Sources of Risk
   Overview: Credit Risk
   Overview: Settlement Risk
   Overview: Market Risk
   Overview: Other Risks


Sources of Risk

Overview: Market Risk

Latest Update January 2000

From January 1st, 1998, internationally-active banks in G-10 countries had to maintain regulatory capital to cover market risk. This is the risk to an institution's financial condition resulting from adverse movements in the level or volatility of market prices of interest rate instruments, equities, commodities and currencies. Market risk is usually measured as the potential gain/loss in a position/portfolio that is associated with a price movement of a given probability over a specified time horizon. This is typically known as value-at-risk (VAR). An institution with a 10-day VAR of $100 million at 99% confidence will suffer a loss in excess of $100 million in one fortnightly period out of 20, and then only if it is unable to take any action to mitigate its loss.

Financial institutions have always faced the risk of losses in on and off-balance-sheet positions arising from undesirable market movements. However, the sharp increase of proprietary trading in many banks has heightened the need among regulators to ensure that these institutions have the management systems to control and the capital to absorb the risks posed by market-related exposures. But as a risk, market risk only gained a high profile when the Basle Committee on Banking Supervision published "The Supervisory Treatment of Market Risks" in April 1993. Although this consultative proposal was heavily criticised and has since been superseded, it was important because it sought, for the first time, to extend the 1988 Capital Accord for credit risks to incorporate market risks. Banks, it was proposed, have to set aside capital to cover the price risks inherent in their trading activities. This document put forward a standardised measurement framework to calculate market risk for interest rates, equities and currencies. For interest-rate related instruments and equities, this framework is based upon a "building-block approach" which differentiates capital requirements for specific risk from those for general market risk. Although the proposal was criticised as overly simplistic and did not take into account option curvature, it is worthwhile reading because it puts into perspective later regulatory documents and the ultimate framework which was adopted.

The "Amendment to the Capital Accord to Incorporate Market Risks (1996)" is the Basle Committee's pronouncement on capital charges for market risk. It sets forth two approaches for calculating the capital charge to cover market risks; the standardised approach (similar to the framework discussed above) and the internal models approach. For banks which choose to use their own internal models, the capital charge will be the higher of:

  • the previous day's value-at-risk;
  • three times the average of the daily value-at-risk of the preceding sixty business days.

The firm's value-at-risk must be computed on an instantaneous price shock equivalent to a ten-day holding period with a confidence interval of 99% (one-tailed) and based on a historical observation period of at least one year. For options, banks initially are be permitted to scale up their one-day value-at-risk measure by the square root of 10 for a certain period of time; however, they are expected to refine their modelling techniques over time. Financial institutions will also be allowed to use correlation offsets both within and across risk factors if they can convince their supervisory authority that they have a sound system for measuring correlations.

Eligible capital under both the internal models approach and the standardised approach consists of shareholders' equity and retained earnings (Tier 1 capital), supplementary capital (Tier 2 capital) as defined in the 1988 Accord, and a new Tier-3 capital, consisting of short-term subordinated debt. The latter is subject to the following conditions:

  • It must have an original maturity of at least two years and will be limited to 250% of the bank's Tier 1 capital that is allocated to support market risk.
  • It is only eligible to cover market risk, including foreign exchange risk and commodities risk.
  • Insofar as the overall limits in the 1988 Accord are not breached, Tier 2 elements may be substituted for Tier 3 up to the same limit of 250%.
  • It is subject to a "lock-in" provision which stipulates that neither interest nor principal may be paid if such payment means that the bank's overall capital would then amount to less than its minimum capital requirement.

Banks which do not meet the qualitative criteria set down by the Basle Committee are not allowed to use their internal models and must use the standardised measurement framework first suggested in 1993, albeit with a few changes. The capital for their option positions can be calculated on a 'simplified', 'delta-plus' or 'scenario' approach, depending on whether the firm buys or/and sells options. They must also take into account commodities risk.

