EBA publishes its regular assessment of EU banks internal model outcomes

 

The HDP report explains the overall level of variability in RWAs and examines the different drivers that explain the dispersion observed. Most of the results are broadly in line with previous exercises on HDP. Given the different approaches used, namely a top-down and an outturn (backtesting) approach, in addition to a distribution analysis, the respective findings should be read concurrently. Two indicators are used to summarise the results, the RW and the ‘global charge' (GC, i.e. considering expected and unexpected losses).

 

A key finding from the top-down approach is that more than 80% of the observed difference in the GC levels across institutions could be explained by few drivers, namely: the proportion of defaulted exposures in the portfolio; the country of the counterparty; and the portfolio mix. The remaining variability could be attributed to differences in riskiness – i.e., idiosyncratic portfolio features, modelling assumptions and risk management practices used by banks, as well as supervisory practices. backtesting approach shows that, in general, the estimated values for PDs and LGDs are above the observed values for defaults and loss rates. Nevertheless, some banks systematically show observed values higher than estimates and will need a closer analysis.

 

This exercise highlighted several areas in which supervisors – and colleges – should investigate further, for instance: the practices regarding defaulted exposures; the definition of default; the use of global models and the interaction with country-specificities for exposures with counterparties from different jurisdictions; unjustified differences between regulatory approaches and possible compensation effects between internal approaches.

The market risk report presents the observed variability measures in terms of the inter-quantile dispersion statistic (IQD). In the initial market valuation results, interest rate portfolios show a lower variability than other asset classes due to a more homogeneity across banks for modelling interest rate risk. In line with the previous exercises on market risk weighted assets variability, a significant dispersion for all the risk measures provided by banks is observed. As expected, the overall variability for Value at Risk (VaR) is lower than that observed for Stressed VaR (sVaR) and more sophisticated measures like Incremental Risk Charge (IRC) and All Price Risk (APR) show a much higher level of dispersion. Modelling choices play an important role in explaining this variability, especially for the most complex risk measures.

 

This report highlights some areas that may require further investigations by Competent Authorities, such as accentuated pricing variability for equity derivatives, commodities trades and credit spreads products.

These annual benchmarking exercises are contributing to the work the EBA is conducting for improving the regulatory framework and restoring confidence in internal models.

 

The EBA is exploring a number of options and publishing guidelines to address specific concerns, such as the report specifying the methodology assessment that competent authorities shall follow when assessing the compliance of an institution with the requirements to use the IRB approach or the guidelines on application of the definition of default. In parallel, the exercises are providing a regular supervisory tool based on benchmarks to support competent authorities' assessments of internal models and produce comparisons with EU peers.