Risk Library
   Documents by Author
     Additional Regulatory Documents
       Framework for Voluntary Oversight
         IV. Evaluation of Risk in Relation to Ca...


Framework for Voluntary Oversight

IV. Evaluation of Risk in Relation to Capital

I. Overview

One of the more complex and important tasks faced by the DPG was its effort to develop (1) a framework for estimating the credit and market risk exposures associated with OTC derivatives activities and (2) an approach for evaluating those risks in relation to capital. This task required the resolution of the following issues:

    First, there must be an agreed-upon definition of the category of OTC derivative products covered.

    Second, there must be an agreed-upon definition of the covered business entities.

    Third, there must be an agreed-upon method or technique for calculating or estimating the market and credit risks associated with portfolios of OTC derivative products as defined in steps one and two above.

    Fourth, depending on the methodologies and techniques adopted for estimating market and credit risks, there must be agreed-upon standards and verification procedures to ensure that such methodologies and techniques will be utilized in a consistent and rigorous manner.

    Fifth, given the estimates of market and credit risk, there must be a prudentially-based framework to enable senior management and supervisors to evaluate the estimates of market and credit risk in relation to capital.

    Finally, all of the above must be achieved in a setting mindful of the competitive environment in which each of the participating firms must operate. Indeed, there is currently no official consensus on market risk measurement and capital adequacy relating to the OTC derivatives activities of banks or securities firms, either nationally or internationally, much less an approach that would apply to banks and securities firms both nationally and internationally.

II. Definitions of Covered Products and Entities

The approach to the definition of products and entices is driven by a number of key considerations, including the following:

  • Based on articulated regulatory and congressional concerns, it was recognized at the outset that the primary focus must be directed at OTC derivatives activities conducted in otherwise unregulated entities.

  • Because markets and practices are far from static, the definitions used must provide an element of flexibility so as to avoid the costs and burdens of seeking to ensure that every last transaction in every last office in every comer of the world is captured.

  • It was also considered essential to maintain consistent definitions in connection with the framework for enhanced reporting and the evaluation of risk in relation to capital.
Relying primarily on the considerations outlined above, the following definitions of covered products and entities were adopted by both the reporting and capital working groups:

  • Covered products include interest rate, currency, equity and commodity swaps, OTC options (including caps, floors, and collars) and currency forwards (i.e., currency transactions of more than a two-day duration, except that firms may elect to include only currency transactions of 14 days or more duration).

  • Covered entities include any affiliate of an SEC-registered broker-dealer ( I ) that is not subject to supervisory oversight with respect to capital, (2) that is primarily engaged in the business of holding itself out to unaffiliated counterparties as a professional intermediary willing to structure and enter into either side of an OTC derivative transaction as principal, and (3) whose OTC derivatives activities are likely to have a material impact, directly or indirectly, on its SEC-registered broker-dealer affiliate.

While the definition of covered products was reasonably straightforward, reaching agreement on the definition of covered entities was more problematic. There was, for example, something to be said for limiting the definition to U.S.-based entities or limiting it to previously identified "material associated persons". Ultimately, the foregoing definition was selected because it came closer to the mark in logic and substance, even if it was somewhat broader than some of the individual firms would have preferred.

The DPG recognizes that the definition of covered entities has a potential weakness in that the "primarily engaged" test and the "materiality" test might permit an entity or a group of derivatives transactions to fall outside the contemplated frameworks for enhanced reporting and risk evaluation. However, this potential weakness is seen by the DPG as minor and can be addressed on a case-by-case basis should such a problem arise.

III. The Technique for Estimating Market and Credit Risk

Once the key definitions had been set, the next step in the process entailed working toward techniques or methodologies for estimating the amount of market and credit risk associated with given portfolios of OTC derivative products as defined above.

In approaching this task, it was recognized that, at the most abstract level, only three generic approaches were potentially viable. The first consists of the so-called "building block" approach published for comment by the international banking regulators in 1993. This was rejected in part because of its rigidity and complexity and in part because it appears increasingly unlikely that this approach will be adopted by the international community of banking supervisors.

A second possible approach consists of the application of low, across the board, tax-like capital charges to the notional value of OTC derivative transactions. While such an approach has the virtue of simplicity, it was rejected for many reasons, including, in particular, the extremely arbitrary results it would produce.

The third approach, and the one adopted by the DPG, consists of the use of statistical and econometric models for estimating the "capital at risk" associated with OTC derivatives activities. Among the major professional intermediaries worldwide, such models have emerged as the technique of preference and necessity in measuring and managing such risks, as well as in valuing certain OTC derivative products. In addition, it appears that the international supervisory community is moving in the same direction.

