Do more popular stocks have lower returns?

The authors could have operated conventionally. The Popularity hypothesis could have been structured to count down, by assigning extra expected returns to stocks that carry undesirable attributes. That approach, ultimately, would arrive at the same conclusions. However, there is merit to their decision. The choice to reward benefits rather than penalize disadvantages emphasizes that their model, unlike their predecessors', does not insist that risks are the sole determinant of stock returns. The term "risk" is unnecessarily limiting.

Small Thinking

For example, large companies may be more popular than smaller firms, and thus possess lower expected returns because they can more readily be evaluated. An institutional investor seeking to put $800 million to work could place that entire amount into 0.1% of Apple’s (AAPL) market capitalization, or it could split that $800 million into 200 positions of 0.1% in companies the size of Morningstar (MORN). One research effort is not like the other.

The authors posit that because detailed third-party information about giant companies like Apple is widely available, while that about firms of Morningstar's size is not, small companies have enjoyed, and will likely continue to enjoy, higher gains. (This effect is separate from any extra returns that they may reap from having higher volatility and lower liquidity, as those are separate factors.)

You get the idea. Active investors consider many things when deciding which equities to hold. Some are rational responses to additional risks; others are rational responses to attributes that are important but cannot be described as risks; and still others are irrational responses. Thus, the authors' subtitle: The Popularity hypothesis seeks to incorporate both classical stock market models that assume fully rational investors, and behavioral models that do not.

Running the Numbers

Enough generalities; time for some specifics. In their Popularity monograph (published, by the CFA Institute), the authors do not attempt to capture all aspects of popularity. However, they do apply their hypothesis to five stock market features that are rarely measured because they are bypassed by traditional risk/return analysis.

I list their findings below, from the largest of the five effects to the smallest. The summary number represents the difference in annual returns (as measured by the geometric mean) between an equal-weighted portfolio composed of stocks that scored in the highest quartile for exposure to that factor and the equal-weighted portfolio of stocks that scored in the lowest quartile.

(Technically, these represent percentage points rather than percentages, but permit me the abbreviation, for the sake of cleanliness.)

Overpaying for Quality?

1) Company Brand--6.08%

This finding requires an asterisk, because it is based on only 75 to 100 companies per year. On the other hand, it's cool. The branding consultancy Interbrand provided the authors with a spreadsheet of its rankings of global brand values over the past 18 years. Within that coterie, small in number but large in market capitalization, the stocks that had the most powerful brands sharply trailed those that were judged to have the weakest brands.

It is not, of course, that brand power is unimportant. Quite the contrary. Unfortunately, as predicted by the Popularity hypothesis, stock investors had already rewarded those companies. Enticed by the allure of owning shares in a business with a great global brand, they had pushed up those stocks' prices, thereby depressing future returns.

2) Company Reputation--5.59%

This result resembles the first. Once again, it is based on a relatively small group of multinational companies, as judged by a third party, with the data covering the years 2000-17. In this case, the third party is The Harris Poll, and the factor assessed was company reputation. Unlike with Interbrand, which calculated its estimates, The Harris Poll surveyed the general public to arrive at its conclusions.

No matter. The outcome was similar. Overall, stocks of the businesses that the public held in highest esteem lagged those of firms that were less admired. It's wonderful to be loved--but not, it seems, for the prospective rewards of equities, as opposed to their past performance.

3) Competitive Advantage--4.25%

A third bite at the apple. This time, the measure for company quality was the Morningstar Economic Moat Rating. The time period was similarly modest, at 16 years, but the field was expanded to an average of 1,039 companies each year. The economic moat is assigned by Morningstar's equity analysts and reflects their view of the size of a company's sustainable competitive advantage.

Wide moats are indeed good for business, but not, as predicted by the Popularity hypothesis, for future stock performances. Narrow-moat issues outgained the wide-moat securities. Fortunately for Morningstar, its overall stock ratings include the effect of price as well as the depth of the moat. As the Popularity hypothesis suggests, one can indeed pay too much for a good thing.

Skewed Perspectives

4) Tail Risk--3.95%

And now for something completely different. All things being equal, investors dislike stocks with "negative coskewness," meaning that they land particularly hard when the market falls. Fancy that. Presumably, then, stocks that have the least amount of negative coskewness are relatively popular. Also presumably, they record lower future returns, because that popularity has already elevated their stocks' prices.

That is indeed what the Popularity authors discovered, this time with the entire universe of U.S. stocks, dating back to 1991. This result is solid.

5) Lottery Effect--(1.72%)

This one didn't entirely work. The authors posited that a subset of investors favors stocks that mimic lottery tickets, in that they carry a small probability of an extremely high payoff. The popularity of that attribute, they believed, would erode the returns of securities that boast those attributes, while improving the returns of their antipodes.

That is not what they found, on an equal-weighted basis. Indeed, the opposite held true. Stocks with the strongest lottery features outgained those with the weakest. However, as the market-cap-weighted computation slightly reversed that finding, and the risk-adjusted results for both calculations were better for the "unpopular" issues (no surprise, lottery stocks tend to be risky), this topic remains far from settled.

Final Thoughts

The Popularity authors' numbers should be regarded as suggestive, not conclusive. Some derive from limited databases. All beg questions. For example: How correlated are the three quality measures? Or, would the findings change if the authors used multifactor analysis rather than quartile buckets?

Nonetheless, the empirical research is encouraging. It offers more than enough support for the Popularity hypothesis to warrant further investigations.

John Rekenthaler has been researching the fund industry since 1988. He is now a columnist for Morningstar.com and a member of Morningstar's investment research department. John is quick to point out that while Morningstar typically agrees with the views of the Rekenthaler Report, his views are his own.