A recent paper correlating days-to-cover with stock returns finds a DTC premium effect and suggests that hedge funds might want to arbitrage it.
This study begins with the crowded trade problem. With so many sophisticated investors in the markets, looking for or at the same arb opportunities, a number of them can sometimes find themselves on the same side of a particular trade, looking for an exit. That’s a quite common observation.
But this paper, by Harrison Hong of Princeton University, and four colleagues, distinguishes between strategy-specific and generic sources of crowding. Trades get crowded “naturally … in price-untethered quantitative strategies such as momentum,” and they cite recent literature on this point, including Jeremy C. Stein’s 2009 address as president of the American Finance Association.
But … Hong et al. distinguish such specific situations from the “generic” crowding of positions that can result from illiquidity irrespective of the strategies pursued by the sophisticated investors, as the aggregate position they hold gets too large for the markets’ liquidity.
Short Sales
To focus on that, they look at short sales. There is a convenient quantitative measure attached to this sort of trade: the ratio of shares shorted to shares outstanding (SR), which is a good predictor of negative returns. Its predictive value suggests that the short ratio represents “informed arbitrageurs trading against mispricing.”
Another important fact about short selling, one that helps illustrate these authors’ concern with the generic cause of overcrowding, is the way in which short sales are “implicitly levered in that the loan of the shares might be recalled at any moment, thereby triggering short-covering,” which in turn can trigger quick price moves.
That implicitly levered character suggests another metric that is of interest to our authors: Days to Cover (DTC). This is the division of the short ratio by the average daily share turnover, and it is a “natural statistic for measuring the crowdedness of short trades.”
The DTC Premium
By way of illustration, consider hypothetical stocks X and Y. X has a short ratio of 5% and an average daily turnover of 1%. This means its days to cover are 5. Y on the other hand has the same short ratio, 5%, but its average daily turnover is also 5%. Y, then, has a DTC of 1. This is monitored by short sellers as a risk management tool. Other things being equal, a short seller will prefer to have borrowed Y rather than X.
Also, and again as a ceteris paribus matter, one would expect that the low DTC stock will be better performing than its counterpart: Y will outperform X. The market has to compensate the arbs that enter the tough-to-exit positions. Thus, a strategy of going long low DTC decile stocks while shorting high DTC decile stocks should yield alpha.
Their data shows that this is, in fact, the case, both with a straightforward DTC-arb strategy and with a value-weighted variant. The former yields 1.19% per month. The latter yields .67% per month. Further, the predictive power of the DTC works “across a variety of return benchmarks.” It isn’t simply picking up a liquidity effect.
Lending Fees
Much of the paper is devoted to distinguishing the DTC-premium effect predicted by the author’s model from other effects and statistical artifacts with which it might be confused. For example, the authors consider how their results might change once lending fees are added into the calculations. For this purpose, they use three distinct measures of lending fees, adroitly labelled Fee1, Fee2, and SIO. The nomenclature reminds one a bit of Dr. Seuss’ “Thing 1” and “Thing 2,” but never mind that.
What is the difference? Fee1 is simply the average of the observed fee in hedge fund borrowing transactions. Fee2 is Markit’s estimate of the lending fee. Fee3, or rather SIO, is the short interest scaled by institutional ownership. On no reading can the predictive power of the DTC be said to vanish into the consequences of lending fees.
Aside from Hong, the authors of the paper are: Frank Weikai Li, Sophie X. Ni, José A. Scheinkman, and Philip Yan. Li and Ni are both with the Hong Kong University of Science and Technology. Scheinkman is with Princeton, Yan with Goldman Sachs.