Much criticism has been leveled at the venerable CAPM. For example, Thomas Schneeweis, Editor of the Journal of Alternative Investments recently told AllAboutAlpha that:
“The CAPM was devised in 1964. But by the time I got my Ph.D. in the 1970s, we knew that model was not really measurable. It’s amazing how long it’s lived on.”
Behavioral economists have led the revolt against CAPM. Economist James Montier recently called CAPM “Completely Redundant Asset Pricing” or “C.R.A.P.” for short.
In its defense, chronicler of the history of modern finance, Peter Bernstein, calls CAPM-inventor William Sharpe “the most important single influence” in his landmark book Capital Ideas. In a recent interview, Bernstein acknowledges empirical shortcomings of CAPM, but points out its significance:
CAPM and the rest provided a place to begin. Just as clinical psychology would not be what it is today if there hadn’t been Freud…Nobody claims that the theories work. In particular, the Capital Asset Pricing Model has been proven, over and over again, even as far back as by Fischer Black, to not work. (But) we know things about how markets work and the centrality of the risk/reward trade off and diversification and so forth that just weren’t part of the investment process before these ideas were set forth.
And for his part, Sharpe essentially says “not so fast” in his recent book Investors and Markets and suggests that CAPM-naysayers such as Fama & French are ignoring certain friction costs associated with the so-called “value premium”.
But here’s a paper we came across recently (written last year) that launches a counter-attack on Fama & French. It attempts to re-frame the CAPM debate, suggesting that the value factor in the Fama & French model is ill-defined and needs to be split in two. The authors point to the fact that price-to-book is used as a proxy for “value”. But, they say, stock price might be temporarily lower due to some external shock. In some situations, a low price-to-book ratio signals temporary distress, not “value” fundamentals – a flat tire, but not a K-car. In their words:
“The flaw in the definition lies, to us, in the fact that firm values are directly related to current stock prices at an instant in time. Stocks are being compared in the rankings without considering that some may be in price equilibrium with respect to their firm fundamentals at the instant, while others may not be for a particular reason. If that is the case, the definition is confusing structural elements with more temporary price effects, as it is ignoring this time aspect of the BP ratios.”
In order to address the difference between actual (fundamental) value and apparent (static) value, this paper proposes that value be split into two variables: “equilibrium” value and divergence from equilibrium value (assumed to be zero by Fama & French). Lest we believe that value stocks are always priced at “equilibrium” (vs. “transitory”) levels, the authors point to the mere fact that analysts recommend value stocks because they believe the stocks are undervalued.
This paper attempts to find a true “equilibrium” price for stocks that is based on a moving average of stock prices, not the price at a point in time. Then they propose a sort of CAPM with a market factor and two value factors – one they call “structural” and the other “transitory”. The transitory value factor is assumed to be the result of short term distress.
The authors then calculate the cumulative excess (above market) returns of the “structural value” factors and of the “transitory value” factor and compare them to the good old-fashioned “value factor”. Their conclusion:
“Transitory value performs better than value (12.7 %) and structural value performs worse (8.7 %)…”
According to their study, the same thing holds true in different markets around the world: “transitory value” performs better than “structural value”. Their conclusion is inescapable:
“Both represent a distinct source of risk, and interestingly, one of them, transitory value, gives rise to systematic outperformance. The decomposed value definition is more significant statistically and is more intuitive to interpret.”
Alas, capturing and isolating the “transitory value” premium may be as difficult as capturing the value premium itself. Say the authors:
“The transitory value stocks are the stocks that are temporarily undervalued in respect to their equilibrium price, independently of their fundamental firm characteristics. The premium on these stocks is evidence that stock prices are mean reverting, i.e. on average temporarily undervalued stocks outperform temporarily overvalued stocks…however, we don’t hold the test results as evidence against efficient market theory! In order to seize the transitory value premium, one should systematically buy undervalued stocks and sell overvalued stocks each month, yet in the way the experiment is set up the transactions costs involved largely outweigh apparent opportunity gains.”