The Current State of the Art: Beyond Beta
In Shakespeare’s Henry IV, Glendower boasts to Hotspur. “I can call spirits from the vasty deep.” “Why, so can I or so can any man,” says Hotspur, unimpressed; “But will they come when you call for them?” Anyone can theorize about how security markets work, and the capital-asset pricing model is just another theory. The really important question is: Does it work?
Certainly, many institutional investors have embraced the beta concept, if only in an attempt to play down the flamboyant excesses of the past. Beta is, after all, an academic creation.What could be more staid? Simply created as a number that describes a stock’s risk, it appears almost sterile in nature. True, it requires large investments in computer programs, but the closet chartists love it. Even if you don’t believe in beta, you have to speak its language because back on the nation’s campuses, my colleagues and I have been producing a long line of Ph.D.s and MBAs who spot its terminology.
By the early 19803. according to a Wall Street Journal article, bet had become so popular that it underlay the investment rationale for $65 billion in U.S. Pension funds. Beta also appeared to provide a method of evaluating a portfolio manager’s performance. If the realized return is larger than that predicted by the overall portfolio beta, the manager is said to have produced a positive alpha. Lots of money in the market sought out the manager who could deliver the largest alpha.
But is beta a useful measure of risk? Do high-beta portfolios always fall farther in bear markets than low-beta ones? Is it true that high-beta portfolios will provide larger long-term returns than lower-beta ones, as the capital-asset pricing model suggests? Do present methods of calculating beta on the basis of past history give any useful information about future betas? Does beta alone summarize a security’s total systematic risk, or do we need to consider other factors as well? In short, does beta really deserve an alpha? These are subjects of intense current debate among practitioners and academics, and not all the evidence is in as yet. We can review the available evidence and discusses the current state of thinking on the new investment technology.
Searching for the Investment Grail
Our guess is that the “turncoat quant” is wrong. The unearthing of serious cracks in the CAPM will not lead to an abandonment of mathematical tools in financial analysis and a return to traditional security analysis. The evidence that supports the efficiency of capital markets and the existence of what is usually a positive relationship between measured risk and return is far too abundant for anyone to reject the new investment technology out of hand. And since academics and practitioners have already made substantial progress in building better theories of the risk-return relationship, the practical consequence of the failures of the CAPM is likely to be more discriminating risk measurement, with the use of even more quantitative tools in risk analysis – not less. Will give the flavor of some of the new approaches to security pricing that have been developed as alternatives to the CAPM, and will present their practical meaning to investment analysts.
The Quant Quest for Better Measures of Risk: Arbitrage Pricing Theory
One of the pioneers in the field of risk measurement is the Yale School of Management’s finance wunderkind, Stephen Ross. Ross has developed a new theory of pricing in the capital markets called APT, or arbitrage pricing theory. APT has had wide influence both in academic community and in the practical world of portfolio management, To understand the logic of the newest APT work on risk measurement, one must remember the correct insight underlying the CAPM: The only risk that investors should be compensated for bearing is the risk that cannot be diversified away. Only systematic risk will command a risk premium in the market. But the systematic elements of risk in particular stocks and portfolios may be too complicated to be capturable by a measure of beta – the tendency of the stocks to move more or less than the market. This is especially so since any particular stock index is a very imperfect representative of the general market. Hence, many quants now feel that beta may fail to capture a number of important systematic elements of risk.
Let’s take a look at several of these other systematic risk elements. Changes in national income, for one, may affect returns form individual stocks in a systematic way. This was shown in our illustration of a simple island economy. Also, changes in national income mirror changes in the personal income of individuals, and the systematic relationship between security returns and salary income can be expected to have a significant effect on individual behavior. For example, the laborer in a Ford plant will find a holding of Ford common stock particularly risky, since job lay-offs and poor returns from Ford stocs are likely to occur at the same time. Changes in national income may also reflect changes in other forms of property income and may therefore be relevant for institutional portfolio managers as well.
