Risk Library
   Documents by Author
     Additional Regulatory Documents
       Net Capital Rule - A Concept Release
         II. Alternatives to the Current Financia...
           A. Modify Current Haircut Approach
           B. Model Based Approach










 

II. Alternatives to the Current Financial Responsibility Regime

B. Model Based Approach

1. Generally A number of broker-dealers, primarily those with large proprietary securities portfolios, have indicated to the Commission that they may be willing to incur the expenses associated with developing and using statistical models to calculate haircuts on their securities portfolios. Under a model based net capital rule, in lieu of taking fixed percentage haircuts, a broker-dealer would use either an external or internal model as the basis for a market risk charge and take a separate charge, or charges, for other types of risk, such as credit risk and liquidity risk.

The Commission could allow firms to calculate market risk capital charges according to external models for specific types of securities that are similar to the options pricing models allowed under Appendix A to the Rule. The benefit of an external model is that all firms would be utilizing the same model. However, the Commission could have difficulty finding a third party (comparable to the Options Clearing Corporation for listed options) that would have access to all the data necessary to facilitate external security-specific models for securities other than options.

With respect to internal models, the Commission would need to prescribe certain minimum quantitative and qualitative standards that a firm's model would have to meet prior to that firm using its internal model for regulatory capital purposes. Currently, several large firms use value at risk ("VAR") models as part of their risk management system. These firms typically utilize VAR modelling to analyze, control, and report the level of market risk from their trading activities. Generally, VAR is an estimate of the maximum potential loss expected over a fixed time period at a certain probability level. For example, a firm may use a VAR model with a ten-day holding period and a 99 percentile criteria to calculate that its$100 million portfolio has a potential loss of $150,000. In other words, the firm's VAR model has forecasted that with this portfolio the firm may lose more than $150,000 during a ten-day period only once every 100 ten-day periods.

In practice, VAR models aggregate several components of price risk into a single quantitative measure of the potential for loss. In addition, VAR is based on a number of underlying mathematical assumptions and firm specific inputs. For example, VAR models typically assume normality and that future return distributions and correlations can be predicted by past returns.14

Given the increased use and acceptance of VAR as a risk management tool, the Commission believes that it warrants consideration as a method of computing net capital requirements for broker-dealers. However, while VAR can be used to manage market risk, broker-dealers that rely solely on VAR for risk management may not have a comprehensive risk management program. VAR models, unlike haircuts, do not typically account for those risks other than market risk, such as credit risk, liquidity risk, and operational risk. Broker-dealers that utilize VAR models should therefore use additional techniques to manage those risks.

Further, while VAR may be useful in helping broker-dealers project possible daily trading losses under "normal" market conditions, VAR may not help firms measure the losses that fall outside of normal conditions during times of market stress. For example, VAR models may not capture possible steep market declines because these models typically measure exposure at the first percentile (or the fifth percentile) and steep market declines are, by definition, below the first percentile. In addition, the most common VAR approaches may pose a problem for those portfolios that utilize options or other products with non-linear payoffs.15

The purpose of the Commission's net capital rule is to protect markets from broker-dealer failures and to enable those firms that fall below the minimum net capital requirements to liquidate in an orderly fashion without the need for a formal proceeding or financial assistance from the

Securities Investor Protection Corporation. The Commission believes that market risk charges must adequately protect a broker-dealer during severe market stress, whether that stress occurs on only one day or over a period of several days, such as the drop in equity prices during the October 1987market break or the Mexican debt crisis in 1994. Because VAR models do not typically reserve capital for severe market declines, it may be necessary to impose additional safeguards to account for possible losses or decreases in liquidity during times of stress. This may include the use of a multiplier or the use of stress tests that firms could apply to their portfolios. A multiplier could be used to account for the other risks in a firm's portfolio that are not captured by VAR models, such as operational, settlement, or legal risk. On the other hand, stress testing could provide a more complete picture of the portfolio's sensitivity to changing market conditions and a more accurate representation of capital needs than a simple multiplier.

