«Analysis of the trading book quantitative impact study October 2009 Copies of publications are available from: Bank for International Settlements ...»
4. Incremental risk capital charge The estimated size of the incremental risk capital charge is reported in Table 3 as a fraction of market risk capital requirements for liquidity horizons of one month, three months and six months. In total, 25 banks from nine different countries submitted results on the estimated impact of the IRC. Of the 25 banks, three included equity exposures into their IRC model.
Since banks that adopt an IRC model will no longer be subject to a specific risk surcharge, the net effect of the IRC must be measured by subtracting the specific risk surcharge, reported in column 2, from the IRC. Using the case of a three-month liquidity horizon as a benchmark, reported in column 4, the net effect of the IRC is estimated to result in an average increase of 103% in market risk capital. This result, however, is affected by a few large outliers as indicated by the significantly lower median of 60%. 7 The bank-level results indicate that the IRC produces a net increase in market risk capital for all but two banks. 8 The size of the IRC does depend on the assumed liquidity horizon. These results indicate that increasing the liquidity horizon from one to six months increases the capital charge, on average, by 20% though there is substantial variation in the size of the increase. 9 Bank-level The median increase of 60% differs from the result obtained by taking the difference between column 4 and column 2 of Table 3. The set of banks reporting a three-month IRC differs from that reporting a specific risk surcharge. The median increase is computed by considering only those banks that report both a three-month IRC and a specific risk surcharge.
The net decrease for two banks is due to a high specific risk surcharge under the current regime for these banks.
Table 3 suggests that the average IRC actually decreases as the liquidity horizon increases from one month to three months. This result is driven by the fact that the set of banks reporting a one-month IRC differs from that Analysis of the trading book quantitative impact study results indicate that four banks actually report a decrease in IRC as the liquidity horizon lengthens from one month to six months while at the other extreme, two banks report roughly a 200% increase in the IRC.
The current version of the IRC guidelines includes both default and migration risks as well as equity risks which may be included in the model. In the final column of Table 3 the results for the default-only IRC at a liquidity horizon of three months are reported. Comparing these default-only charges with the full IRC charges of the same liquidity horizon indicates that the inclusion of non-default risks in the IRC results in an average increase of roughly 33% over a default-only IRC. Finally, in the case of three banks adding non-default risks to the IRC is estimated to actually reduce the capital requirement. In each of these cases, the bank’s credit exposure is net short. Moreover, in one case the bank has confirmed that the methodology used to compute the IRC assumed that credit migrations and defaults were independent rather than mutually exclusive. This assumption tends to overstate the effect of migrations as migrations may occur in the event of default and was employed by the bank in order to comply with the survey deadline; it would not be employed in a more carefully specified IRC model.
5. Stressed value-at-risk Results of the stressed VaR calculations were provided by 38 banks from 10 countries. The data period used to calculate the non-stressed VaR is in most cases the period ending on 31 December 2006. This should represent a VaR number for a non-stressed period in terms of market risk factor movements.
reporting a three-month IRC. Comparing results across banks that report both a one-month and three-month IRC estimate indicates an average increase in IRC of 3%.
The second column relates the stressed VaR to the non-stressed VaR, both calculated for the same portfolio. It can be seen that the stressed VaR is on average 2.6 times the nonstressed VaR, with extremes as high as seven times and as low as two-thirds the nonstressed VaR. 10 The last column shows the diversification effect within the stressed VaR relative to the diversification effect within the non-stressed VaR. Diversification effects were estimated by comparing the total VaR number including diversification effects to the sum of the VaR numbers per defined sub-portfolios, which neglects diversification between these subportfolios. The results were not conclusive since the non-stressed VaR exhibits sometimes more and sometimes less diversification benefit than the stressed VaR.
In particular, the supposition that the stressed VaR would reflect lower diversification effects than the non-stressed VaR was generally not confirmed, as the average of 99.8% for the last column shows.
6. Re-securitisation capital charges Twenty-eight banks from 10 countries submitted results on the impact of the securitisation charges. Re-securitisations positions accounted for an increase of 118% in market risk capital. The rated re-securitisation positions accounted for 72.7% of the increase and the unrated for 27.3% of the increase. These positions will continue to be capitalised under the standardised method under the Committee’s July 2009 revisions. Ten of the 28 banks had significant re-securitisation positions which would result in an increase in market risk capital requirements of over 50%.
