BERKELEY, Calif. - Dissecting the risk of individual securities in a mutual fund is a better predictor of risk than traditional measures, which look at entire portfolios, according to a new study by BARRA Inc.
BARRA looked at 29 domestic equity and bond mutual funds and tested the accuracy of two different risk forecasts. For both, BARRA analyzed the risk of mutual funds for the three-year period ended Dec. 31, 1993 and used the data to project risk in the subsequent year.
Both methods looked at monthly data. One looked at the volatility of a entire portfolio's total return, while the other aggregated the volatility of the individual securities within the portfolio to arrive at an aggregate risk figure.
The test showed negative returns would seldom have surprised investors if they had known the fund risks based on the holdings. But in many cases, negative returns did surprise investors when the portfolio-based risk measure was used.
Using holdings-based risk forecasting, just 3% of the realized returns were surprising to investors. With portfolio risk forecasting, more than 17% of the realized returns were surprising. To quantify the notion of surprise, BARRA defined the standardized outcome as the ratio of the realized excess return to the forecast risk. Both methods measured standard deviation.
According to Paul Green, strategic marketing manager, analyzing the standard deviation of the total return of a whole portfolio would not capture the effects of a manager change or shift in portfolio strategy.
Such a method "depends on a long series of observed monthly variations. It would forecast moderate risk based on moderate changes even though the portfolio may have shifted toward more volatile stocks. The preferred approach looks at (the standard deviation) of individual holdings, which can shift month to month and aggregates the risk of all those holdings," he said.
"It's a superior method. Now we have a study to back that up. Our clients . . . measure risk that way, and so do our competitors," including Wilshire Associates and Vestek Systems, he said.
While many pension funds require their money managers to furnish them with risk analysis of individual portfolio holdings, the practice is not common in the mutual fund industry.
"Instead, mutual funds tend to look at the month-to-month returns of the whole portfolio. That makes the assumption that a whole portfolio's past volatility will be the same in the future. If you had used the preferred method, you would have been surprised much less frequently about subsequent volatility," he said.
He said the difference in results between the two methods is even more crucial to bond fund investors, "because there you really are surprised when things happen differently from the past. The unpleasantness of the shock is even more unpleasant."
Bond holdings of the funds were provided by CDA Spectrum BondWatch; stock holdings came from Morningstar Inc., Chicago.
BARRA began its study with a list from Stephen J. Butler of Pension Dynamics Corp., Lafayette, Calif., of funds that are the most popular constituents of 401(k) plans, and added other large funds as well as funds whose performance in 1994 was much worse than their performance in prior years.
For the study, BARRA defined risk as the standard deviation of excess returns above the risk-free Treasury bill rate of return.
Ron Kahn, BARRA's director of research, wrote the study.