Are Value Stocks Riskier than Growth Stocks?

One of the cornerstones of modern finance is the nexus between return and risk. These two characteristics are joined at the hip—you simply don’t get one without the other. It’s also well accepted that value stocks have higher returns than growth stocks. Not only is this empirically so (decades of work on the stock markets of dozens of nations, including the U.S. back to 1927, demonstrate that doggy companies have higher returns than glamorous ones), but value stocks must have higher returns because they are riskier companies.

And so, without mangling syllogistic logic, must it not follow that because they have higher returns, a portfolio of value stocks must indeed have higher risk?

The problem is that this risk is not readily apparent. Let’s start with the most widely used measure of risk, standard deviation (SD). Utilizing the Fama-French (FF) and S&P indexes, here are the annualized returns and SDs for monthly data from July 1963 through April 2002:

	Return	SD
S&P 500	11.04%	14.88%
CRSP Universe	11.00%	15.42%
FF Large Growth	10.25%	16.65%
FF Large Value	13.71%	15.39%
FF Small Growth	9.68%	24.60%
FF Small Value	17.59%	19.20%

If anything, value seems to have lower risk than growth, especially for small stocks, where small growth stocks have by far the highest risks and lowest returns of any cross section. In the words of Ben Graham, "Why do folks buy this junk?"

Aha, say the academics: There are dimensions of risk not measured by simple standard deviation. They correctly point out that value stocks are "poor earners"; they are "distressed," with low profitability and tenuous financial strength. "Just look at these companies—some of them will fall over in a strong breeze." True. But the key question is how systematic is this risk? Take the example of Kmart: Suppose there were a 75% chance that it would be bankrupt within one year. In order to repay its investors, it would need a greater than 300% payoff if it survives. Let us assume, for the sake of argument, that its payoff were 400%. Its expected one-year return would then be 25%. [0.75 x -100% + 0.25 x +400%.] Thus, in a portfolio of 100 such stocks, in order to lose money, 81 or more companies would have to fail; binomial probability tells us that the odds of this happening are only 10%. (Purists will argue that bankruptcy is not a necessary dimension of risk. Agreed, but bankruptcy is a handy paradigm—switch to negative earnings surprises or persisting poor growth and the math changes, but the basic concept does not.)

The above paradigm also grossly overstates value risk; most value companies, although distressed, are not bankruptcy risks, and most in fact have earnings. However, the above example made one strong assumption—that the risk of each company is independent, that the odds of one company failing tells us nothing about the odds of another company failing. Thus, in our example, the risk is almost completely diversifiable and, therefore, not a real risk to the holder of a large number of securities.

In actuality, of course this is not so: Adverse economic conditions can affect all companies, particularly value companies. It follows, then, that the risk of value stocks is "business-cycle risk"—the possibility that value companies as a group will be disproportionately affected by an economic downturn. Thus, one would predict that during economic downturns, growth should beat value.

The record in this regard is mixed. During the Great Depression, it was indeed the case: from September 1929 until June 1932, Ken French’s data show that large growth stocks lost "only" 82% of total return versus a loss of 89% for large value. Similarly, for the 12 months from October 1989 to September 1990, large growth and value lost 7% and 19%, respectively.

On the other hand, from 1973 until 1974, the reverse occurred, with large growth stocks losing 45%, versus only 26% for value. Similarly, from April 2000 to July 2002, large growth lost 44% versus only 27% for large value.

Not only are the real-world data ambiguous about the nature of value risk, but recent events suggest that growth stocks possess a risk all their own—bubble collapse. Bubbliness, of course, is in the eye of the beholder. Devout efficient marketeers sneer at the very concept: bubbles don’t exist, they are evident in retrospect, and failing all else, if they do exist, they are "rational," whatever that means.

But no matter what your financial religion, if the Internet/tech scene of the late 1990s wasn’t a bubble, then nothing ever was. And almost by definition, growth stocks are the heart and soul of a bubble—it is difficult indeed to spin a convincing story around a distressed company in an out-of-favor industry. Bubbles, by their very nature, revolve around the supposedly unlimited growth possibilities of the transformative technologies of the age—the Internet in the late 1990s, mainframe computers and airlines in the 1960s, radio and electrical utilities in the 1920s, and British railroads in the 1840s. Although the technologies prospered, investors lost their shirts by hideously overpaying for their growth.

Finally, there are behavioral issues involved. Even efficient marketeers will admit that because of the lack of persistence of earnings growth, growth stocks are priced higher than the present value of their future earnings and dividends. Further, it is well established that negative earnings surprises hit growth stocks harder than value stocks and, in the same vein, positive surprises benefit value stocks more than growth stocks.

I submit, then, that although value and growth stocks have their own unique risks, those of growth stocks are more regular and pervasive. During a depression, growth companies may hold up better than value companies, though history has shown this to be an unreliable phenomenon. But when bubbles burst, you can take to the bank that growth will get whacked more than value. And, as long as there are human beings, there will be bubbles.

 

Copyright © 2002, William J. Bernstein. All rights reserved.

The right to download, store and/or output any material on this Web site is granted for viewing use only. Material may not be reproduced in any form without the express written permission of William J. Bernstein. Reproduction or editing by any means, mechanical or electronic, in whole or in part, without the express written permission of William J. Bernstein is strictly prohibited. Please read the disclaimer.