While the use of models for risk estimation has obvious attraction, such techniques raise a number of difficult issues when applied to a group of individual firms. The most difficult of such issues that had to be addressed by the DPG are briefly outlined below in Sections "A" through "D" below.

  1. The Measure of "Capital at Risk" due to Market Risk Considerations

    The technical, statistical and behavioral characteristics of the models used by major derivatives intermediaries are, in principle, well-equipped to provide estimates of "risk of loss" or "capital at risk" associated with a given portfolio of OTC derivative products. That is, given any portfolio of OTC derivative products on any day of the week, it is a relatively straightforward matter to estimate with a particular statistical probability what future losses might be associated with the portfolio, based on the correlations and covariances that lie behind the parameters and coefficients in such models.

    After careful consideration and considerable debate, the DPG agreed upon a common definition of capital at risk. The definition of capital at risk is the maximum loss expected to be exceeded with a probability of one percent over a two-week period.

    While this definition is quite conservative and appears to be gaining favor in the supervisory community, it nevertheless does not encompass the low probability events that are of greatest concern to many observers. It is, of course, possible to define capital at risk in statistical terms that would capture more of these low probability events than are captured by the above definition. Obviously, there is a point on this continuum where the estimate of capital at risk could meet virtually all (but never all) of these low probability events, but the implied capital needs at such a point would not be necessary and would not be commercially viable in terms of the ability of any firm to earn the return needed to attract and retain such a level of capital.

    In all of these circumstances, the one percent/two-week point for estimating capital at risk due to price changes in portfolios of OTC derivative products is a workable balance between practicality and prudence, with a clear tilt toward conservatism.

  2. Minimum Standards for Models

    Because the techniques used to estimate capital at risk are based on proprietary models, they will obviously differ in their technical and behavioral characteristics. Because of this, firms that compete with each other as well as supervisory authorities must have an appropriate degree of confidence that all of the models possess broadly similar performance characteristics. To accomplish this objective, the DPG instructed its capital working group to work with the risk managers and model experts at each of the six firms in order to develop certain minimum standards that any model must possess if it is to be used for the purpose of estimating capital at risk as outlined above. Those minimum standards include the following:

    1. A model should capture all of the material sources of market risk which might impact the value of the positions under consideration. While it is impossible to enumerate all such risks in advance, there is a group of core risk factors (outlined below) which all models should capture.

    2. In addition to the capital at risk calculation, firms will report the current market exposures for certain specified core risk factors. This information will be useful to the agencies both as a reality check on the capital at risk numbers (though in themselves, the core risk factors do not provide enough information to derive capital at risk) and to enable the agencies to evaluate issues of systemic risk.

      The core risk factors should include: (i) parallel yield curve shifts, (ii) changes in the steepness of yield curves, (iii) parallel yield curve shifts combined with changes in the steepness of yield curves, (iv) changes in yield volatilities, (v) changes in the value of equity indices, (vi) changes in equity index volatilities, (vii) changes in the value of key currencies (relative to the U.S. dollar), (viii) changes in foreign exchange rate volatilities, and (ix) changes in swap spreads in at least the G-7 countries plus Switzerland.

      The model would compute the change in value of all positions as the result of specified market movements (shown below) both up and down in each market. This process would capture the delta and gamma associated with each risk factor. These moves are all large moves relative to an overnight time horizon, and are not intended to be equally likely. The specified movements are:

      1. parallel yield curve shifts of 100 basis points up and down;

      2. steepening and flattening of the yield curves (2's to 10's) by 25 basis points;

      3. each of the four permutations of a parallel yield curve shift of 100 basis points concurrent with a tilting of the yield curve (2's to 10's) by 25 basis points;

      4. increase and decrease in all 3-month yield volatilities by 20 percent of prevailing levels;

      5. increase and decrease in equity index values by 10 percent;

      6. increase and decrease in equity index volatilities by 20 percent of prevailing levels;

      7. increase and decrease in the exchange value (relative to the U.S. dollar) of foreign currencies by 6 percent, in the case of major currencies, and 20 percent, in the case of other currencies;

      8. increase and decrease in foreign exchange rate volatilities by 20 percent of prevailing levels; and

      9. increase and decrease in swap spreads by 20 basis points.

    3. For both the capital at risk computations and the core risk factor computations, the models may use historical or market-implied volatilities and correlations. There are many generally accepted approaches to the estimation of such volatilities and correlations. (In general, for trade entry purposes, transactions, contracts and instruments within a firm's portfolio will not be entered into the model on an aggregate basis as a single category of transaction, contract or instrument, unless the individual transactions, contracts and instruments within the category exhibit an appropriately high degree of positive price correlation.) Regardless of the method used, the estimates of volatilities and correlations should apply over the forthcoming two-week period. Of course, the DPG recognizes that volatilities and correlations may break down in stress scenarios.