The primary advantage of incorporating models into the net capital rule is that a firm would be able to recognize, to a greater extent, the correlations and hedges in its securities portfolio and have a comparatively smaller capital charge for market risk. Accordingly, if the Rule is amended to permit models to be used to calculate market risk in lieu of taking the haircuts currently imposed by the rule, the Commission solicits comment on how the Rule may be modified to include separate capital requirements to cover sources of risk other than market risk. Other issues associated with incorporating models into the Rule are the need for management controls necessary to ensure that the firm is collecting accurate and comprehensive information on its proprietary positions and the effectiveness of those controls to monitor the risk assumed by the firm.

2. Two Tiered Approach One way that the Commission could incorporate models into the net capital rule would be to have different net capital requirements based on certain standards ("Two Tiered Approach"). Under the Two Tiered Approach, broker-dealers meeting certain minimum threshold levels would be required to use models to determine capital compliance. For example, broker-dealers with net capital exceeding a certain amount and currently using models for in-house risk management purposes could use models to determine their market risk capital charge under prescribed circumstances. Firms with less than the prescribed level of net capital and those firms with net capital greater than the prescribed level but not using models for risk management could be required to continue to follow the current Rule's haircut methodology. These haircut percentages could either be the same as the current percentages or modified versions.

A Two Tiered Approach potentially has two primary benefits. First, the Commission could structure a Two Tiered Approach to limit the use of models to those firms that currently use sophisticated models such as VAR, thereby not requiring other firms to incur the cost of implementing such models. Second, the Commission could design a Two Tiered Approach that establishes appropriate limits on which firms can utilize models to determine capital compliance.

A potential weakness of a Two Tiered Approach is that it could inhibit competition between large and small firms because models may give large firms more flexibility in determining their net capital requirements.

However, this advantage could be small if smaller firms did not have to incur the start-up and maintenance costs associated with models and the risk management infrastructure to support their use. Additionally, a Two Tiered Approach could still allow firms with simple portfolios to easily calculate the applicable haircuts on their portfolios.

3. Base Approach with Pre-Commitment Feature Another option for incorporating models into the Rule could be to combine the current haircut methodology using fixed percentage haircuts with a model-based approach (the "Base Approach"). The Base Approach could combine the strengths of both haircuts and models and at the same time possibly address the weaknesses of each. The Base Approach would include three primary components. First, broker-dealers could be required to maintain a certain minimum base level of net capital for each of their business activities, similar to the minimum requirements under the current rule. For example, higher capital levels could apply to broker-dealers that hold customer funds and securities as opposed to those firms that only introduce customer accounts to clearing firms. Second, broker-dealers could take a fixed percentage haircut for each security in their portfolio. This haircut would be similar to the haircut requirements under the current net capital rule; however, the size of the haircut would be lower due to the additional charge for market risk obtained from the third component.

The third component of the Base Approach could consist of a capital charge based on the firm's model and include a pre-commitment feature that could require a broker-dealer to take capital charges based on the realized performance of its models ("pre-commitment feature"). The pre-commitment feature could have two steps. First, at the start of a pre-determined time period (i.e., one month or one quarter), a broker-dealer could be required to represent that its losses, as computed by its model, would be within certain parameters over the fixed time period. Second, at the conclusion of each fixed time period, the firm's minimum net capital level could increase by an amount equal to the difference between the actual portfolio gains and losses and those projected based on its model. These additional capital contributions would be required because differences between the actual results and those projected by the model could indicate that the firm's models may not be accurately assessing the risk of the firm's portfolio. By incorporating haircuts and models into the Base Approach, the inherent strengths and weaknesses of each could potentially offset each other. Additionally, the Base Approach may be a viable capital standard for firms with diverse portfolios and those that use more sophisticated methods of risk management. The pre-commitment feature would create additional incentives for broker-dealers to manage risk effectively. On the other hand, a Base Approach may be too complicated for firms to apply. In balance, however, the Base Approach could potentially provide firms with flexibility in developing models and control systems, encourage the development of accurate forecasts, and still ensure that firms reserve sufficient amounts of net capital.

4. Comments on the Potential Use of Models The Commission solicits comment on the following specific topics, including the appropriateness of using proprietary models generally and the recent initiatives of both the DPG and the U.S. Banking Agencies.

    a. Models as a means to determine broker-dealer regulatory capital

Question 8: Should the Commission permit the use of models to calculate regulatory capital for registered broker-dealers? If yes, please explain whether the Commission should allow firms to utilize internal models or whether the Commission should establish an external model approach similar to the treatment of options under Appendix A to the Rule.

