EBA releases its annual assessment of the consistency of internal model outcomes for 2020

  • Press Release
  • 15 March 2021

The European Banking Authority (EBA) published today two Reports on the consistency of risk weighted assets (RWAs) across all EU institutions authorised to use internal approaches for the calculation of capital requirements for 2020. The Reports cover credit risk for high and low default portfolios (LDPs and HDPs), as well as market risk. The results confirm that the majority of risk-weights (RWs) variability can be explained by fundamentals. These benchmarking exercises are a fundamental supervisory and convergence tool to address unwarranted inconsistencies and restoring trust in internal models.

Credit Risk exercise

The horizontal analysis of the data collected in the course of the 2020 Benchmarking (BM) exercise is summarised as usual in the credit risk Report. The results of the 2020 regular benchmarking analyses for credit risk are largely comparable to those of the last exercises. This can be seen as an indication of the general stability of bank portfolios and internal model outcomes. As in the past years, the observed variability (in terms of Global Charge ) of institutions’ overall IRBA exposure can mostly be explained by the different share of defaulted and non-defaulted exposures and by the portfolio mix of the individual institutions (i.e. the different distribution of IRB-exposures over exposure classes). For HDP as well as for LDP, around 60% of the total variability is explained via these two drivers.

Furthermore, the Report includes an analysis on the newly introduced portfolios on specialised lending exposures (SLE). This analysis reveals that variability of RWAs calculated under the IRB approach for SLE stems primarily from an unequal distribution across banks in terms of type and volume of investment into SLE. Of particular interest may be the variability stemming from the use of the four different regulatory approaches available for SLE (SA, FIRB, AIRB and the supervisory slotting approach). In line with the expectation, the analysis shows that the average RW consumption of SLE exposures is lower for exposures under AIRB than for FIRB exposures or those under the supervisory slotting approach.

Equally, analysis is provided on the newly introduced split for the large corporates (LCOR) portfolios, which confirms that about three quarters of the exposure at default (EAD) of this benchmarking exposure class is expected to be under the FIRB approach under the revised Basel reform.

Market Risk exercise 

The 2020 exercise considered the same instruments applied in 2019, which are mostly plain vanilla financial instruments. This stability facilitated the understanding of the benchmarking portfolio and contributed to an observed reduction in overall dispersion. 

However, at portfolio level, the variability increases with the risk metric’s complexity, and stressed value at risk, incremental risk charge and all price risk show higher levels of dispersion. Across asset classes, except for commodity exposures, the overall variability for value at risk (18%, it was 21% in 2019 exercise) is lower than the observed variability for stressed VaR (29% it was 30% in 2019 exercise). More complex measures such as incremental risk charge (IRC) and all price risk (APR) show a higher level of dispersion (49% and 45% respectively, compared with 54% and 37% in 2019).

The quantitative analysis, was complemented by a questionnaire to competent authorities on banks participating in the exercise. Although the majority of the causes were identified, and actions put in place to reduce the unwanted variability of the hypothetical RWAs, the effectiveness of these actions can be evaluated only with on-going analysis.

Note to the editors


EBA Report results from the 2020 Credit Risk Benchmarking Report

(1.25 MB - PDF) Last update 3 May 2023

Annex - Chart Pack from the 2020 Credit Risk Benchmarking Exercise

(6.3 MB - PDF) Last update 3 May 2023

EBA Report results from the 2020 Market Risk Benchmarking Report

(4.72 MB - PDF) Last update 3 May 2023

Press contacts

Franca Rosa Congiu