In January, banks in the U.S. became subject to new capital requirements for market risk arising from trading activities. These requirements are notable because the capital charge is based on banks' internal value-at-risk models. They are based on VaR estimates calibrated to a common supervisory standard of a 10-day holding period and a 99th percentile confidence level. They distinguish between two forms of market risk: general market risk, which reflects the potential for losses from movements in the general level of market rates and prices, and specific risk, which reflects the risk from price movement relating to the characteristics of individual security issuers.
EFFECT ON CAPITAL LEVELS AND RATIOS
The effect on banks' required capital ratios is difficult to calculate precisely with the data currently in the public domain. Such calculations require both information on banks' VaR estimates--calibrated to the 10-day, 99th percentile supervisory standard--and information about distribution of trading assets and liabilities among various specific risk categories. It is possible to make a rough estimate of the impact by using information reported in banks' annual reports.
Table 1 shows a sample of large bank holding companies that presented annual average VaR estimates in their 1996 annual reports along with sufficient detail to identify the holding period and percentile underlying the estimate. All the estimates were based on a one-day holding period, with percentiles ranging from the 95th to the 99th. The divergence in these parameters, as well as other aspects of the estimates such as correlation assumptions, makes direct comparisons of these figures across institutions difficult.
Nevertheless, these figures suggest the impact of the market risk capital charge on required capital levels and capital ratios is likely to be small. We calculate the estimated increase in the level of required capital from the general market risk component of the new capital charge ranges as roughly 1.5-7.5% for these banking companies. We find the impact on the capital ratios is also fairly modest, with an average decline of about 30 basis points and 40 basis points in tier 1 and total capital ratios, respectively. These calculations are rough estimates and could differ significantly from the impact at the time the capital charge becomes effective. Such differences would reflect both estimation error in translating the reported figures to the supervisory standards and changes in the banks' portfolios.
Once we account for the capital treatment of specific risk, the overall impact of the market risk capital charge is likely to be smaller than our calculations suggest. Many traded debt and equity positions subject to the credit risk capital requirements under the original capital guidelines are now subject to specific risk capital requirements based on the output of banks' internal models. This "specific risk carve-out" will offset the impact of the additional general market risk capital charge. Unfortunately, the data needed to make reasonably precise estimates of this effect is not available currently. Given the significant positions some institutions hold in instruments that will become subject to specific risk capital requirements, this carve-out may result in a net reduction of capital levels for some institutions.
ADVANTAGES OF INTERNAL MODELS
Whatever the effect of the new standards on the level of overall required capital, capital requirements based on internal models should produce minimum regulatory capital charges closer to banks' true risk exposures. This closer relationship is important not only for determining the risk facing an institution at a particular moment, but also for tracing the evolution of risk over time. While VaR estimates underlying the market risk capital charge are useful for assessing the level of risk undertaken by a bank at a given moment, they potentially are even more beneficial for understanding changes in risk exposure over time. By extension, the key benefit of the market risk capital charge is that required capital levels will evolve with risk exposures over time.
In addition to tightening the link between risk exposures and capital requirements, a capital charge based on internal models may provide supervisors and the financial markets with a consistent framework for making comparisons across institutions. As the information in Table 1 makes clear, VaR figures presented in the annual reports of various bank holding companies are calculated using different parameters, especially the percentile of the loss distribution. These differences make comparisons across institutions difficult without additional calculations to convert the figures to a common basis. Typically, these calculations require assumptions that may be only approximately correct, introducing additional noise in the comparisons.
By contrast, the market risk capital charge provides a common standard for VaR estimates making comparisons across institutions easier and more reliable. VaR estimates underlying banks' capital charges will be based on a uniform set of parameters and accurately reflect assumptions and specifications of each bank's internal model. Further, the financial markets may gain information about the performance and accuracy of these models over time if banks make public the results of their backtests. While disclosure of the details of these results is purely discretionary, this information is consistent with the type of disclosures advocated in several recent discussion papers.
CHALLENGES FOR SUPERVISORS
The actual benefits to be derived from VaR estimates depend crucially on the quality and accuracy of the models on which the estimates are based. To the extent these models are inaccurate and misstate banks' true risk exposures, then the quality of the information derived from any public disclosure will be degraded. More important, inaccurate models or models that do not produce consistent estimates over time will undercut the main benefit of a models-based capital requirement: the closer tie between capital requirements and true risk exposures. Thus, assessment of accuracy is a key concern for supervisors.
IMPLICATIONS FOR THE FUTURE
Market risk capital requirements based on internal models have drawn considerable attention since the initial proposal for these requirements was released. During this time, supervisory interest in models has encouraged banks in the U.S. and abroad to direct resources and attention toward further developing these models and their full integration with the risk management process.
In coming years, some key issues facing banks in modeling will concern the extension of these models to cover a broader range of risks. For example, can quantitative risk models be applied to credit, operational, and legal risks? If so, should supervisors expand use of internal models to derive capital charges for these exposures? These issues already have surfaced in banks' efforts to model specific risk. In measuring specific risk, banks face a number of technical and conceptual problems--how to measure the probability and likely impact of events that occur infrequently and how to quantify the effects of complex events that depend on the interrelated actions of many parties. These problems, at the frontier of thinking about regulatory capital and banks' internal capital allocation, need to be resolved if quantitative risk models are to be used systematically to gauge other forms of risk.
At present, banks are still in the early stages of developing methods for quantifying other risks and for integrating these into a unified capital allocation framework. Understanding the ways models can and cannot be used is clearly one of the most significant challenges facing banks and their supervisors today. The market risk capital requirements may further this understanding by providing a test case for the supervisory use of internal models.
This week's Learning Curve was written by Darryll Hendricksand Beverley Hirtleof the Federal Reserve Bank of New York. It was adapted from "Bank Capital Requirements for Market Risk: The Internal Models Approach," published in the December issue of the Economic Policy Review. The opinions expressed in this article are those of the authors and do not necessarily reflect the views of the Federal Reserve Bank of New York or the Federal Reserve System.
