%, much lower than VaR/SVaR DB (70-80%), but higher dispersion than VaR.

Dispersion in Capital Outcomes

• The most inclusive portfolios (29 and 30) show lower variability (26-28%) compared to individual and aggregated portfolios.
• Variability is not driven by complexity in aggregated portfolios, as portfolio 30 (excluding bespoke positions) has lower variability than portfolio 29.
• Regulatory add-ons did not influence variability, but applying 3 multipliers increased the variation coefficient due to lower average capital.

Main Points and Key Information

• Objective: Assess variability in RWA for market risk portfolios due to banks' internal models.
• Portfolio Types: 35 general portfolios and 7 correlation trading portfolios.
• Risk Metrics: VaR, SVaR, IRC, and APR.
• Data Sources: P&L vectors, IMV data, and regulatory inputs.
• Variability Drivers:
  • Modeling Choices: Differences in how banks model risk factors.
  • Supervisory Actions: Regulatory treatment of sovereign exposures.
• Limitations:
  • Limited sample size (maximum of 10 banks).
  • Inconsistent variability observed across portfolios.
  • Non-homogenised stressed period for SVaR hinders meaningful analysis.
• Outlier Handling: Excluded portfolios with outlier values and those not provided due to model or internal restrictions.

Conclusion

The report highlights that variability in RWA for market risk portfolios is influenced by both modeling choices and supervisory actions. While the variability is generally lower for more standardised portfolios, the inclusion of complex instruments and bespoke portfolios increases it. The use of P&L data allows for a deeper understanding of the drivers behind variability, particularly for HS-based VaR models. The findings suggest that there is a need for greater consistency in modeling approaches and regulatory treatment of risk factors to reduce variability across banks.