Tuesday, August 10, 2004

SSRN-How the Inflation Illusion Killed the CAPM by Randolph Cohen, Christopher Polk, Tuomo Vuolteenaho

SSRN-How the Inflation Illusion Killed the CAPM by Randolph Cohen, Christopher Polk, Tuomo Vuolteenaho

Wow! It seems like I say that frequently, but what this paper definitely deserves a wow.

CAPM is dead. We know that. Only the most optimistic of those (myself included) hold out hope of its recovery.

Its death has been widely reported since Fama and French (1992). However, that does more to say that CAPM is dead, not why or how it died.

The CAPM had been on its sickbed for years. Looking back we see the evidence. In addition to a slew of anomalies (size, market to book, calendar to name but a few), there was more serious trouble in the findings that suggested CAPM worked sometimes and didn’t work at other times. For instance CAPM did a bad job of explaining returns in the in the 1950s and in the post 1980 period. Additionally, previous literature (for instance Black, Jensen, and Scholes (1972) had found the empirical SML to be flatter than expected and to have an intercept (zero beta portfolio return) that is greater than the Risk Free rate.

Maybe, just maybe we now know some of the details of CAPM’s death thanks to the work of the great detectives Cohen, Polk, and Vuolteenaho (who are in reality professors but will play the role of detectives in this report). They released their findings in the paper “How the Inflation Illusion Killed the CAPM.” (http://papers.ssrn.com/abstract=548402)

The short version of the work is as follows:

To set the stage a bit: remember that the Expected market risk premium is the expected return on the market minus the expected inflation rate (which is largely the risk-free rate). So (ERm – RF) equals the expected market risk premium.

They begin off by further investigating Modigliani and Cohn’s 1979 suggestion that the market incorrectly values assets because of irrationally accounting for expected inflation rates. While this irrationality was difficult to accept when the M&C paper was published in 1979, there has been enough time-series evidence since that time to make the hypothesis more palatable (Sharpe 1999, Asness (2000), Campbell and Vuolteenaho (2004) etc.).

In simple terms, Modigliani and Cohn (1979) stated that investors irrationally accounted for inflation by overemphasizing recent inflation rates. Thus, when inflation is high, investors expect inflation to remain high. This is ok as far as it goes, but the error comes when they (the investors) fail to realize that earnings will also grow with inflation. Consequentially, when inflation is high, the market risk premium is too low (since the expected inflation rate is subtracted in the risk premium calculation).

To test this, the authors (err detectives) “propose a behavioral hypothesis that the market uses the Sharpe-Lintner Capital Asset Pricing Model (CAPM, Sharpe 1964, Lintner 1965) but suffers from inflation illusion.” They then proceed to “ show that an implication of this joint hypothesis is that the security market line (the relation between an asset’s average return and its CAPM beta) is steeper [i.e there is a larger risk premium] than predicted by the Sharpe-Lintner CAPM when inflation is low or negative. Conversely, when inflation is high, the security market line is shallower [i.e. there is a smaller risk premium] than the Sharpe-Lintner CAPM’s prediction.”

The results? Sure enough, the empirical evidence supports the hypothesis of Modigliani and Cohn. And the author/detectives conclude that “we find that the CAPM fails predictably with inflation, consistent with the Modigliani-Cohn inflation illusion hypothesis.”

So what does this all mean? For starters, we may have an important piece of the puzzle as to how assets are priced. Moreover, there is more evidence that investors are not always perfectly rational and that they may over emphasize recent conditions.

A question that merits future attention is whether this finding could help to explain the growth and value anomalies. For example, if value stocks do better when the economy is doing well (and hence inflation may be higher), is the apparent outperformance a market problem or a model problem? I do not know but definitely look forward to finding out!

BTW this is definitely going to be required reading for my classes, I hope you read it as well and do not just rely on this review.

As an aside, Richard Cohen (yes one of the authors) sugggested the following which makes the paper more readily understandable:
"One point might be worthy of clarification if it can be done without making things too complicated. It's not that investors are wrong to assume inflation will be similar to recent inflation. Their mistake is forgetting that earnings will go up with inflation. Thus if T-Bills pay 12% because of 10% expected inflation, and earnings yields are at 11%, investors look and say, hey, the (safe) T-Bill return beats the yield available in the market [I'm oversimplifying here]; let's sell our stocks and buy T-Bills." They should be saying, "not only do stocks yield 10%, but the earnings will grow at 11% next year because of high inflation, making stocks very attractive indeed." "

Cohen, Randolph B., Polk, Christopher Keith and Vuolteenaho, Tuomo , "How the Inflation Illusion Killed the CAPM" (May 12, 2004). http://ssrn.com/abstract=548402

A newer version of this, entitled "How Inflation Illusion Killed CAPM" is available at

No comments: