In a new paper, three quants with Robeco suggest that the “short” side of the activity of many long/short equity trades is pointless.
These quants are David Blitz, Guido Baltussen, who is also affiliated with Erasmus University, Rotterdam, and Pim Van Vliet.
They have broken down the common equity factor strategies into their long and short parts, and have concluded that (1) most added value comes from the long side; (2) the long also offers more diversification that the short, and (3) the performance of the short is generally subsumed by the long. All of this is true even before accounting for the fact that the implementation costs of shorting are substantially higher than those of taking a long position.
There has been some media coverage of this story to the effect that it disputes the validity of short selling in general. That isn’t quite accurate. It is only about the use of shorts in combination with longs in equity factor strategies. Forensic shorting—the strategy of researching a publicly owned company with the thoroughness that Bill Ackman researched MBIA, finding a skeleton in its closet that would undermine value if brought out, and then going short the stock prior to the un-closeting of that skeleton—is not something the authors question. It is simply not on their radar.
What exactly is on their radar?
The paper, entitled “When Equity Factors Drop Their Shorts,” begins with the observation that the long-short strategy, often used in conjunction with the Fama-French factors, presumes that “both legs contain information that is relevant for investor portfolios and for understanding asset prices.” That is the assumption the paper is devoted to shooting down.
And they are not the first to see this on the screen. In 1993 Jegadeesh and Titman argued that there is an asymmetry in the gains from buying winners on the one hand and selling losers on the other. Quite recently (2017), Chu, Hirshleifer, and Ma (2017) explained how the various constraints on short sales account for such an asymmetry.
The authors make their addition to the literature by isolating the two legs statistically. They do this by hedging the beta of a long-only portfolio with the liquid derivatives of long market indexes. This produces the factor performance of the long side. They perform the same procedure on the short side. They apply this method to the following factors: value (high minus low), momentum (winner minus lower), profitability (robust minus weak), and investment (conservative minus aggressive). Of the Fama-French five, then, the one they’re leaving out is size (small minus big) which “is already constructed in a long-minus-market fashion.” They add instead a fifth suggested by Frazzini and Pederson, high/low volatility (also known as “betting against beta”).
Numbers
Blitz et al. use the Sharpe ratio as a key metric. (The Sharpe ratio is a measurement of how much return an investor can expect for each unit of increased risk.) They find that the individual Sharpe ratios of the five factors listed above range from 0.40 to 0.61. If we combine each of the five long legs into an equally weighted portfolio, the Sharpe ratio for that portfolio is 1.10.
When the same is done on the short side, the Sharpe ratio of the equally weighted portfolio is only 0.69.
When the long and short are reunited, the Sharpe ratio for the portfolio is 0.86.
What is key though is that not only do the short legs consistently have worse performance than the long, but they “barely provide any diversification” to the mix, either.
The factor exposure one gets from the short legs had already been subsumed by the long legs. There is, then, no obvious reason not to drop the short from the long/short factor strategies altogether.
Blitz et all do not explicitly consider the extra costs associated with shorting, except to observe that if they had, the results would most likely have strengthened their conclusions that the short side doesn’t add value.
Further research might of course move beyond the familiar Fama-French terrain and look into the hundreds of other factors that have been identified in the literature. As these authors acknowledge, it may well be that for some of them the short side does account for most of the performance of the respective long/short portfolios.