Changes in interest rates also systematically affect the returns from individual stocks and are important non diversifiable risk elements. To the extent that stocks tend to suffer as interest rates go up, equities are a risky investment, and those stocks that are particularly vulnerable to increases in the general level of interest rates are especially risky. Thus, many stocks and fixed-income investments will tend to move in parallel, and these stocks will not be helpful in reducing the risk of a bond portfolio. Since fixed-income securities are a major part of the portfolios of many institutional investors, this systematic risk factor is particularly important for some of the largest investors in the market. Clearly, then, investors who think of risk in its broadest and most meaningful sense will be sensitive to the tendency of certain stocks to be particularly affected by changes in interest rates.
Changes in the rate of inflation will similarly tend to have a systematic influence on the returns from common stocks. This is so for at least two reasons. First, an increase in the rate of inflation tends to increase interest rates and thus tends to lower the prices of equities, as just discussed. Second, the increase in inflation may squeeze profit margins for certain groups of companies – public utilities, for example, which often find that rate increases lag behind increases in costs. On the other hand, inflation may benefit the prices of common stocks in the natural-resource industries. Thus, again there are important systematic relationships between stock returns and economic variables that may not be captured adequately by a simple beta measure of risk.
Statistical tests of the influence on security returns of several systematic risk variables have shown promising results. Better explanations than those given by the CAPM can be obtained for the variation in returns among different securities by using, in addition to the traditional beta measure of risk, a number of systematic risk variables, such as sensitivity to changes in national income, in interest rates, and in the rate of inflation. Of course, the evidence supporting many-risk-factor models of security pricing has only begun to accumulate. It is not yet certain how these new theories will stand up to more extensive examination. Still, the preliminary results are definitely encouraging.
If, however, one wanted for simplicity to select the one risk measure most closely related to expected returns, the traditional beta measure would not be my first choice. The best single risk proxy turns out to be the extent of disagreement among security analysts’ forecasts for each individual company. Companies for which there is a broad consensus with respect to the growth of future earnings in dividends seem to be considered less risky (and hence have lower expected returns) than companies for which there is little agreement among security analysts. It is possible to interpret this result as contradicting modern asset pricing theory, which suggests that individual security variability per se will not be relevant for valuation. The dispersion of analysts’ forecasts, however, may actually serve as a particularly useful proxy for a variety os systematic risks.
Consider, for example, two companies. One a machinery manufacturer, is heavily in debt and extremely sensitive to systematic influences. The other, an all-equity pharmaceutical firm, is quite insensitive to economic conditions. It could be that Wall Street analysts agree completely on how economic conditions will affect the companies, but differ greatly on their economic forecasts. If so, there could be a big dispersion in earnings forecasts for the machinery manufacturer (because of the difference in economic forecasts and the extreme sensitivity of the company to economic conditions) and very small differences in the forecasts for the drug company (because economic conditions have little effect on that company). Thus, if two different analysts have very different forecasts for GNP, inflation, and interest rates, a highly debt-leveraged company in a heavy industry would be greatly affected by differences in underlying economic forecasts, while an unleveraged drug company might show no effect whatsoever. Hence, differences in analysts’ forecasts could be a most useful proxy for systematic risk in the broadest sense of the term.
While we still have much to learn about the market’s evaluation of risk, I believe it is fair to conclude that risk is unlikely to be captured adequately by a single beta statistic, the risk measure of the CAPM. It appears that several other systematic risk measures affect the valuation of securities. In addition there is some evidence that security returns are related to size (smaller firms tend to have higher rates of return) and also to price-earnings multiples (firms with low P/Es tend to produce higher returns). Whether individual risk plays any role at all in the valuation process is still, however, an open question.
Individual security variability does play a role in the valuation process. This would not be hard to explain. Because of transaction and information costs, a large number of individual portfolios may not be diversified. Individuals own between one-half and two-thirds of all NYSE stocks and an even larger fraction of stocks traded on other exchanges. Thus, these security holders might well be concerned with the variability of individual stocks. Even well-diversified institutional investors may worry about the behavior of individual stocks when they must report to finance committees the breakdown of their performance results over the preceding period. Still, there is a powerful argument on the other side. Any role in the valuation process that may consistently be provided by individual security variability will create an arbitrage opportunity for investors able to diversify widely. It is difficult to believe that these arbitrage opportunities will not eventually be exploited and “true value will out”.