Researchers from AQR Capital Management, a Connecticut-based asset manager, claim they have uncovered the source of Warren Buffett’s alpha. They believe Buffett-like performance can be achieved by constructing a portfolio with exposure to certain “factors.” But their theory hinges on a crucial assumption, which, as I will show, is highly tenuous.

Buffett’s returns, based on Berkshire Hathaway’s stock price, didn’t come from successful stock-picking, AQR claims. Rather, the source of his outperformance was from his exposure to low-beta and high-quality stocks.

AQR and Dimensional Fund Advisors (DFA) have begun to incorporate those factors into their portfolio construction. This research raises the hope of duplicating Buffett’s success in a conventional, open-ended mutual fund.

I’ll look at the findings of this research and why I am skeptical of its implications.

Buffett’s alpha

Buffett has outperformed every stock and mutual fund on an absolute and risk-adjusted basis over the last 30 years. Previous research has failed to identify the sources of Buffett’s returns through the standard four-factor Fama-French model. Those studies found that Buffett’s returns could not be explained by exposure to value, market capitalization or momentum. His returns were mathematically equivalent to alpha.

Whether that alpha is due to luck or skill has been a matter of debate. Buffett’s position, stated in a 1984 article based on a talk he gave at Columbia University and in other writings, is that it is skill. Buffett attributes his skill to a rigorous application of the value-investing principles laid out by his one-time teachers, Benjamin Graham and David Dodd, most notably in their 1934 book, Securities Analysis.

Andrea Frazzini, David Kabiller and Lasse Heje Pedersen of AQR published Buffett’s Alpha in November of last year. The AQR researchers essentially expanded the Fama-French framework from a four- to a six-factor model, adding exposure to low-beta stocks and exposure to high-quality stocks. Those two factors, they claimed, explain Buffett’s success, eliminating his alpha.

They wrote -

“Buffett’s performance appears not to be luck, but an expression that value and quality investing can be implemented in an actual portfolio (although of course, not by all investors who must collectively hold the market).”

The average beta of Buffett’s public stocks was 0.77, and Berkshire Hathaway’s (BRK-B) beta was 0.68. The AQR researchers attribute some of Buffett’s excess return to this low-beta exposure, consistent with other research. But a greater contributor to returns was Buffett’s exposure to high-quality stocks, according to the authors. High-quality stocks are defined as those of companies that are profitable, stable and growing and have high payout ratios.

They also attribute some of Buffett’s success to his access to cheap leverage. Through its insurance operations, Berkshire Hathaway invests the cash flow from policy premiums, in advance of when claims need to be paid.

The authors wrote -

“In essence, we find that the secret to Buffett’s success is his preference for cheap, safe, high-quality stocks combined with his consistent use of leverage to magnify returns while surviving the inevitable large absolute and relative drawdowns this entails.”

The authors constructed a portfolio of public securities with characteristics (factor loadings) that mirror Berkshire Hathaway’s to demonstrate that investors can achieve returns similar to Buffett’s. They wrote -

“Buffett’s success shows that the high returns of these academic factors are not just ‘paper returns,’ but these returns could be realized in the real world after transaction costs and funding costs, at least by Warren Buffett.”

Will it work in the real world?

Michael Edesess previously criticized this line of research, claiming that excess returns cannot be attributed to exposure to high-quality stocks. I agree with Edesess. But let’s put that aside for now and assume that high-quality (and low-beta) stocks outperform the market and that this effect can be documented in out-of-sample tests.

My problem is that none of this reflects the way Buffett selects stocks. In some cases, Buffett has advocated the opposite of what the AQR researchers presented.

Buffett’s guiding precept has been to purchase companies with prices below their intrinsic values – companies that trade “at 40 cents on the dollar.” Nowhere will you find a statement from Buffett that he prefers low-beta stocks, or even that he considers beta a relevant characteristic in stock selection. In his 1984 article, he even wrote that under certain circumstances, a higher beta would be preferable. He cited the example of the Washington Post Company, which was worth $400 million in 1973, although it was trading for $80 million.