Question 9: If the Commission permits the use of internal models, should the model conform to certain objective criteria, or should they be subjective? When could the assumptions upon which models rest be challenged? Should internal or external auditors periodically review and approve the model and its application? If so, how much should regulators rely on auditors' application of models? Could the self-regulatory organizations adequately surveil and examine for net capital compliance utilizing models?

Question 10: Should the Commission impose limits on the types of firms that can use models? Should there be certain additional minimum criteria a firm must satisfy in order to use a proprietary model? Should firms that meet the minimum criteria for using models have the option of using an alternate standard approach (i.e., not using models) to calculate regulatory capital? If so, what should that approach be?

Question 11: Is VAR an appropriate method of using models as the basis for calculating capital requirements for broker-dealers? The Commission understands there are several approaches to calculating VAR that are currently used by firms (e.g., Monte Carlo, variance/covariance, and historical simulation approaches). Given the various methods, the Commission seeks comment on whether minimum criteria should be established for models used for regulatory capital purposes? If not, how can the Commission provide for the ability to compare levels of risks among firms or understand the significance of levels of risk reported by firms when determining their net capital requirements?

Question 12: The Commission believes that any approach that uses models for setting regulatory capital requirements should result in broadly consistent results for firms with similar portfolios. Can consistent results for similar portfolios be obtained without the Commission requiring firms to use a standard model? How else can consistency of capital standards among firms with similar portfolios be achieved?

Question 13: Some firms use different types of statistical models to measure risk from different types of businesses, such as fixed income securities and foreign equities. Should the Commission permit firms to use more than one model to calculate regulatory capital? If yes, would the inefficiencies in each model get accentuated or mitigated when the results of the different models are aggregated?

Question 14: Should the Commission allow the use of models gradually(i.e., first allow models for debt securities, then allow models for equity securities and other securities)?

Question 15: What will be the costs of implementing models? How do the costs of implementing models compare to the current costs of computing net capital? At what level would it be economical for firms to try to use models? How do the start-up costs of implementing models compare to the ongoing costs of managing models incurred by firms that currently use models? How does the availability (or anticipated future development) of software packages and databases impact cost estimates? Will the costs of implementing models be a barrier to firms not currently using models? Please provide relevant data to support your response.

Question 16: Will firms not currently using models be at a competitive disadvantage to those firms that currently use models? Please provide relevant data to support your response.

Question 17: If the Commission permits the use of models, what additional reporting or record keeping requirements would the Commission need to impose on broker-dealers using models? Should firms using models have to file additional reports with the Commission or their DEA? Should the Commission amend its books and records rules to require firms using models to maintain certain books and records that they are currently not required to maintain? How can the Commission ensure that it has access to information regarding a firm's models that is not maintained by the broker-dealer (i.e., information maintained at an unregistered entity)? What measure could the Commission require to ensure broker-dealers would not be able to modify the model (or data inputs) to avoid falling out of net capital compliance? Should the Commission require models to be stored with third-parties subject to escrow arrangements?

Question 18: If the Commission permits the use of models, should firms using models be subject to modified forms of Commission and DEA inspections? Should the models themselves be subject to review and approval by the Commission or DEA?

    b. Abnormal Market Conditions

Question 19: Because the purpose of VAR is to provide an estimate of losses over a short period under normal conditions, is it possible for VAR models to ensure an adequate capital cushion during unusual market stress or structural shifts in the economy given the nature, size, and liquidity of a broker-dealer's portfolio? Given the complexity of models, could an accurate and rapid assessment be made of a firm's true financial condition? Please provide relevant data to support your response.

Question 20: Would models be more effective during times of severe market fluctuations if stress testing were required? Should the Commission specify what stress tests should be used by the firms? Please provide relevant data to support your response.

Question 21: If stress testing were required, should a firm be required to use the same parameters when conducting stress testing on each of its business units (i.e., apply the same levels and stress the same movements in the relevant securities, markets, and indexes)?

