Читать книгу Competitive Advantage in Investing - Steven Abrahams - Страница 28

To start, CAPM predicted that the return on any asset begins with the riskless return and then adds a multiple of the excess return on the market portfolio of risky assets:

Michael Jensen (1967), then at the Harvard Business School, noted that if investors stripped out the riskless rate from the periodic return on each asset, then CAPM predicted the only thing left would be the multiple of the excess return on the market portfolio:

In other words, the return on any investment above and beyond the riskless rate was simply the excess return of the market portfolio scaled up or down by a multiplier, . That pointed to a CAPM prediction that investors could test directly:

In this test, if CAPM described the world well, the value of α_i should be zero. That was one place where analysts could test CAPM theory against actual investment results. A number of analysts did exactly that (Douglas, 1968; Black, Jensen, and Scholes, 1972; Miller and Scholes, 1972; Blume and Friend, 1973; Fama and MacBeth, 1973; Fama and French, 1992).

When investors followed Jensen's lead and tested CAPM they found that not only did α_i have a value other than zero but also it followed a surprising but reliable pattern: had a positive value when is low and a negative value when is high. In other words, more stable assets had a better return than predicted by CAPM and more volatile assets had a worse return. The capital markets line in the real world was flatter than predicted by theory.

As recently as 2004, Eugene Fama at the University of Chicago and Kenneth French at Dartmouth again ran the numbers on portfolios of stocks built to be either high or low beta (Fama and French, 2004). Assembling these portfolios at the beginning of each year, they looked at performance over the year that followed. CAPM would predict that the safest portfolio of stocks would return only slightly more than the riskless rate for the year. As portfolios became riskier, their return simply would be the excess return of the market portfolio that year multiplied by each portfolio beta. Testing from 1928 through 2003, they compared actual returns on these portfolios to returns predicted by CAPM. They found that low beta stocks had returns higher than predicted, and high beta stocks had returns lower than predicted (figure 3.1). Again, the line was too flat.

Higher beta investments still showed higher returns, but the flatness of the observed capital markets line suggested that CAPM alone might not explain it. The theory had a mysterious flaw.

Figure 3.1 On stock portfolios that vary in their beta, realized returns are too flat compared to the predictions of CAPM.

Note: Average annualized monthly return versus beta for value-weight portfolios formed on prior beta, 1928–2003.

Source: From Fama and French (2004).