EIOPA publishes the results from its yearly study on the modelling of market and credit risk

The European insurance and Occupational Pensions Authority (EIOPA) published today the results from its yearly study on the modelling of market and credit risk (MCRCS). This study is based on simplified asset-liability-portfolios and also focusses on the analysis of interest rate ‘down’ movements, comparison of model results to historical market experience, and preliminary analyses on dependency structures.

As in past studies, the overall results continue to show significant variations in asset model outputs, which are partly attributable to model and business specificities already known by the relevant national competent authorities, but also indicate a need for further supervisory scrutiny. 

The results, tools and experience feed into the Supervisory Review Process on internal models, and are also used to assess model changes or models in pre-applications.

The report shows the following conclusions

  • Credit risk charges for sovereign bonds across groups of modelling approaches show relatively low variation for bonds issued by Germany, Netherlands, Austria, Belgium, and France. The variation is greater for the bonds issued by Ireland, Portugal Spain, and Italy. These results are influenced by firms which show zero or low credit risk shocks across the instruments.
  • Credit risk charges for corporate bonds are generally higher for bonds with lower credit ratings and the variation increases materially with worsening credit quality. The variation becomes substantial for BB-rated bonds. This demonstrates the variety of modelling assumptions being made by firms, particularly for low rated bonds.
  • Equity and real estate risks. With respect to equity risk, undertakings in general show less variation in the risk charges for major equity indices compared to risk charges applied to the strategic equity participations. Risk charges applied to the five real estate benchmark investments vary to a larger extent compared to equity. Additionally, for real estate category, model calibrations tend to place more emphasis on the risk profile of the undertakings’ actual investment portfolio and less on publicly available indices.

This year’s study includes also the following new elements

  • Effects on the modelling of market and credit risk of the COVID-19 crisis. The study provides a first evaluation and analysis will be on-going. This first evaluation is presented as a comparison of model outcomes against historical experience. Although the observed market impacts are significant, no immediate conclusions can be drawn regarding the appropriateness of models under such circumstances. 
  • Sustainability. Participants were asked about how sustainability is considered in their modelling approach. Of 21 respondents, only 1 explicitly uses a taxonomy for “green assets” in its model. This will be monitored in next editions.
  • Dependency structures. An analysis of dependency structures was performed for the first time and leads to observations which will be taken up in further work and need further scrutiny. 

The year-end 2020 survey was launched on 15 January 2021 and the results will be published in early 2022.

Download the results