Question 22: If stress testing were required, should a firm be required to test its models based on a predetermined number of volatile days of market movements (i.e., models would have to be stress tested based on the 100 most volatile days of market movements during the last ten years)?

Question 23: Should the results of stress testing impact the calculation of a firm's capital requirements (i.e., through the use of some type of multiplication factor)? Please provide relevant data to support your response. Question 24: Does the use of a minimum multiplier, as endorsed in the Basle Standard and by the U.S. Banking Agencies, adequately address risks arising from severe market movements? Please provide relevant data to support your response.

Question 25: Should back-testing (i.e., ex post comparisons between model results and actual performance) be required and, if so, to what extent? Should back-testing results be used to determine a multiplier for minimum capital amounts? Could back-testing results be used to raise minimum capital levels for the firms?

    c. Qualitative and Quantitative Criteria for Models

Question 26: Will setting minimum qualitative and quantitative criteria prevent a firm from adjusting its model to encompass changing market conditions, the firm's structure, or the firm's business lines?

Question 27: Two important components of models are the length of time over which market risk is to be measured and the confidence level at which market risk is measured. The definition of "capital at risk" as used in the DPG Framework is the maximum loss expected to be exceeded with a probability of one percent over a two-week period. Is this definition appropriate for regulatory capital purposes?

Question 28: What should be the minimum criteria for models, including pricing accuracy, correlations, netting factors, and observation periods? Please provide relevant data to support your response.

Question 29: Are the minimum standards for the use of models, the separate calculation of capital at risk due to shocks to the core risk factors, and the audit requirements used in the DPG Framework appropriate? Please provide relevant data to support your response.

Question 30: VAR models typically assume normality and that future return distributions and correlations will behave similar to the way they behaved in the past. For these reasons, the Commission needs to ensure that VAR models can withstand steep market declines. Other than by specifying minimum qualitative and quantitative criteria, how can regulators assure themselves that the proprietary models used by the firms are adequate for capital purposes?

Question 31: Should the Commission require that broker-dealers utilizing models manage these models from a risk management division that is separate from the firm's business divisions?

Question 32: Should the Commission require that broker-dealers utilizing models use the same model for both computing net capital and internal risk management purposes?

Question 33: Currently, firms utilize a wide variety of risk management techniques. Should the Commission mandate specific minimum risk management standards for firms that wish to use models?

Question 34: Should the Commission require that firms using models manage risk on either a firm-wide, legal entity, or business basis?

    d. Additional Risks

Question 35: Usually, VAR models do not handle options products well because the returns on an options portfolio are not typically normally distributed. How should the non-linear nature of options be adequately addressed? For firms with substantial options positions, is a standard approach (similar to the Commission's amendments to Appendix A of the net capital rule) more appropriate? Is the approach set forth in the Commission's recent amendments to Appendix A a viable alternative?

Question 36: Models typically measure losses by assuming that assets can be sold at current market prices. However, if a firm has a portfolio which includes illiquid assets, highly customized structured products (including, for example, some CMOs), or aged items, the Commission is particularly concerned that models may underestimate the true losses since these assets may have to be sold at a discount. Given the importance of liquidity risk, the Commission solicits specific comment with respect to how this risk should be addressed if models are permitted for regulatory purposes.

Question 37: Is it possible to include a credit risk analysis in a model based methodology? Please provide relevant data to support your response.

Question 38: As mentioned above, models may not properly account for additional risks, including credit risk, liquidity risk, operational risk, settlement risk, and legal risk. How should these additional risks be treated? Can the Rule be modified to include separate capital requirements to cover these sources of risk? Please provide relevant data to support your response.

Question 39: Is there an alternative to using a multiplier to account for operational risk, legal risk, and other risks that are difficult to quantify? Is the use of insurance to cover these risks a viable option? Please provide relevant data to support your response.

Question 40: In order for a firm to calculate VAR effectively, data must be aggregated from all its departments worldwide. Also, there is often incompatibility of trading and back-office accounting computer systems that operate from different regions of the world. How can this problem of integration be adequately addressed?

    e. OTC Derivatives Dealer

Question 41: Should the Commission amend the Rule so that all broker-dealers are eligible to use the methodology for calculating market and credit risk as in proposed Appendix F to the Rule?

