As some investors found out the hard way this year, the value of an asset may fluctuate wildly as market conditions change.
To prevent being blindsided, investors and derivatives dealers use stress testing to simulate the performance of their portfolios under many different market situations.
The tests, generally done with computer programs, value and revalue the assets in a portfolio or swaps in a swap book under various interest rate scenarios. Some tests even vary rates at random to measure portfolio performance under unexpected market conditions.
Why should a derivatives holder worry about extremely unlikely simulated market conditions? Because the simulations may reveal a chink in the holder's hedging strategy.
An industry report put out by the Group of 30, a Washington think tank backed by financial institutions, recommended last July that dealers regularly perform stress tests to uncover hidden risk.
"Simulations of improbable market environments are important in risk analysis because many assumptions that are valid for normal markets may no longer hold true in abnormal markets," the report said.
In addition to trying out portfolios in different rate environments, the tests may vary other market conditions as well.
For example, in a Volatile market, liquidity dries up and bid-ask spreads widen. If a simulation tested a portfolio's performance in a volatile market without also varying hid-ask spreads, it would less accurately predict actual performance.
David Askin, a fund manager who invested in mortgage-backed derivatives, designed his funds to pay a steady rate of return regardless of market rate changes. His computer models predicted that losses in some kinds of securities from rising rates, for example, would be off-set by gains in other securities.
But Askin had leveraged his funds, borrowing money at short-term rates and putting up his fund's assets as collateral.
When rates gyrated earlier this year, demand for mortgage-backed securities dropped and liquidity plummeted. The value of Askin's funds dropped more than predicted because the models had not sufficiently accounted for the liquidity drop.
As the value of Askin's collateral dropped, the lenders made margin calls. When Askin didn't have the cash to meet the calls, the lenders liquidated some of his securities at distressed prices.
In early 1993 the Group of 30 surveyed derivatives market participants and found that 15% blamed a loss in derivatives on "errors in the input assumptions in the pricing/risk management model used."
The group also found room for improvement in the stress-testing area. The survey asked about testing of "crash or shock scenarios on your derivatives portfolio."
Only 16% of respondents tested each of their portfolios and their combined portfolio with multiple scenarios, while 20% tested each portfolio alone with multiple scenarios and 25% tested with a single scenario.
The survey respondents said they planned to increase shock testing. Ultimately, 37% plan to use multiple tests of each portfolio and the total portfolio, and 25% plan to use multiple scenarios on each portfolio.
