Alert July 29, 2008

Basel Committee, in Conjunction with IOSCO, Publishes Proposals on Capital for Incremental Risk in Trading Book and Value-at-Risk Models

In furtherance of the recommendations set forth in the Financial Stability Forum’s report in April, 2008 (see the April 15, 2008 Alert), as well as the perceived significant losses financial institutions have incurred in their trading book, the Basel Committee on Banking Supervision (the “Basel Committee”) published a proposal providing guidelines for computing capital for incremental risk in the trading book (“IRC Proposal”), and a proposal with certain changes to the Basel II framework (the “Framework”) concerning internal Value at Risk (“VaR”) models (the “VaR Proposal”).  Both proposals were developed jointly with the International Organization of Securities Commissions and will apply to investment firms as well as to banking institutions.  Comments on both proposals are due by October 15, 2008.

IRC Proposal

Purpose and Scope.  The purpose of the IRC Proposal is to address a number of perceived shortcomings in the risks captured in, and associated capital charges for, the current 99%/10-day Value at Risk (“VaR”) framework.  In October, 2007, the Basel Committee published for comment guidelines meant to compute capital for incremental default risk in the trading book.  In light of market events since then, the Basel Committee decided to expand the scope of the capital charge to include not only default risk, but also other sources of price risk, such as significant moves in credit spreads, referred to in the IRC Proposal as the incremental risk charge (“IRC”), as more fully defined below.  The IRC would apply only to banks that are subject to the 1996 Market Risk Amendment (“MRA”) and that seek to model specific risk in the trading book.  Banks subject to MRA but not modeling specific risk would continue to be subject to the current rules.

Calculation of the IRC.  Under the IRC Proposal, the trading book charge would consist of three components:  (1) a general market risk charge; (2) a specific risk charge (each of (1) and (2) using the current VaR model); and (3) the IRC.  The separate surcharge for specific risk under the current trading book framework would be eliminated. 

As to the IRC, it would include all positions subject to the MRA (e.g., debt securities, equities, and securitizations), except for positions whose valuations depended solely on commodity prices, foreign exchange rates, or the term structure of default-free interest rates.  IRC covered risks would include default risk, credit migration risk, credit spread risk, and equity price risk.  The IRC must capture these risks regardless of whether they are already captured in a bank’s 10-day VaR. 

The IRC must incorporate these risks at the 99.9 percent confidence interval over a capital horizon of one year, taking into account the “liquidity horizons” applicable to individual trading positions or sets of positions.  The liquidity horizon represents the time required to sell a position or to hedge all material risks in a stressed market, and the IRC Proposal sets floors as to these horizons based on the greater of specified time periods and the bank’s actual experience.  For example, the liquidity horizon for equities generally is the greater of one month or the bank’s actual trading experience; for re-securitizations, the floor is one year. A bank would be permitted to assess liquidity on a position or bucketed basis.

Intra- and inter-obligor hedging may be incorporated into an IRC model, subject to specific guidelines set forth in the IRC Proposal.  As indicated above, the IRC Proposal currently does not offset for double-counting of risks inside both the current VaR model and the IRC.  However, the IRC Proposal does provide for the possibility of a bank developing an approach to address the double counting issue, subject to regulatory approval.

Internal Risk Measurement Modeling.  The IRC Proposal does not mandate any specific modeling approach for the IRC, and indeed anticipates that banks will develop different IRC approaches.  The bank’s ultimate approach is subject to the “use test” (i.e., it must be used in the bank’s risk management framework).  The IRC Proposal states that ideally the principles in the IRC Proposal would be incorporated into internal models for calculating trading book risk, however this is not a requirement.  Nonetheless, the bank must be able to demonstrate that its internal approach provides a conservative estimate of capital charges incurred for the risks identified in the IRC Proposal.  Banks must calculate the IRC at least weekly.

Validation.  The IRC Proposal provides that banks would apply the same validation procedures developed for the Basel II and MRA rules for the IRC.  Banks would not be required to use the backtesting regime applied to trading risk VaR models, but quantitative validation should remain important to a bank’s internal validation process.

Phase-in/Disclosure.  If a bank already has a specific risk model, in order to retain that model after January 1, 2010 it would have to, by that date, have an approved IRC model incorporating at least default and migration risks, and would have to have an IRC model incorporating all risks by January 1, 2011.  The IRC Proposal does provide a “temporary fallback option”, subject to supervisory approval, for those institutions unable to meet the January 1, 2010 deadline.

As to disclosure, the IRC Proposal would require that the IRC be disclosed publicly alongside and at the same frequency as disclosures of a bank’s market risk capital calculation. 

VaR Proposal

Purpose and Scope.  The purpose of the VaR Proposal is to create an updated framework for internal VaR modeling that is responsive to risks to banks that have been brought to light by recent market events.  Banks would be expected to comply with the revised Framework in order to receive approval for using internal models for the calculation of market risk capital requirements.

Proposed Changes to VaR Modeling Methodology.  The VaR Proposal would require banks to justify to their supervisors any factors used in pricing which are not included in the calculation of VaR.  In addition, under the VaR Proposal the bank’s VaR model would have to capture certain nonlinearities beyond those inherent in options (e.g., mortgage-backed securities, tranched exposures or n-th loss positions), as well as capturing correlation risk and basis risk.  Moreover, under the VaR Proposal, the bank’s supervisor would have to be satisfied that proxies were being used that show a good track record for the actual position held.  The VaR Proposal also would require a bank that uses VaR numbers calculated according to holding periods shorter than ten days that are scaled up to ten days to periodically justify the reasonableness of its scaling-up method to the satisfaction of its supervisor.

Under the VaR Proposal, banks would also be required to use hypothetical backtesting at least for validation, to update the market data used in the VaR modeling at least monthly and to be in a position to update it in a more timely fashion if deemed necessary.  Finally, the VaR Proposal clarifies that it is permissible to use a weighting scheme for historical data that is not fully consistent with the requirement that the “effective” observation period must be at least one year, as long as that method results in a capital charge at least as conservative as that calculated with an effective observation period of at least one year.

Proposed Changes to Prudent Valuation Methodology.  To complement the above changes to the VaR methodology, the Basel Committee has made the language with respect to prudent valuation for positions subject to market risk more consistent with existing accounting guidance.  The VaR Proposal clarifies that a bank’s supervisors would retain the ability to require adjustments to a bank’s current valuation of a position beyond those required by financial reporting standards where there is uncertainty around the liquidity of that position.  Furthermore, the VaR Proposal indicates that actual market prices or observable inputs should be considered for the purposes of valuation even when the market is less liquid than historical market volumes and that any use of mark-to-model valuation must be shown to be prudent.