Question 42: What minimum capital requirements should the Commission require a broker-dealer to meet to be eligible to use proposed Appendix F? Should the criteria be based on tentative net capital, net capital, or both? Are the $100 million tentative net capital and$20 million net capital requirements appropriate?

Question 43: Assuming that the Commission were to allow all broker-dealers to utilize Proposed Appendix F, what sections in Proposed Appendix F need to be modified for all broker-dealers? Are the market risk and credit risk sections in Proposed Appendix F appropriate for all broker-dealers? Are the qualitative and quantitative requirements for VAR models in Proposed Appendix F appropriate to VAR models used by non-OTC derivatives dealers?

    f. Two Tiered Approach

Question 44: Is a Two Tiered Approach a viable alternative to the current net capital rule? If so, what standards should the Commission utilize to determine which broker-dealers are required to utilize statistical models? Should the tier limits be based on capital, amount of customer business, level of proprietary trading, or some other factor(s)? Should these minimum net capital amounts be fixed dollar amounts or be based on financial ratios such as aggregate indebtedness or aggregate debit items as in the current rule? Please provide relevant data to support your response.

Question 45: Should the current haircut percentages be maintained? If not, what modifications should be made to the current haircut percentages? Please provide relevant data to support your response.

Question 46: What will be the impact on competition among firms indifferent tiers? In this regard, the Commission seeks comment on the effects of creating a two-tiered system from broker-dealers that do not currently use models in their risk management system and from broker-dealers that currently use models for risk management purposes but either lack sufficient capital or sufficiently diverse securities portfolios to use models for net capital purposes.

    g. Base Approach with Pre-Commitment Feature

Question 47: Is the Base Approach a viable alternative to the current net capital rule?

Question 48: Should the Base Approach only apply to firms that meet certain standards? If so, what are the appropriate standards?

Question 49: What minimum capital requirements should the Commission establish for certain broker-dealer activities? Should these minimum net capital amounts be fixed dollar amounts or based on financial ratios such as aggregate indebtedness or aggregate debit items as in the current rule? Should the current minimum levels be retained?

Question 50: What modifications should the Commission make to the current haircut percentages? Please provide relevant data to support your response.

Question 51: What should be the parameters for the pre-commitment feature? Should firms be penalized for differences between actual results and the results as projected by VAR models? If so, what criteria should be used to determine the additional capital requirements for these differences?

Footnotes:

(14) The Commission recognizes that there is a wide variety of secondary source information discussing both the positive and negative aspects of VAR. See

  • Philippe Jorion, Value at Risk: The New Benchmark for Controlling Market Risk(1996) (explaining how to use VAR to manage market risk) ;
  • JP Morgan, RiskMetrics-Technical Document(1994) (providing a detailed Description of RiskMetrics, which is JP Morgan's proprietary statistical model for quantifying market risk in fixed income and equity portfolios) ;
  • Tanya Styblo-Beder, VAR: Seductive but Dangerous, Financial Analysts Journal, September-October 1995, at 12(giving an extensive analysis of the different results from applying three common VAR methods to three model portfolios) ;
  • Darrell Duffie and Jun Pan, An Overview of Value at Risk, The Journal of Derivatives, Spring 1997, at 7(giving a broad overview of VAR models) ;
  • Darryll Hendricks, Evaluation of Value-at-Risk Models Using Historical Data, Federal Reserve Bank of New York Economic Policy Review, April 1996, at 39(examining Twelve approaches to value-at-risk modelling on portfolios that do not include options or other securities with non-linear pricing) ; and
  • Robert Litterman, Hot Spots and Hedges, Goldman Sachs Risk Management Series(1996) (giving a detailed analysis on portfolio risk management, including how to identify the primary sources of risk and how to reduce these risks) .

(15) See Autoro Estrella et al., Options Positions: Risk Measurement and Capital Requirements, Federal Reserve Bank of New York Research Paper number 9415, September 1994(evaluating different methods of measuring the market risk of options and analyzing the capital treatment of the market and credit risk of options) .

Contact us * Risk Library * Documents by Author * Additional Regulatory Documents * Net Capital Rule - A Concept Release * II. Alternatives to the Current Financial Responsibility Regime