A Stupid Question About Stocks Yields a Surprising Answer

University of Miami economics professor Alex Horenstein asked a stupid question, and got a surprising answer. He wanted to know how the capital asset pricing model affected stock returns. At first blush, that sounds sensible. Academics have always thought about how their research might affect markets.

One of the earliest and strongest patterns turned up by researchers was how small-capitalization stocks have better risk-adjusted returns than large-capitalization stocks. Once that was published, you might have expected investors to sell large cap stocks and buy small cap stocks, pushing down large cap prices and increasing their future returns while pushing up small cap stock prices and reducing their future returns, until the two had equivalent expected returns.

That didn't happen for small cap stocks, nor for other similar findings. It's true that many patterns turned up by academic researchers proved ephemeral or temporary, but in no case did publication seem to play a role in their demises. It's also true that products are launched to exploit these anomalies, but they don't seem to eliminate – or even attenuate – them. Moreover, a handful of patterns, such as size, value, momentum, quality and others have been found in nearly every financial market studied and have persisted as far back as we can get data.

But the capital asset pricing model, or CAPM, is not like those patterns. It claims that all stocks are fairly priced. It doesn't matter which stocks you buy, as long as you buy enough to get a diversified portfolio set to the risk level you choose. It's stupid to ask if it affects stock choices, because it says those choices don't matter. If theories that clearly help investors get ignored, why would theories that give no stock selection advice affect decisions?

Shockingly, Horenstein found that after publication of CAPM – but not before – stocks with low alpha over a five-year period did significantly better over the following year than stocks with high alpha over the same five-year period. Moreover, when academic consensus moved in the early 1990s from single- to multi-factor CAPM, multi-factor alpha replaced single-factor alpha in that pattern.

Alpha is the CAPM measure of risk-adjusted performance. For example, over the last year the S&P 500 is up about 7 percent and low-risk investments about 2 percent. CAPM predicts that stocks should return about 2 percent (the low-risk return) plus their betas times 5 percent (the market return minus the low-risk return). Coca-Cola with a 0.4 beta should have returned 4 percent, while Celgene with a 1.8 beta should have returned 11 percent. Coca-Cola actually returned 6 percent for an alpha of positive 2 percent, while Celgene returned 9 percent for an alpha of negative 2 percent. Celgene did better but Coca-Cola had better risk-adjusted performance. Horenstein discovered that stocks like Celgene seem to be better future investments than stocks like Coca-Cola - but only since 1964.

This means Horenstein has a new factor, which he calls “betting against alpha.” Investors should be cautious about exploiting it. Hundreds of new factors are proposed every year, and almost all turn out either not to work in the future or are subsumed by better-documented factors. Horenstein has done solid research to support his innovation but the odds are against it entering the pantheon of consensus factors.

The best factors have been documented by hundreds of researchers in many financial markets and eras and proven in decades of application. Betting against alpha correlates with accepted factors like long-term reversal (0.53), value (0.43) and conservative (0.32). Horenstein documents excess performance after adjusting for these factors, but given the high correlations, it may turn out that his analysis leads to improvements in the measurement of other factors rather than adding a new factor.

Many investors did take to heart the lessons from the body of work that includes CAPM was well as modern portfolio theory (MPT) and the efficient market hypothesis (EMH). They increased diversification, cut the fees they were willing to pay, benchmarked their advisers and moved to more passive and index investments.

But lots of investors apparently learned the math, but not the economics. They switched from chasing stocks that had gone up in the past to chasing stocks that improved risk-return ratios of diversified portfolios—i.e stocks with positive alpha in the past. This doesn't require CAPM specifically, because any reasonable way of evaluating risk-adjusted return could explain Horenstein's results. The key is investors accept MPT—investments should be evaluated by their effect on portfolio statistical properties rather than non-statistical or investment-specific criteria—and most importantly, that they believe past performance is indicative of future results.

After six decades of fruitless searching for evidence that investors exploit useful academic knowledge to pick stocks, someone has demonstrated that investors pick stocks using academic research that stock picking is useless. Horenstein calls it an “unintended impact,” but that's an understatement. Long before there was CAPM, EMH or MPT, we knew that chasing past results is a bad strategy in investing, gambling and life. MPT appears to have redirected the direction of the chase, but it hasn't taught it investors not to chase.