CAMBRIDGE, Mass. - Some think beta is an economist's dream and a portfolio manager's nightmare. In reality, beta - which measures the risk of a stock to the entire market - is like cholesterol, coming in both bad and good forms, according to a draft paper by two Harvard University economics professors.
This is no semantic distinction. These academics have come up with a way of explaining why small-cap and value stocks have outperformed large-cap and growth stocks over long periods of time. And it has important implications for both asset allocation and performance measurement for long-term investors, such as pension funds.
John Y. Campbell and Tuomo Vuolteenaho address a major flaw the Capital Asset Pricing Model has had since it was developed independently by William F. Sharpe and John Lintner in the mid-1960s.
Their attempt to explain the finance puzzle provides a revamped tool investors can use in looking at the relative riskiness of stocks by style and market capitalization. While they are not the first to tackle problems with CAPM, their research applies an intertemporal model, created in 1973 by Robert C. Merton, to CAPM.
In other words, it incorporates changing market conditions and investor demand.
Everyone knows that the valuation for, say, Lucent Technologies Inc. is far lower today than it was a year ago, and the stock is also now viewed as a much riskier bet than the market as a whole. This approach adds in the changing risk relative to the entire market.
The academics divide beta into two forms. "Bad" beta, which reflects news of expected cash flows into the stock market, is a more long-term phenomenon. It is influenced by long-run changes in corporate profitability.
"Good beta," by contrast, reflects expected changes in discount rates, which account for much shorter-term market movements. Using the intertemporal asset-pricing model, good beta should equal the variance of the market portfolio itself, Mr. Campbell explained in an interview.
Fails after 1963
The original CAPM worked fine worked fine with historical data in the period from 1929 through June 1963 to explain the outperformance of small-cap and value stocks, but it is widely recognized that the model falls down after that time.
The reason, according to the new research: Value stocks in the more recent period have disproportionately more "bad" beta, while growth stocks have disproportionately more "good" beta.
In fact, the draft paper suggests that value stocks are far riskier than is generally believed. Over the entire 73-year period measured, the authors found that "bad" cash flow beta is seven times riskier than "good" discount-rate beta.
"There's more risk to the Warren Buffett strategy" than is generally acknowledged, Mr. Campbell said.
What does this mean for pension funds? "The more conservative long-term investors are, the more they will dislike the bad beta..." he explained. That means they "should not necessarily have a strategic permanent tilt toward value stocks" and instead should focus on the broader market, Mr. Campbell said. However, shorter term, tactical shifts toward value stocks are viable.
Growth stocks are more volatile and more correlated with the market, but that shouldn't upset long-term investors such as pension funds, he added.
Not so radical
Focusing on the broader market is not a radical conclusion and is borne out in most pension fund asset mixes.
Where the new model may be more helpful is in analyzing the risk exposure of money managers. Now, institutional investors measure the performance of a value manager against a value-based benchmark. But many value indexes contain a sizable element of growth stocks, while growth indexes contain some exposure to value stocks.
A more sophisticated measure of beta would give pension executives a better idea of where their risk exposures are.
"What's helpful here is to crystallize what drives returns," said Michael A. Rosen, principal, Angeles Investment Advisors, Santa Monica, Calif. "And I think he's (Mr. Campbell) right to note it's two factors, discount rates and cash flows, and that's what matters."
Mr. Rosen said it makes sense that growth companies, whose earnings are further out in the future, are more sensitive to discount rates.
`A simpler explanation'
William Jacques, chief investment officer of Martingale Asset Management LP, Boston, said there's "a simpler explanation why value stocks outperform growth stocks. Maybe it's because of the high price-to-earnings of growth stocks, which reflect high expectations that do not materialize."
Research by Martingale, a value manager, shows that large-cap value outperformed large-cap growth by 260 basis points per year for the 75-year period ended last Dec. 31. What's more, the small-stock effect appears to have disappeared: large-cap growth stocks outperformed small-cap growth during the entire period.
Meanwhile, research by Jon Christopherson and Andrew L. Turner of the Frank Russell Co., Tacoma, Wash., and Wayne E. Ferson, the John L. Collins Chair of Finance at Boston College's Carroll School of Management, looks at how changing market conditions affect both alpha and beta. Their research on "conditional performance evaluation" is helpful in picking managers who can consistently add value, Mr. Christopherson said. Frank Russell is close to implementing a model that its manager research staff will use, he added.