EBA欧洲银行-Public-hearing-on-CPs-on-estimation-and-identification-of-an-economic-downturn-in-IRB-modelling-310518_27页_1mb

Summary of Consultation Papers on Estimation and Identification of an Economic Downturn in IRB Modelling

Core Content

The European Banking Authority (EBA) published two consultation papers on the estimation and identification of an economic downturn in the Internal Ratings-Based (IRB) approach, aiming to enhance the accuracy and consistency of loss given default (LGD) estimates during economic downturns. The consultation period ran from 22 May 2018 to 22 June 2018, building on an earlier consultation launched on 1 March 2017.

The documents are based on Articles 181(3)(a) and 182(4)(a) of Regulation (EU) No 575/2013 (CRR), which require EBA to specify the nature, severity, and duration of an economic downturn for LGD and credit loss (CF) estimation.

Main Views and Key Information

1. Concept of an Economic Downturn

• Nature: Defined by a set of relevant economic factors, including those listed in Article 2(1) and explanatory variables specific to the type of exposure.
• Severity: Determined by the most severe value of a relevant economic factor over a minimum 20-year period. If the severity is observed over a longer period, the minimum observation period may be adjusted.
• Duration: Generally one year, with exemptions to cover adjacent peaks or troughs on one economic factor or multiple correlated factors within the same downturn period.

2. Identification of Downturn Period(s)

• Downturn periods are identified based on the most severe values of economic factors.
• If multiple economic factors are significantly correlated, their downturn periods are combined into one.
• Examples include:
  • Downturn Period A: Unemployment rate peak in 2003
  • Downturn Period B: GDP and Productivity Index trough in 2009

3. General Requirements for Downturn LGD Estimation

• Integration: Downturn LGD estimation is integrated into the EBA/GL/2017/16 Guidelines on PD and LGD estimation.
• Level of Application: At least the same level as used for LGD calibration.
• Multiple Downturn Periods: DT LGD estimates must be provided for each downturn period.
• Selection of Estimates: Institutions should choose the downturn period that provides the highest average DT LGD, with exceptions to prevent overwriting.
• Stability: Downturn LGD estimates should remain stable throughout the economic cycle.

4. Downturn LGD Estimation Methods

4.1 When Loss Data is Available

• Institutions should perform an impact assessment to reflect the effects of the downturn on:
  • Realised LGD levels
  • Annual recoveries
  • Default rates
  • Time in default
• Annual data is preferred, but more frequent data can be used if available.
• Time lags between the downturn and its impact should be considered.

4.2 When Loss Data is Not Available

• Haircut Approach:
  • Precondition: LGD model includes a relevant economic factor as input.
  • Mechanics: Use the downturn period value of the economic factor instead of current values.
• Extrapolation Approach:
  • No precondition required.
  • Mechanics: Use statistical dependency to extrapolate realised LGD from historical data.
• Combined Use: Both approaches can be used together to estimate intermediate parameters.

4.3 Limitations and Policy Considerations

• Institutions should choose the most appropriate combination of methodologies based on:
  • Applicability to estimate realised LGD, intermediate parameters, or risk drivers.
  • Ensuring the final estimate reflects the downturn effect on all relevant economic loss components.
• Mandatory Use of Haircut Approach: If the LGD model includes market value or index related to a relevant type of collateral, and that variable is a relevant economic factor of the downturn period.
• Backstop Approaches:
  • Backstop 1: Any approach can be used, provided it includes a minimum model confidence level (MoC) of 20 percentage points above the long-run average LGD, and does not exceed 105%.
  • Backstop 2: Use the average of the two years with the highest ratio of total economic loss to total exposure in default.

5. Terminology

• Rating System: Methods, processes, and data used to assess credit risk and assign exposures to rating grades or pools.
• Type of Exposures: Homogeneous group of exposures managed similarly.
• Model Development: Process of identifying risk drivers, building statistical/mechanical methods, and estimating intermediate parameters.
• LGD Calibration: Ensuring LGD estimates correspond to the long-run average or a more conservative downturn estimate.
• Calibration Segment: A uniquely identified subset of the model's scope used for joint calibration.

Conclusion

The consultation papers aim to provide clear guidelines on how to estimate and identify economic downturns in IRB models, ensuring that LGD estimates are appropriately conservative and reflect real-world impacts. They propose methodologies such as the haircut and extrapolation approaches, and include backstop mechanisms for cases where data is insufficient. The EBA seeks feedback on the clarity of these approaches and their limitations, with particular focus on the use of methodologies and the establishment of a minimum model confidence level.