EBA欧洲银行-QIS-report-on-default-definition-October-2016_73页_3mb

Summary of the EBA Data Collection Exercise on Default Definition

Core Content

The European Banking Authority (EBA) conducted a qualitative and quantitative impact study (QIS) to assess the impact of proposed regulatory changes on the definition of default used by institutions. The study aimed to harmonise the definition of default across the EU, with the goal of reducing variability in risk-weighted assets (RWA) and improving comparability of regulatory metrics.

The QIS involved 72 institutions, with 8 participating only in the qualitative analysis and 64 in both qualitative and quantitative analysis. These 64 institutions account for 44% of the total EU institutions' credit risk-weighted assets, though the representativeness varies by jurisdiction. The data was collected on a voluntary and best-effort basis, and some institutions did not provide estimations for all policy options.

Main Areas of Analysis

1. Qualitative Analysis - Current Practices

The qualitative part of the study focused on understanding current industry practices in defining default. Key findings include:

• Default Definition Differences: Institutions use varying default definitions, with 46% using more than one. The main reasons for differences are:

  • Materiality Threshold (52% of cases)
  • Number of DPD (15% of cases)
  • Additional Indications of Unlikelihood to Pay (3% of cases)
  • Criteria for Return to Non-Defaulted Status (3% of cases)
  • Other (e.g., treatment of SCRA, etc.)

• Retail Exposure Application: Institutions apply the default definition either at the facility level or the obligor level. Key results:

  • 40% of institutions apply the definition only at the facility level.
  • 30% apply it only at the obligor level.
  • 30% apply it at both levels depending on entities, exposures, or jurisdictions.

• Pulling Effect: When the definition is applied at the facility level, some institutions use the pulling effect, where the default of a significant part of exposures to an obligor leads to the default of the remaining exposures. Key findings:

  • 75% of institutions do not apply the pulling effect.
  • 15% apply it partially.
  • 12% apply it fully, with 5% using a 0% threshold and 7% using a 20% threshold.

• Contagion Rules: When the definition is applied at the obligor level, contagion rules are used to determine whether defaults of one obligor affect others. Key findings:

  • 30% of institutions do not use contagion rules.
  • 18% use rules in line with the EBA's proposed guidelines.
  • 37% use other specific rules, including case-by-case assessments, and some differentiate between retail and non-retail exposures.

• Technical Default Definitions: Although not explicitly defined in Regulation (EU) No 575/2013, technical default is a common concept used to describe situations that may trigger default status. The EBA identified several types of technical default definitions used by institutions.

• Materiality Thresholds: Institutions use different materiality thresholds for retail and non-retail exposures. The analysis showed significant variability in the structure, reference, and application of these thresholds.

• Indications of Unlikelihood to Pay: Institutions use various indicators, including SCRA, impaired exposures, and sale of credit obligations. Some use alternative indicators not prescribed in Article 178(3) of the regulation.

• Probation Period: The use of probation periods for returning to non-defaulted status varies. Some institutions apply it partially, while others do not use it at all.

2. Quantitative Analysis

The quantitative part of the QIS aimed to estimate the impact of the proposed policy options on capital requirements. Key findings:

• Impact on Capital Ratios: The introduction of a harmonised default definition is expected to lead to a modest increase in capital charges, with an average reduction of CET1 capital ratio by about 20 basis points for IRB institutions, and a limited reduction for SA institutions.

• RWA Variability: The analysis confirmed that differences in default definitions are a major driver of RWA variability. Harmonising these definitions is expected to reduce this variability.

• Policy Scenarios Tested:

  • Materiality Threshold
  • Technical Defaults
  • Specific Credit Risk Adjustments (SCRA)
  • Sale of Credit Obligations
  • Probation Period
  • Contagion Effect

• Methodological Considerations:

  • The quantitative analysis was based on representative samples of exposures, weighted by the institutions' total assets.
  • The results are subject to simplifying assumptions and institutional subjectivity, which may affect accuracy.
  • Data quality issues were identified, particularly regarding sample representativeness and approximations in RWA and capital ratio calculations.

Key Information

• The harmonisation of default definitions is expected to lead to more comparable metrics and reduced RWA variability.
• Materiality thresholds, DPD criteria, and indications of unlikelihood to pay are the main sources of variability.
• The quantitative analysis is based on simplified assumptions and representative samples, which may not fully reflect real-world scenarios.
• IRB institutions are more likely to use the obligor-level definition of default, while SA institutions tend to use the facility-level definition.
• Contagion rules and probation periods are applied inconsistently across institutions.
• The EBA used data from the QIS and COREP to assess the impact of the proposed changes, but the results should be interpreted with caution due to data quality and methodological limitations.

Conclusion

The study highlights the divergence in default definitions among institutions and identifies materiality thresholds, contagion rules, and probation periods as the primary sources of variability. Harmonisation of these definitions is expected to lead to more consistent regulatory capital requirements and reduced RWA variability. However, the quantitative results should be treated as indicative, due to the simplifications and subjectivity involved in the analysis.