|
|
The risk measurement model’s accuracy is reviewed using back-testing. The daily backtesting is stipulated in section 318 (1) of the Solvency Regulation. The value-at-risk (VaR) measure calculated by the model, based on a holding period of one business day, is compared with the change in value of the individual financial instruments or groups of financial instruments included in the model's calculation. The hypothetical and real changes in value are to be used here.
To determine the hypothetical change in value (clean P/L), the financial instruments or groups of financial instruments included in the institution’s portfolio at the close of business on the previous day are revalued using current market prices at the close of business. A calculation of clean P/L through back calculation from the accounting or economic P/L is permitted if as a result compliance with the requirements of section 318 of the Solvency Regulation is guaranteed.
The actual change in value (dirty P/L) is to be calculated on the basis of the accounting and economic P/L, but excluding fees, commissions and the balance from interest income and interest expenditure. No additional adjustments to the economic P/L are to be made. If a loss occurs in the clean P/L or dirty P/L and it exceeds the value-at-risk measure, BaFin and the Deutsche Bundesbank shall be notified promptly of this overshooting, its scale and its cause.
Back-testing is a key instrument in model validation. It is also used to establish the multiplication factor pursuant to section 314 (3) of the Solvency Regulation. The number of during the last 250 business days on the basis of clean P/L and dirty P/L is calculated separately and the larger of the two numbers is used.
BaFin generally takes all overshootings into account when determining the multiplication factor pursuant to section 318 (2) sentence 3 of the Solvency Regulation. In determining the multiplication factor, BaFin will allow, on a case-by-case basis under section 318 (2) sentence 3 of the Solvency Regulation, overshootings of the clean P/L to be the sole criterion if an institution demonstrates that the dirty P/L overshootings are not due to inadequate accuracy of the risk measurement model.
This Notice defines how the overshootings are to be reported and analysed in future. It is designed to help achieve the prompt reporting of overshootings, as well as a suitable depth of analysis. The design of the reporting process as described here will enable the implementation of these aims.
The analysis of overshootings is an important aspect of the supervision of market risk models. Using the overshootings, the institution should test the suitability of its risk measurement model in order to find possible model shortcomings in the light of the overshooting and to make them transparent to the Deutsche Bundesbank and BaFin. Communication of the institution’s analysis makes the supervisory process more efficient and examinations can be omitted or built on these analyses.