Arnold Kling

The Risk Premium

PRINT
Consumption Taxation... Comment of the Week...

Surfing while snowbound, somehow I started with this post on Asymmetrical Information and wound up reading this article by Robert D. Arnott and Peter L. Bernstein on the equity risk premium. It is an exhaustive study that covers many fundamental issues in finance, so it is not easily summarized here.

Their main conclusion is that the high returns earned on stocks over the last 75 years on average are not indicative of high returns in the future. They expect growth in real dividends to be modest (two to four percent), and they say that a normal risk premium is only about 2-1/2 percent (other estimates have been as high as 5 percent). Finally, they suggest that the stock market as of the time that they article was written appeared to be overvalued, insofar as the apparent risk premium was close to zero.

I am one of the few economists who does not like the concept of the risk premium, which is measured in the same dimension as yield. I prefer Benjamin Graham's concept of "margin of safety," which gets at the same issue but is measured in the same dimension as price. I discuss the margin of safety and the value of the stock market in this essay, which admittedly is not nearly as thorough as the Arnott-Bernstein paper.

For Discussion. Economists and others who try to calculate market risk premiums (or the margin of safety) often use these calculations to draw conclusions about whether or not the stock market is overvalued. According to the Efficient Markets Hypothesis, what is the value of these market-timing conclusions?



COMMENTS (2 to date)
ruben cohen writes:

Yield and price are connected directly to each other. If you have one, you can compute the other. So it does not make any difference which one is used.

Comments appreciated on the above-mentioned paper, which concludes that the ex-ante risk premium is the expected dividend yield. The ex-post risk premium, on the other hand, involves the historical volatility of the market.

Bob Coleman writes:

This is an example of the confounding of PRICE (risk premiums) with VALUE (margin of safety). The definition of "risk" in each of these two radically different methods is crucial.

The reference to THE risk premium, of course, reveals a monocausal Capital Asset PRICING Model (the conventional CAPM). A long-reigning fad in academic finance is the addition of two fallacious atheoretical risk factors: Size and Value, which allegedly have their own risk premia.

Market timing or technical analysis, sometimes openly based on charting techniques such as Moving Averages, is scientifically not valid. It is essentially curve-fitting, which derives two metrics: slope and intercept. In contrast, statistical hypothesis testing adds a third metric: variance. With inferential statistics, all of the moments of each independent variable and of the error term or residual are included for analysis and testing.

Comments for this entry have been closed
Return to top