"The profitability of momentum strategies does indeed decline sharply with market capitalization."
In Bad News Travels Slowly: Size, Analyst Coverage, and the Profitability of Momentum Strategies (NBER Working Paper No. 6553) , Harrison Hong, Terence Lim, and Jeremy Stein cite a number of studies that show that stock returns exhibit momentum: in the medium term past winners continue to perform well, while past losers continue to perform badly. They then go on to ask, what drives momentum?
Standard economic explanations do a poor job of providing an answer, they argue. For example, the evidence simply does not support a standard risk-based approach to explaining momentum. So Hong, Lim, and Stein turn to a "behavioral" approach: they ask whether momentum reflects the gradual diffusion of firm-specific information which, unlike earnings data, is not made publicly available to all investors simultaneously. They predict that stocks with slower information diffusion will exhibit more pronounced momentum.
Their sample, which runs from 1980-96, separates stocks into different classes according to the speed of information diffusion. They use data on stock returns from a file which includes NYSE, AMEX, and NASDAQ stocks, along with data on analyst coverage.
In the first set of tests, the authors sort firms into 10 classes by size and ask whether momentum strategies will be more profitable with the smaller firms. Information about small firms seeps out more slowly than information about big firms, the authors suggest. One reason is that investors face fixed costs for acquiring information. This means that, in the aggregate, investors will devote more effort to learning about those stocks in which they can take larger positions. The authors demonstrate that, aside from some unusual behavior among truly tiny stocks, the profitability of momentum strategies does indeed decline sharply with market capitalization.
In the full sample, a baseline strategy that buys winners, the top 30 percent of firms based on past performance, and shorts losers, the bottom 30 percent, generates 0.53 percent per month return. The researchers break this result down to show that a momentum strategy generally works better for smaller firms. However, one exception is that for the very smallest class of firms, a momentum strategy actually yields a negative result. The researchers argue that this likely reflects very limited investor participation in these tiny stocks - otherwise known as "thin trading" - which can lead to more pronounced supply-shock-induced reversals.
For the second smallest group of firms, momentum profits are significantly positive, though. Profits reach a peak at the third smallest group of firms, in which market capitalization averages $45 million. Here, profits are a striking 1.43 percent per month, three times the value for the sample as a whole. Above this third class, though, momentum profits decline to the point where they are effectively zero for the largest firms.
In a further set of tests, the authors use analyst coverage as an alternative proxy for information flow, on the hypothesis that with less analyst coverage, information gets out more slowly to the investing public. The researchers hold size constant to show that momentum strategies work particularly well among stocks with low analyst coverage. In the low-residual coverage subsample, a momentum strategy yields a profit of 1.13 percent per month. For a high-residual-coverage subsample, it is only 0.72 percent. Thus the evidence confirms the predicted result, that stocks with slower information diffusion exhibit more pronounced momentum.
Hong, Lim and Stein go on to show that there is a strong asymmetry in terms of the effect of analyst coverage for good and bad news. Low coverage stocks react more sluggishly to bad news than to good news, or said differently the effect of analyst coverage is far more pronounced for stocks that are past losers than for past winners. This is consistent with the hypothesis that firm-specific information, particularly bad news, diffuses only gradually across the investing public. The authors suggest that, to the extent that managers prefer higher to lower stock prices, they will seek to publicize good news when they have it. However, when the same managers are sitting on bad news, they clearly have less of an incentive to keep investors up to date.
-- Andrew Balls