The small-cap premium is entrenched in academic and professional finance.

Smaller stocks do not bring about bigger returns.

Yes, you read that correctly.

If there is a financial equivalent to what goes under the jargon “political correctness,” the good folks at Research Affiliates, specifically Vitali Kalesnik and Noah Beck, have just committed it in a new report called “Busting the Myth About Size.”

As the current fashion in today’s infantilized academic environment is to include so-called “trigger warnings” before saying something to which the audience might take offense, readers reaching for the smelling salts can revive themselves with the authors’ front-of-the-article assurance that small stocks might still be desirable investments.

But just as a Ptolemaic conception of the universe gave way to a Copernican one, which in turn yielded to a relativistic, Einsteinian one, the Research Affiilates duo have come to announce that the evidence supporting a return advantage for small stocks is largely spurious, despite its cherished status as an axiom of investment.

Kalesnik, head of equity research, and Beck, a researcher at the firm, based in Newport Beach, California, early on state that small stocks may add diversification value, and further that they are more subject to mispricing, and thus serve as “an alpha pool” for active managers and rules-based systems.

The purpose of this de facto trigger warning doubtless stems from the knowledge that the small-cap premium is entrenched in academic and professional finance. Indeed, it is one of the big three sources of return advantage in the original 1993 statement of the Fama-French three-factor model.

But taking a fresh look at the data, Kalesnik and Beck expose some anomalies therein that appear to shatter the “myth” of the small-cap premium.

One such anomaly derives from reduced impact of the small-cap effect in international data sets versus its seemingly robust impact in the U.S. But even in the U.S., there are problems.

So in a data set extending from July 1926 to July 2014, the small-cap premium is a whopping 3.4%, which is statistically significant.

But an international data set consisting of 18 countries including the United States from January 1982 through July 2014 shows a mere 1% return advantage on average. In some of those countries, small stocks underperformed. Though the average effect of small-caps was positive, and in the U.S. during that period added 1.9% to equity returns, in no place was the advantage of small-cap stocks statistically significant.

That finding impelled Kalesnik and Beck to look still closer at the long-term U.S. data set, which is the source of the idea that small-cap stocks confer a return premium, and there they discovered a number of serious problems.

The first and most important problem is that the data is not clean. By not accounting for small-cap stocks that were eventually delisted, the database essentially overestimates small-cap returns by omitting from the record companies whose delisting likely emanated from poor corporate performance.

The notion that delisted stocks tend to have strongly negative returns is not just intuitive. The Research Affiliates duo review data on small stocks listed on the Nasdaq and find that “after adjusting for the delisting bias, the statistical significance of the size premium completely disappears.”

A second problem involving transaction costs is straightforward. “Small stocks by definition have much lower trading capacity and, correspondingly, much higher transaction costs,” which are not factored into the tested portfolios, Kalesnik and Beck write.

A third problem involves data mining that turns up patterns that attract the attention of publications but whose unobserved methodological flaws may invalidate that data. Kalesnik and Beck propose that the standard threshold used to test the small-cap premium’s statistical significance is too low given the nature of factor analysis, which searches for statistical irregularities. A fourth problem, and perhaps the key flaw the Research Affiliates duo identifies, involves four months during which small-caps produced extreme outlier premia over their larger-cap brethren, casting severe doubt on any assumptions of normality in the data set.

In one of those months (January 1934), the 23.6% premium that occurred would count as a “one-in-67 million-year event, like the one that wiped out the dinosaurs,” the authors write using statistical “sigma” analysis.

But the sigma scale only escalates with the other three outliers. The 27.2% premium return in September 1939 would on a normal distribution “have about a one-in-five chance of occurring in the 4.5 billion years since the planet earth came into existence.”

The two higher premia of 33.8% (August 1932) and 51.6% (May 1933) are improbable in the extreme: “each would have much less than a one-in-a-hundred chance of occurring in the entire 13.8 billion years the universe has existed,” the authors write.

What’s more, each of the four outliers occurred in the 1930s, which tends to make the effect seem more anomalous. And removing them from the data set shrinks the small-cap return premium from 3.4% to 1.9%, which no longer counts as statistically significant.

As Kalesnik and Beck put it, “if the size premium is predicated on exceedingly rare events, then we’ll have to wait many lifetimes to determine with confidence whether or not it exists.”

A fifth, and fairly straightforward problem with the small-cap premium idea, involves the markedly higher volatility they introduce to a portfolio. Since professional investors are after “risk-adjusted return,” factoring in volatility–which the authors calculate using Sharpe ratios—reveals that “the added risk of small-cap stocks is essentially uncompensated.”

While some critics of the small-cap premium have viewed it as a sort of ersatz, analytically less useful cousin of value investing, Kalesnik and Beck go farther, questioning [trigger warning!] whether it even exists as a phenomenon.

They summarize their case that the evidence for a small-cap premium is weak by reminding readers of the absence of statistically significant premia in international markets; the presence of extreme outlier performance driving U.S. performance in the 1930s; and the presence of several biases, particularly delisting, that make the premium vanish.

The authors conclude [trigger warning!]:

“If the size premium were discovered today, rather than in the 1980s, it would be challenging to even publish a paper documenting that small stocks outperform large ones.”

– Related on ThinkAdvisor: