EBA欧洲银行-AFME-on-LGD_20170117_final_for-EBA-website_9页_676kb

AFME Summary on LGD Estimation and Defaulted Assets

Core Content

The document provides AFME's feedback and comments on the European Banking Authority (EBA) guidelines regarding the estimation of Loss Given Default (LGD) and the treatment of defaulted assets. It outlines key considerations and questions raised by AFME members during an EBA workshop, focusing on the practicality, consistency, and alignment of LGD and ELBE (Expected Loss on Bank Exposure) methodologies with current industry practices and regulatory expectations.

Main Views and Key Points

LGD Estimation and Defaulted Assets

• Collateral vs Unsecured Exposures: AFME suggests that LGD approaches should not unnecessarily differentiate between collateralised and unsecured exposures, as the recovery process is often managed at the borrower level.
• Time Horizons and Recovery: Time horizons are important, and the recovery process should consider more than just repossession. Portfolio diversification adds value and should be factored into the analysis.
• Single LGD for Multiple Vintages: Some portfolios may apply a single LGD to multiple vintages of performing facilities, even with varying Loan-to-Value (LTV) levels.
• Repossession and Recovery Strategies: Banks should be allowed to optimise repossession and recovery strategies over several years.

Data Representativeness

• Data Comparisons: AFME supports the idea of comparing defaulted contracts (modelled vs observed) rather than between defaulted and performing portfolios.
• Intrinsic Differences: There is a need for clarification on how to handle a lack of representativeness due to differences between defaulted and performing exposures.
• Consistency Between Databases: AFME calls for more clarity on the consistency between PD and LGD databases, especially regarding periods of observation and risk profiles.

Additional Drawings and Capitalized Interest

• Additional Recoveries: Additional recoveries, such as restructuring fees, should not be neutralised if they represent economic gains.
• Interest and Fees Capitalized: Interest and fees capitalized after default should be included in the calculation of realized LGD.

Discounting Rate and Add-on

• Single Rate Proposal: The proposal of using a single rate (BOR + add-on) is seen as simple and precise, though there is debate over whether it should consider the nature of the loan.
• Alternative Approaches: Using the cost of funding or BOR + loan margin could be more accurate.
• Convergence with Provisioning Standards: There is a call for convergence with provisioning standards, such as referencing contract rate or effective interest rate.

Collateral Treatment

• Recovery Cash Flows: Recovery cash flows from collaterals not recognized by the Capital Requirements Regulation (CRR) should be considered.
• Haircuts and Downturn Component: There is a need for further work to clarify the link between collateral haircuts and the downturn component in LGD to avoid double counting of adverse events.

LGD Estimation: Long-Run Average and Downturn Adjustment

• Incomplete Recovery Processes: AFME supports the EBA's approach to include unresolved cases in LGD estimates, as it reflects current industry practices.
• Minimum Observation Period: A minimum period of observation is recommended to exclude too recent or inconsistent data.
• Recovery Procedures: Relatively long recovery procedures should be allowed, reflecting the legal environment in certain countries.
• Future Recoveries: Future recoveries linked to collaterals for open cases should be included in the estimation.

Downturn Adjustment

• Additional Guidance Needed: AFME believes that more guidance is necessary to clarify the concept of downturn adjustment.
• Prudential Considerations: LGD in default should reflect a downturn cycle, which is not clearly addressed in the RTS 2016-03.

ELBE and LGD in-Default Estimation

• Diverse Practices: There is a wide diversity of practices in estimating ELBE and LGD in-default, with some firms using LGDD models directly and others using provisions.
• Backtesting: Proper backtesting of ELBE, specific provisions, and final loss is essential.
• No Full Convergence: AFME does not support full convergence of methodologies between LGD for performing exposures and LGDD, as they serve different purposes.

Reference Date and Economic Circumstances

• Reference Date: The reference date notion is meaningful and should be used in ELBE and LGD in-default estimation.
• Economic Factors: Economic factors are usually considered via expert judgment, by sector, or by the nature of collateral. Unexpected losses may require the inclusion of a forward-looking component like MoC (Market Outlook Component).

Provisioning Models

• Use of Provisions: AFME supports the use of provisions for ELBE estimation and suggests that the conditions for using them should be widened to avoid the need for redevelopment of IFRS9 models.
• Auditor Scrutiny: Provisioning models are subject to auditor scrutiny and public disclosure, and changes would not meet a cost/benefit test.
• Alignment with EBA Orientation: AFME recommends aligning the use of provisions with EBA orientations from the CP2016-10 (ECL accounting), as the current conditions are too restrictive.

Conclusion

AFME's feedback highlights the need for practical and consistent approaches to LGD and ELBE estimation, while also advocating for clarity and alignment with existing regulatory and accounting frameworks. The association supports the EBA's proposals but calls for further refinement and guidance to ensure they are operationally feasible and reflective of real-world conditions.