  3. Estimating Capital At Risk for Current and Potential Credit Risk

    While credit risk and market risk can be related, they are quite different. Credit risk, for example, is unique to a particular counterparty, or group of counterparties, whereas market risk relates to changes in classes of financial asset or in OTC derivative product prices. Moreover, where appropriate internal controls are in place, actual defaults of economic entities, especially those entities which are users of OTC derivative products, are more likely to arise from factors other than market risks associated with OTC derivative products. Finally, with relatively few exceptions, defaults of major firms generally occur with some warning, while sharp changes in financial asset prices can occur with little notice.

    For all of these reasons, the design and technical characteristics of models are not yet well-equipped to estimate the likelihood of individual counterparty default, except to the extent that such defaults may be related to changes in financial asset or OTC derivative product prices or positions. Accordingly, as described below, the DPG had to develop a hybrid approach to estimating the credit risk associated with portfolios of OTC derivative products.

    The DPG recognized that an approach to estimating credit risk must involve two measures. First, it must address current exposures. Second, it must anticipate potential exposures.

    1. Current Credit Risk

      The approach adopted by the DPG for estimating the capital at risk associated with current credit risk exposures is straightforward. The capital at risk for these purposes would be the net replacement cost by counterparty multiplied by the applicable historic default ratios published by the rating agencies (based on the firm's internal credit ratings). The historic default ratio used would take into account an estimate of average maturity, based on the types of contracts outstanding, and the credit rating of the individual counterparty. However, in no case (including the estimation of potential credit risk as described below), would a default ratio below 0.001 be used. This methodology would permit bilateral netting with individual counterparties where such netting arrangements are legally enforceable in accordance with accepted norms. This methodology would also permit reduction of exposures to individual counterparties if they are collateralized in an appropriate manner. A hypothetical illustration of the application of this methodology is contained in Appendix II.

    2. Potential Credit Risk

      Reaching agreement on the approach to estimating future or potential credit risk exposures proved to be much more difficult. The DPG considered and rejected the interim use of the notional value/add-on grid that has been used by banking supervisors for some time. This approach was rejected primarily for two reasons: first, it was seen as inconsistent with the DPG's goal to move as quickly as possible in the direction of model-based estimates of credit exposures; and second, there is great reluctance to using notional values as a part of the calculation of potential credit exposure because of a concern about implicitly sanctioning the notional value concept as a relevant measure.

      The DPG engaged in extensive discussion on the subject of potential credit risk and concluded that a dynamic methodology based on the application of counterparty default ratios to model-derived estimates of value at risk would be preferable to more static, fixed-charge "add-on" methodologies.

      Under the approach adopted by the DPG, the OTC derivatives portfolio of each counterparty (regardless of whether the aggregate net replacement value of the counterparty's portfolio is positive or negative) would be subject to the capital at risk calculation to be used for estimating portfolio market risk. Thus, for each counterparty, a firm would calculate, as a proxy for potential credit exposure to the counterparty, the maximum loss likely to be exceeded with a probability of one percent over a two-week period, ignoring for these purposes any legally enforceable right of the firm to call for the transfer of collateral on a prospective basis.

      The resulting estimate of potential credit exposure would be multiplied by the counterparty's applicable default ratio, based on the firm's internal credit rating for the relevant counterparty, to produce an estimate of the potential credit risk associated with the counterparty's OTC derivatives portfolio. Aggregate potential credit risk would equal the sum, for all counterparties, of these calculations of individual counterparty potential credit risk. Aggregate potential additional credit risk would equal aggregate potential credit risk less the aggregate current net credit exposure to all OTC derivatives counterparties. (See Appendix II.)

  4. Conclusion

    It must be recognized that the resulting estimates of capital at risk associated with potential credit risk (as well as market risk) are not intended, in themselves, to constitute capital standards. Rather, they are imperfect measures of the potential losses associated with credit and market risk against which judgments by managers and supervisors must be made as to the adequacy of capital resources. This is the essence of the supervisory oversight process.

    It must also be recognized that the quantitative analysis of credit risk assumes the existence of a diversified portfolio of counterparty credit risk. Accordingly, in the event that a firm's portfolio is concentrated in particular counterparties, particularly those rated below investment grade (a circumstance that would be revealed by the credit concentration information to be supplied under the proposed enhanced reporting framework), it may be necessary for senior management and supervisory authorities to focus heightened attention on such concentrations.

