EBA Releases Results of 2019 Benchmarking Exercise for Internal Models

The credit risk report examines the different drivers leading to the observed dispersion across banks' models. The results are broadly in line with the previous exercises, with 50% of the difference in variability explained with simple risk drivers, a risk-weighted deviation on low-default portfolios below 10 percentage points and estimates for high-default portfolios generally on the conservative side when compared with empirical observed metrics. Furthermore, this year, for the first time, on high-default portfolios, EBA performed a comparison with the standardized approach risk-weights. The overall observed variability under the standardized approach is at a similar level than the one observed on internal rating-based (IRB) approach. Within a single exposure class, the variability under the IRB approach follows, in a conservative manner, the empirical variability of risk (observed via default rates). In addition to a questionnaire filled in by supervisors and interviews conducted with seven institutions, a survey was conducted among institutions to better assess the variability of practices in terms of rating scales. This survey highlights the variability of practices on the type of calibration of the probability of default.  

The market risk report presents the results of the 2019 supervisory benchmarking and summarizes the conclusions drawn from a hypothetical portfolio exercise conducted by EBA during 2018-19. Compared to the previous exercises, the 2019 analysis shows a substantial reduction in terms of dispersion in the initial market valuation and some reduction in risk measures, especially for the aggregated portfolios. This improvement was expected and is likely due to the simplification in the market risk benchmarking instruments. The remaining dispersion is probably the result of new benchmarking instruments being used by banks for the first time. The quantitative analysis, which has been extended in terms of scope with respect to the previous exercises, was also complemented by a questionnaire to competent authorities. Although the majority of the causes were identified and actions were put in place to reduce the unwanted variability of the hypothetical RWAs, the effectiveness of these actions can be evaluated only with ongoing analysis. The 2019 exercise is the first exercise with the new set of hypothetical instruments and portfolios. The new set of instruments mainly consists of vanilla instruments and is more extensive in terms of the number of instruments to model with respect to the three previous benchmarking exercises.

Related Links

• Press Release
• Results of Credit Risk Benchmarking (PDF)
• Annex: Charts from Credit Risk Benchmarking (PDF)
• Results of Market Risk Benchmarking (PDF)

Keywords: Europe, EU, Banking, Credit Risk, Market Risk, Benchmarking, Internal Models, 2019 Benchmarking Exercise, Regulatory Capital, EBA