A seminal regulatory document on market risk is the Basle Committee's "Proposal to Issue a Supplement to the Basle Capital Accord to cover Market Risks (1995)." This paper is significant because it marks a significant shift in thinking on the part of banking regulators. This proposal is the result of two years' dialogue between the regulatory community and the banking industry over the measurement of market risk (and resultant capital charges.) It represents a victory for the banking industry because the Basle Committee agreed to banks' requests that they be allowed to calculate market risk capital based on proprietary in-house models, rather than the standardised measurement framework mooted two years before.

Winning over the Committee was not easy because each in-house market risk model is different and there is no definitive standard on the inputs into these models. By setting out its concerns and its reasons for recommending certain quantitative standards, the Basle Committee clearly explains some of the complex issues in the measurement of market risk. They include the thorny problems of how to account for the non-linear price behaviour of options, the holding period over which the value of positions can change, the historical time horizon over which the prices are observed, the confidence interval for the level of protection judged to be prudent, and whether correlation offsets should be allowed when aggregating within and across risk classes.

Many banks calculate one-day VAR numbers which are then compared with actual daily profits and losses. It is much easier for them to scale up these numbers by the square root of 10 to meet the 10-day holding period laid down by the Basle Committee than to start afresh. But scaling up daily numbers to reflect a longer unwind period is fine for linear products such as forwards and swaps but not good enough for options and products with embedded options. Extrapolating daily VAR numbers does not capture the gamma or curvature risk of such products. The best way of capturing this curvature is to use longer holding periods which explains why the Basle Committee opted for a 10-day holding period. But the Committee wanted to limit industry burden in complying with its pronouncements and thus decided that banks could scale up their one-day numbers for a limited period. These banks are however asked to assess the option risk in their portfolios by applying Monte Carlo simulation and/or stress testing.

A paper entitled "An Internal Model-Based Approach to Market Risk Capital Requirements (1995)", explains the considerations which have to be borne in mind when specifying the risk factors in a market risk measurement system. The system should be broad enough to capture all the risks inherent in a portfolio. These risks would typically include interest rate risk (outright, curve, spread), currency, equity, commodity and options. Generally the number of risk factors and the level of detail involved per asset class is greatest where the bank has the largest exposure. The paper also shows worked examples of value-at-risk calculations using the three main approaches:(i) a variance-covariance approach (ii) a Monte-Carlo simulation (iii) an historical simulation, as well as alternative approaches to incorporating the curvature of options.

Although the Basle Committee has accepted the validity of using internal models to calculate market risk capital, it felt that potential weaknesses in these models had to be provided for in the capital charge. These weaknesses exist because:

  • market price movements are not always normally distributed because they display 'fat tails', i.e. they tend to have a wider dispersion of extreme events.
  • models cannot adequately capture event risk arising from exceptional market circumstances
  • value-at-risk estimates are typically based on end-of-day positions and generally do not take account of intra-day trading risk
  • funding risk is not assessed in a value-at-risk measure
  • the past is not always a good approximation of the future, for example, correlation assumptions may break down.

So the Committee deemed that all internal value-at-risk numbers had to be multiplied by a minimum factor of three (a capital buffer), provided the predicted value-at-risk numbers accurately reflected realised daily profits and losses. If there were a significant discrepancy between actual trading and model-generated numbers, then a plus factor would be added to the minimum number of three. The regulators are thus using the threat of a plus factor to give financial institutions an incentive to building models with good predictive quality to help internal risk control and not ones with the sole purpose of facilitating a lower capital requirement.

The criteria for avoiding the plus factor are set out in the three zones contained in the document titled " Supervisory Framework for the Use of "Backtesting" in Conjunction with the Internal Models Approach to Market Risk Capital Requirements (1996)" . The three zones are green (no plus factor), yellow (plus factor of 0.4 to 0.85) and red (plus factor of 1.0). An institution falls into the green zone if the actual figures exceed the predicted numbers 4 or less times (out of 250 observations); yellow five to nine times, and red 10 or more times. This document also spells out the Committee's backtesting framework. Essentially, banks will have to formally test and account for exceptions on a quarterly basis using the most recent twelve months of data. So this means that the first backtesting programme will occur by year-end 1998.