Buffett wrote -

“Now if the stock had declined even further to a price that made the valuation $40 million instead of $80 million, its beta would have been even greater. And to people who think beta measures risk, the cheaper price would have made it look riskier. This is truly Alice in Wonderland.”

The AQR authors present only weak evidence that Buffett prefers the type of high-quality companies they modeled in their research. They quote Buffett’s 2008 annual report, where he wrote,

“Whether we’re talking about socks or stocks, I like buying quality merchandise when it is marked down.”

They quote his 1989 annual report, where he wrote,

“I could give you other personal examples of ‘bargain-purchase’ folly but I'm sure you get the picture: It's far better to buy a wonderful company at a fair price than a fair company at a wonderful price.”

Profitability, stability, growth and high-payout ratios are surely among the criteria Buffett considers when evaluating a company, but they only matter in the context of the price he must pay. We can turn to Graham and Dodd to see the terms Buffett in which thinks of quality:

The untrained buyer fares best by purchasing goods of the highest reputation, even though he may pay a comparatively high price. But, needless to say, this is not a rule to guide the expert merchandise buyer, for he is expected to judge quality by examination and not solely by reputation, and at times he may even sacrifice certain definite degrees of quality if that which he obtains is adequate for his purpose and attractive in price. This distinction applies as well to the purchase of securities as to buying paints or watches. It results in two principles of quite opposite character, the one suitable for the untrained investor, the other useful only to the analyst.

Graham and Dodd emphasized the need to estimate future corporate profits, not the historical profits that are captured by the AQR models. Much of Graham and Dodd’s text is devoted to how to analyze competitive positioning and the stock market’s preference for certain types of companies at a given point in time. Dividends are appropriate to consider, according to Graham and Dodd, but the analysis should be directed to the influence of interest rates upon them.

Indeed, Graham and Dodd warned against an over-emphasis on quality:

“The great majority of common stocks of strong companies must be considered speculative during most of the time, simply because their price is too high to warrant safety of principal in any intelligible sense of the phrase.

We must warn the reader that prevailing Wall Street opinion does not agree with us on this point; and he must make up his own mind which of us is wrong.”

The ultimate irony

The question advisors must answer is whether the mathematical models created by firms like AQR and DFA will lead to Buffet-like performance in the future, as the AQR paper claims. If those models reflected Buffett’s investing strategy and, for example, were able to identify companies trading at 40 cents on the dollar, I might be persuaded to have faith in them.

But the models don’t do that, so their past results could reflect correlation without causation, lacking any predictive properties.

AQR’s analysis neglects what is surely Buffett’s underlying advantage. Buffett is not constrained by the burdens of an open-ended mutual fund. He is not concerned with short-term performance and the effect it might have on asset flows, nor with being measured against a benchmark or having his returns compared to a peer group. He doesn’t have to deal with daily redemptions. He isn’t concerned with how Morningstar rates his fund, and his revenue model is not based on assets under management. Investors can buy Berkshire Hathaway at essentially no cost.

Whatever the chances are that AQR's approach in a mutual fund could approximate Berkshire's returns on a pre-tax basis, they are far lower on an after-tax basis.  For example, Berkshire's shareholders enjoy the benefit of paying taxes at capital-gains rates only when securities are sold.

We can be confident what Buffett would say about the type of research published by AQR. In his 1984 article, he wrote,

“Of course, the reason a lot of studies are made of these price and volume variables is that now, in the age of computers, there are almost endless data available about them. It isn’t necessarily because such studies have any utility; it’s simply that the data are there and academicians have worked hard to learn the mathematical skills needed to manipulate them.”

The ultimate irony will unfold as funds that are based on low-beta and high-quality stocks accumulate assets. That will present the type of herd- and momentum-driven buying that creates opportunities for value-driven investors like Buffett. He and the other disciples of Graham and Dodd will find their 40-cent bargains among the securities disdained by those funds.