Why Popular Stocks Are a Sucker's Bet
A recent study points out, yet again, how fickle we humans are.
Any student of the market is by necessity a student of human psychology and the behavior of crowds. With that in mind, consider a study published in the Feb. 10 issue of Science magazine. The authors, Matthew J. Salganik and Peter Sheridan Dodds of Columbia University, and Duncan J. Watts of the Santa Fe Institute, studied how people judged music when left to their own devices, versus how they judged music when they knew how popular the music was among their peers. It offers a nice demonstration of people's suggestibility and dovetails with our own thoughts about market behavior and the behavior of crowds.
Can you guess the results?
First, the Study
The paper's authors created an artificial music market on the Internet, recruiting more than 14,000 participants. They invited each participant to listen to a set of songs by various obscure bands, and offered them the option of downloading any song they liked.
The authors divided these 14,000 individuals randomly into nine groups. One group--called the "independent" group--simply saw a list of 48 songs in random order. They then sampled this music and downloaded the songs they liked. As you would expect, this resulted in a wide dispersion in the number of downloads. A few songs got downloaded a lot, and a few songs were hardly downloaded at all.
The other eight groups--the "social influence" groups--had the same list of songs to choose from, but also saw the number of times each song had been downloaded by other members of their respective groups. Within each of these eight groups, two factors therefore contributed to whether a song was downloaded: the quality of the song, and the influence of the popularity of the song on each participant.
Why eight different social-influence groups? This way, the authors could essentially create eight different worlds in order to see whether each world evolved along similar lines, or if their paths diverged. In many real-life situations--like the stock market--we only have one history to study, and we don't know if that one history was the way things had to be, or whether it was the result of blind, stupid chance.
The study's key results are threefold, and have some rather obvious parallels with investor behavior.
The addition of peer influence widens the gap between the winners and losers.
In other words, peer influence makes the most popular songs even more popular, and the least popular songs less popular. The authors used the Gini coefficient to measure this. Economists use the Gini coefficient to measure things like the income inequality across different countries (the United States versus France, for example) or within the same country over time (the U.S. of 1900 versus the U.S. of 2000). Based on this measure, the gulf between the most-popular and least-popular songs widened in each of the eight "peer influence" groups versus the independent group.
Does this remind you of the stock market? A given industry or asset class--Japan in 1989, emerging markets in 1994, Internet stocks in 1999, or (dare we say) commodity stocks right now--soars far beyond what the fundamentals justify and then comes crashing back to earth. As Benjamin Graham told us many years ago, in the short term markets are voting machines. And as this study and others like it suggest, the very fact that something is popular makes it even more so, for a while at least.
When you introduce peer influence, you increase the unpredictability of the results.
Not only does the difference between winners and losers widen, but it gets tougher to predict who the winners and losers will be. The authors calculated the differences in "market share" for each song in the eight different social-influence worlds and compared this to differences in market share in various random samplings of the "independent" group. The unpredictability between different social-influence worlds was dramatically larger than in the independent worlds.
This is further confirmation, if any were needed, that our knowledge of the future is severely limited. What stocks will go up this year? Where will interest rates head? Where will the Dow finish? Easy questions to ask, but impossible ones to answer. (This doesn't stop the answers from coming, unfortunately.) As the authors put it: "We conjecture...that experts fail to predict success not because they are incompetent judges or misinformed about the preferences of others, but because when individual decisions are subject to social influence, markets do not simply aggregate pre-existing individual preferences. In such a world, there are inherent limits on the predictability of outcomes, irrespective of how much skill or information one has."
The greater the peer influence, the stronger are the first two results.
The researchers repeated the experiment twice. The second time, the social-influence groups saw the songs ranked in descending order of downloads, whereas in the previous experiment they had appeared in random order. This made it easier for participants to see which songs were most popular. In this case, the first two results--a more extreme dispersion between the most- and least-popular songs, and an increase in unpredictability--were even more pronounced. The authors' conclusion: "Increasing the strength of social influence increased both inequality and unpredictability of success."
Profiting from the Crowd's Madness
In any given year, it's impossible to predict which stocks (or bonds, or mutual funds, etc.) will be popular or unpopular, but the great thing about investing for the long term is that it doesn't matter. We don't have to predict what will be popular next year. We just have to wait for great companies to become unpopular. In fact, the more bipolar the market is, the more profitable a disciplined long-term investment strategy will be. Ultimately, stock prices depend on the cash a company generates. And if you keep your eyes focused on that, and not the best-performers' list, you'll do just fine.
Profiting from the ebbs and flows of popularity is what the Morningstar Rating for stocks is designed to do. The table below breaks out some characteristics of Morningstar's current list of 5-star and 1-star stocks. As you can see, we like what happens to be out of favor. Looking at the stocks currently rated 1 star, the median return over the past 12 months is 61%, as opposed to negative 5% for the median 5-star stock. (These results are not the performance of the 1-star and 5-star ratings, we hasten to clarify. They're simply the median returns of the stocks that happen to be 1 star and 5 star today.)
|Current 5-Star Stocks Versus 1-Star Stocks|
|5 Star||1 Star|
|Trailing 3-month return||-5.00%||10.20%|
|Trailing 1-year return||-5.20%||60.74%|
|Return on equity||15.70%||11.50%|
|Percent carrying above-average or speculative risk||9%||60%|
|Data as of 04-28-06. Figures are medians.|
The numbers also show that lower-quality businesses have been popular. Many marginal businesses have seen their stocks soar far above what we feel are reasonable prices--thus their 1-star ratings. By contrast, the market is currently offering excellent prices for quality companies with high returns on capital and powerful business models. These include eBay (EBAY) (a recent addition to our Hare Portfolio), Wal-Mart (WMT), Dell , Berkshire Hathaway (BRK.B), Maxim Integrated Products (a recent purchase in our Growth Portfolio), and dozens more.
By allocating capital to such firms, you can turn the fickleness of the crowd to your advantage.
* The article is "Experimental Study of Inequality and Unpredictability in an Artificial Cultural Market," by Matthew J. Salganik, Peter Sheridan Dodds, and Duncan J. Watts, Science 10 February 2006 311: 854-856.
Haywood Kelly, CFA has a position in the following securities mentioned above: BRK.B, EBAY, DELL. Find out about Morningstar’s editorial policies.