For one bank, stress VaR is materially lower than the non-stress value-at-risk because of the specific portfolio composition of that bank.
Analysis of the trading book quantitative impact study The incremental modelled specific risk charge from re-securitisations (calculated by excluding them from the internal models) showed that under the current framework resecuritisations accounted for 10.5% of the total market risk capital charge. In some cases removing re-securitisations from the VaR model would increase the capital charge as diversification benefits would be lost. However, for calculating the summary statistics in the table below and the overall results in Section 3 it was assumed that banks would not remove the exposures from their VaR model if this would occur.
7. Equity specific risk The capital charges resulting from the preferential treatment for liquid and well-diversified equity portfolios under the standardised measurement method to specific risk was reported by 12 banks from eight countries. The preferential treatment differs between those countries with standardised capital charges as low as 1.2% of the exposure amount. The removal of the preferential treatment will result in a uniform standardised specific risk capital charge of 8% of the exposure amount.
The contribution of those equity risk positions which currently receive preferential treatment, to the overall market risk capital requirements will on average increase by roughly 200%.
One bank might see an increase of more than 500%.
The effect of the removal of the preferential treatment on overall market risk capital requirements in most cases is very small. For two banks though, the percentage of required market risk capital due to equity position specific risk is significant under the current rules and will double to about 30% under the new rules.
Calculation of capital requirements for market risk The capital charges according to the market risk framework comprise interest rate related instruments and equities in the trading book as well as foreign exchange risk and commodities risk irrespective of whether the position is held in the banking or in the trading book.
Under the previous market risk framework, the capital requirements for market risk are
calculated as the sum of the following elements:
• The capital charge according to the standardised measurement method to the extent a bank does not use internal models, covering • general and specific interest rate risk;
• general and specific equity position risk;
• foreign exchange risk;
• commodities risk;
• The capital charge according to the internal models approach, which is the higher of (1) its previous day’s VaR number; and (2) an average of the daily VaR measures on each of the preceding sixty business days, multiplied by a multiplication factor.
To the extent a banks’ internal model does not cover specific risk, the specific risk capital charges of the standardised measurement method apply. Under the revised market risk framework, the capital requirements for market risk are calculated as the sum of the
* The multiplier may be adjusted up to 4 based on backtesting results. Banks may use one VaR model jointly modelling general and specific risk. ** The Committee will evaluate a floor for the comprehensive risk capital charge which could be expressed as a percentage of the charge applicable under the standardised measurement method.
Analysis of the trading book quantitative impact study Glossary Capital charge for incremental risk: The capital charge for incremental risk is calculated based on a risk measure that includes default risk as well as migration risk for unsecuritised credit products held in the trading book at a 99.9% confidence level, a one-year capital horizon, and a minimum liquidity horizon of three months.
Capital charge for comprehensive risks: The capital charge for comprehensive risk is calculated based on a risk measure that can be applied to banks’ so-called correlation trading portfolios and captures not only incremental default and migration risks, but all price risks at a 99.9% confidence level and a one-year capital horizon.
Specific risk surcharge: The specific risk surcharge for banks including both general and specific risks in their internal VaR models is defined as the difference between (i) the capital charge for both general and specific risks when applying the higher multiplier of four to the VaR measure for specific risk; and (ii) the capital charge for both general and specific risks applying a uniform multiplier of three.
Stressed value-at-risk: A measure intended to replicate a VaR calculation that would be generated on the bank’s current portfolio if the relevant market factors were experiencing a period of stress. Therefore, it should be based on the 10-day, 99th percentile, one-tailed confidence interval VaR measure of the current portfolio, with model inputs calibrated to historical data from a continuous 12-month period of significant financial stress relevant to the bank’s portfolio. The period used must be approved by the supervisor and regularly reviewed. As an example, for many portfolios, a 12-month period relating to significant losses in 2007/2008 would adequately reflect a period of such stress; although other periods relevant to the current portfolio must be considered by the bank.