Other reasons for capital buffers are justified by IOSCO in a 1998 document, titled "Methodologies for Determining Minimum Capital Standards for Internationally Active Securities Firms Which Permit the Use of Models Under Prescribed Conditions." The committee maintains that a capital cushion is needed to cover not only model risks but also operational and non-modelled risks.

The Basle Committee conducted a survey of 40 banks in nine countries for the third and fourth quarters of 1998 to assess whether the internal models approach generated sufficient capital cover for market risk. The survey results showed that for the period covered in the survey, a period of high market volatility, the internal models approach provided an adequate buffer against trading loss. In a report titled, "Performance of Models-Based Capital Charges for Market Risk (1999)", the Committee notes that none of the 40 institutions surveyed reported trading losses over any 10-day consecutive period that exceeded the capital requirement in force at the start of the period.

Almost half of the firms surveyed reported no instances where one-day losses exceeded the daily VaR estimate. Of those who experienced exceptions, the majority reported between one and four incidences. Three banks reported five and seven exceptions during the second half of 1998.

Given the increased use and acceptance of VAR as a risk management tool, the Securities and Exchange Commission of the United States believes that it warrants consideration as method of computing net capital requirements for broker-dealers. But since VAR can only be used to manage market risk, and thus to calculate charges for market risk, the SEC has to look for additional methods to calculate capital for other risks such as credit, liquidity and operations. The SEC is therefore soliciting comments from broker-dealers on how in-house statistical models should be used to determine net capital requirements and whether it should replace or modify its current haircut methodology.

The SEC's net capital rule, adopted in 1975, requires every broker-dealer to have enough liquid assets to enable those firms that fall below the net minimum to liquidate in an orderly fashion. In calculating these requirements, broker-dealers have to deduct from their net worth certain percentages, known as hair-cuts, of the value of the firm's securities and commodities positions. These haircuts, derived from multiplying the market value of the positions by prescribed percentages, are to afford protection from market risk, credit risk and other risks inherent in particular positions.

The SEC puts forward a new way of calculating haircut percentages and recognising additional offsetting positions in its concept release, "Net Capital Rule" (1997). For broker-dealers with large proprietary positions, the SEC is considering allowing them to use an internal or external model to calculate the market risk charge and to take a separate charge, or charges, for other types of risk, such as credit and liquidity risks. If a firm uses an internal model, the Commission, like the Basle Committee, would prescribe certain minimum quantitative and qualitative criteria. The SEC envisages the incorporation of models into capital calculations in two ways:

  • The two-tiered approach requires broker-dealers meeting certain minimum threshold levels to use models to determine capital compliance. 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 will continue to follow the current Rule's haircut methodology.
  • The base approach with a pre-commitment feature has 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. Second, broker-dealers could take a fixed percentage haircut for each security in their portfolio. This haircut would be similar but lower than the current haircut requirements because the additional charge for market risk would be obtained from a third component. This would consist of capital based on the firm's model and include a pre-commitment feature that would require a broker-dealer to take capital charges based on the realised performance of its models.

The pre-commitment feature itself has two steps. First, at the start of a pre-determined time period, a broker-dealer could be required to represent its losses, as computed by its model, to 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 be increased by an amount equal to the difference between the actual portfolio gains and losses and the model's projections.

See also capital adequacy

  • The Supervisory Treatment of Market Risks (1993)
  • Proposal to Issue a Supplement to the Basle Capital Accord to cover Market Risks (1995)
  • An Internal Model-Based Approach to Market Risk Capital Requirements (1995)
  • The Implications for Securities Regulators of the Increased Use of Value at Risk Models by Securities Firms (1995)
  • Amendment to the Capital Accord to Incorporate Market Risks (1996)
  • Supervisory Framework for the Use of "Backtesting" in Conjunction with the Internal Models Approach to Market Risk Capital Requirements (1996)
  • Net Capital Rule (1997)

Latest Update January 2000


Key Risk Concepts * Sources of Risk