47. Supervisors will be interested in some or all of the following data: (a) position data that would allow independent supervisory assessment of market risk through the use of some supervisory model or monitoring criteria and (b) data derived from an institution's own internal estimates of market risk. To minimise burden, supervisory assessment of market risks based on position data or internal models should start with and draw as much as possible on the information that institutions must collect for supervisory capital purposes. For example, in the case of the banking sector, information that banks use to determine their compliance with the Basle Committee's market risk capital requirements should be considered by supervisors when assessing banks' market risks. In addition, supervisors should assess the position information and internal estimates of market risk that institutions use for other risk management purposes that go beyond information related to market risk capital requirements.

48. The collection of position data could be carried out at various levels of detail, depending on the nature and scope of the institution's trading and derivatives activities. The detail can range from a broad measure of exposure at the portfolio level to a finer disaggregation by instrument and maturity. For certain institutions, particularly those that are not major dealers, it may be appropriate to obtain position data (e.g., equities, debt securities, foreign exchange and commodities), which could be drawn from the framework of the Basle Committee's standardised approach for market risk ¹³, or from other approaches adopted by national banking and securities supervisors.

49. For example, banking organisations using the Basle Committee's standardised approach to determine their minimum market risk capital charge typically would develop the following position data ¹⁴:

Interest rate risk positions

• General market risk - long and short positions broken down by time-bands according to residual maturity or to duration; breakdown of positions by currency (main currencies relative to the activity of the firm [for supervisory purposes] or all G-10 currencies [for G-10-wide aggregation exercise]).
• Specific risk - breakdown of positions according to issuer (government, qualifying, other) and, to some extent, maturity.

Equity risk positions

• Long and short positions broken down by major markets; breakdown by issuer types; futures-related and index-based arbitrage strategies.

Foreign exchange risk positions

• Net long or net short position by currency (including gold).

Commodities risk positions

• Net long or short position by commodity type.

Options Risk (for all risk categories)

• Delta equivalents of portfolios of options.
• Gamma and vega risk.

This information is used by banks to determine their minimum capital charge for general and specific market risk under the standardised approach. This information is illustrated further in Annex 5.

50. As an alternative or supplement to assessing position data, supervisors could evaluate available information on an institution's internal estimates of market risk. For some institutions, this information could be derived from their internal value-at-risk methodology, which involves the assessment of potential losses due to adverse movements in market prices of a specified probability over a defined period of time. As an alternative to value-at-risk, supervisors may find it useful on a case-by-case basis to assess internally-generated information on earnings-at-risk ¹⁵, duration analysis, stress scenario analyses, or any other appropriate approach that sheds light on an institution's market risk. Whatever the approach taken, supervisors should consider the measure of market risk exposure in the context of the institution's limit policies.

(a) Value-at-risk estimates

51. If a firm uses value-at-risk models for measuring market risks, the supervisor should evaluate in detail the methodology used, including its main parameters, for both market risk capital purposes and other risk management purposes. Key parameters for evaluating value-at-risk estimates include: (1) position sensitivities, (2) the market risk volatility and correlation assumptions of the underlying model (using historical volatilities), (3) the holding period over which the change in portfolio value is measured, (4) the confidence interval used to estimate exposure, (5) the historical sample period over which risk factor prices are observed, (6) the method of estimation, (7) the approach to nonlinear risk, and (8) the approach to specific risk.

52. Value-at-risk measured solely at a point in time may not provide appropriate insights about market risk due to the speed with which positions in derivatives and other instruments can be altered. Such difficulties may be addressed by the use of summary statistics for the period over which the institution is reporting. For example, supervisors could assess information on the highest value-at-risk number measured during the reporting period, together with monthly or quarterly averages and related ranges of value-at-risk exposures. By comparing end-of-period value-at-risk with these other measures, supervisors can better understand the volatility which has occurred in these measures during the period. However, time series of daily value-at-risk estimates are more informative than averages or ranges.

53. Supervisors could also encourage or require institutions to convey comparisons of daily value-at-risk estimates with daily changes in actual portfolio value over a given period ¹⁶. Internal models should be validated by comparing past estimates of risk with actual results and by assessing the models' major assumptions (often referred to as "backtesting"). For example, an institution could periodically compare its one-day, 99 percent confidence interval value-at-risk estimates with the daily profits and losses for the entire trading portfolio. Institutions should also periodically evaluate the major assumptions underlying their internal models used for market risk capital purposes and other risk management purposes. Time series of value-at-risk estimates, histograms of daily trading profits and losses, and other internally-produced backtesting results can be very useful to supervisors in assessing the accuracy of value-at-risk estimates used for market risk capital purposes and other risk management purposes ¹⁷.