    As noted earlier, this approach for estimating the capital at risk associated with current and potential credit risk is seen by the DPG as an interim approach, pending the development of more rigorous model-based techniques for evaluating credit risk. As such, the DPG contemplates further refinements over time, particularly in the case of potential credit risk. The DPG also contemplates cooperating with requests by the agencies from time to time to compute potential credit risk using other methodologies.

    Nonetheless, the DPG believes that the foregoing approach is superior to a fixed-charge "add-on" approach in significant respects. By using actual historic default ratios, the resulting credit risk estimates more accurately reflect differences in the credit quality of individual counterparties. In addition, the model-derived estimates that serve as a proxy for credit exposure are dynamic and empirically-based and, accordingly, these estimates will tend to be more accurate than fixed charges in reflecting the market risk associated with specific products, particularly in the context of markets characterized by significant volatility.

IV. Audit and Verification Procedures for Models

While the common approach to estimating capital at risk and the minimum standards go a long way in addressing the problems inherent in using models for the purposes contemplated in this exercise, the DPG was strongly of the view that credibility considerations also pointed to the need for common approaches to audits and verification of the technical and performance characteristics of the models. Such procedures, which rely on both internal and external audits, are summarized below.

The firms would be responsible for making all computations necessary for purposes of assessing risk in relation to capital on a regular basis and be able to provide such computations on a current basis upon request. The firms would ensure that inventory pricing and modeling procedures are reviewed at least annually by independent auditors or consultants. Internal auditing departments would perform reviews at intervals consistent with the firms' own audit policies. The external auditors would provide reports summarizing the results of their reviews, and each agency would receive the annual audit report covering these issues.

An integral part of the firms' reliance upon models and related systems is adequate control over their use. Each firm should maintain:

  1. documentation describing each model used, including designation of those responsible for developing and testing the model;

  2. documentation adequate to assess the comparability of computations between firms, including the volatilities and correlations assumed and a description of relevant databases used;

  3. procedures for recording alterations to the model made after its introduction;

  4. policies and procedures for maintaining the systems supporting the models; and

  5. a disaster recovery plan for the relevant systems.

Additionally, each firm would be required to document its procedures for testing the models used for product pricing and portfolio evaluation prior to their use for purposes of evaluating risk in relation to capital.

To facilitate audits, the output of each model would permit the identification of the empirical assumptions used by the model in computing market and credit risk.

Taken together, (i) the common approach to estimating capital at risk, (ii) the minimum standards, and (iii) the audit and verification procedures provide a solid base upon which models can be used effectively by both management and supervisors to analyze better and more rigorously and to control market risk associated with OTC derivatives activities. However, at least for the present, there are limits on the extent to which models can be used to develop estimates of capital at risk for credit risk.

V. Oversight Policies and Practices

Under the framework of initiatives developed by the DPG, the SEC and the CFTC will soon begin to receive from the participating firms, on a confidential basis, ( 1 ) enhanced quantitative reports, (2) estimates of capital at risk for both market and credit risks, and (3) a series of core risk factor stress tests of the portfolios.

The DPG believes that the resulting information flows to the SEC and the CFTC will substantially enhance the Commissions' understanding of the OTC derivatives businesses and the related risks associated with those businesses.

In the face of this dramatically improved flow of information to the SEC and the CFTC, the DPG has concluded that rigid formulas linking capital at risk (or some multiple of capital at risk) to capital levels should be avoided. Thus, for a transitional period, the DPG advocates an oversight approach that requires the agencies and senior managers of the individual firms to evaluate the estimates of credit and market risk in relation to capital on a case-by-case basis, taking into account the factors listed below:

  • each firm's organizational structure;

  • each firm's internal controls and risk management systems;

  • the quality of management at each firm;

  • each firm's desired risk profile and credit standing;

  • each firm's actual day-to-day loss experience;

  • each firm's ability to manage risk as indicated by its ability to perform and document both stress and contingency analysis; and

  • each firm's overall compliance with all of the policies and practices advocated by the DPG.

As the firms and the agencies gain more experience with this overall framework, and depending on the evolution of thinking and policies within the international community of supervisors, this approach to oversight may evolve. For now, however, gaining experience with these new tools, techniques, and approaches is the key to better risk management and better oversight.

Consistent with that objective, on a case-by-case basis, the senior management of each firm should arrange periodic meetings with the agencies in order to evaluate the information and analyses being provided to the agencies, especially as it pertains to the views of both management and the agencies on the level of market and credit risk in relation to capital, in the context of the entire integrated framework advocated by the DPG.

* * *

An illustration of the estimation of credit risk
is contained in AppendixII.

Contact us * Risk Library * Documents by Author * Additional Regulatory Documents * Framework for Voluntary Oversight