54. Value-at-risk estimates may be provided on an aggregate basis for the entire trading portfolio. In addition, value-at-risk estimates are particularly informative when provided on the basis of major trading risk categories (e.g., interest, foreign exchange, equity or commodity) or business line of the institution. Moreover, some institutions supplement their value-at-risk estimates for trading activities with those for their end-user activities, as well as a consolidated measure for the entire institution.

(b) Stress test information

55. Institutions with significant trading activities should subject their portfolios on a regular basis to stress tests using various assumptions and scenarios ¹⁸. Institutions' stress scenarios need to cover a range of factors that can create extraordinary losses or gains in trading portfolios, or make the control of risk in those portfolios very difficult. These factors include low-probability events in all major types of risks. Stress scenarios should provide insights into the impact of such events on positions that have both linear and non-linear price characteristics.

56. These analyses of the portfolio under "worst-case" scenarios should preferably be performed on an institution-wide basis and should include an identification of the major assumptions used. Quantitative information on the results of stress scenarios, which could be specified by supervisors or institutions themselves (or a combination of both approaches), coupled with qualitative analyses of the actions that management might take under particular scenarios, would be very useful for supervisory purposes. Examples of scenarios for interest rate risk include a parallel yield curve shift of a determined amount, a steepening or flattening of the yield curve, or a change of correlation assumptions. Simulations could include testing the current portfolio against past periods of significant disturbance, such as those involving the largest one-day, five-day, and 30-day gains and losses, significant past "events" (e.g., the 1987 major stock market decline, the ERM crises of 1992 and 1993, and the 1997 Asian financial crises), and other stress events.

(c) Information on the quality of market-risk information processes

57. The quality of the processes and models that generate value-at-risk estimates, stress scenarios, and other measures of market risk, including the adequacy of related internal controls, may be evaluated in reports prepared by the institutions' independent risk management/control units, internal auditors, external auditors, consultants and other experts. Supervisors can gain important insights into the quality of market risk information by reviewing reports on these topics (and, when available, supporting documentation for these reports).

Footnotes:

13. "Amendment to the Capital Accord to incorporate market risks", Basle Committee on Banking Supervision, January 1996.

14. In evaluating this type of position data, supervisors should understand the qualitative criteria underlying these reported amounts and implications for the comparability of position information across institutions, and should adapt their reporting requirements accordingly. For example, under the Basle Accord's standardised approach, national supervisors may allow either full offsetting of positions or restrict offsetting of positions between different entities within a banking organisation. In assessing this information, supervisors should understand whether the market risk capital information is presented by the institution on a group-wide, consolidated basis and the extent of off-setting between entities within the group.

15. Under mark-to-market accounting, value-at-risk will equal earnings-at-risk because changes in value are reflected in earnings. If accrual accounting is applied to certain positions, value-at-risk and earnings-at-risk will differ because all changes in value are not reflected in earnings.

16. The report of the Euro-currency Standing Committee, a discussion paper entitled, "Public Disclosure of Market and Credit Risks by Financial Intermediaries", issued in September 1994 (Fisher Report), discusses factors to consider in interpreting value-at-risk measures, among other topics.

17. For banking organisations, the Basle Committee's discussion paper entitled, "Supervisory Framework for the use of "Backtesting" in Conjunction with the Internal Models Approach to Market Risk Capital Requirements", issued in January 1996, explains issues related to the use of backtesting in assessing the accuracy of model-generated risk measures.

18. Banks that use the internal models approach for meeting the Basle Committee's market risk capital requirements must have in place a rigorous and comprehensive stress testing programme. Stress testing is a key component of a bank's assessment of its capital position. Under the Basle Committee's market risk capital requirements, supervisors may ask banks using the internal models approach to provide information on stress testing in three broad areas: (a) supervisory scenarios requiring no simulations by the bank, (b) scenarios requiring a simulation by the bank, or (c) scenarios developed by the bank itself to capture the specific characteristics of its portfolio.