web server statistics
American Accounting Association

Ben Haimowitz (212-233-6170)

Short-selling is a far better guide for investors than the recommendations of analysts,
study finds
When shorts and analysts disagree, it pays to buy when the latter advise to sell

The research started out as an exploration of a trading strategy based on agreement between stock analysts and short-sellers: buy stocks with highly favorable analyst recommendations and a low level of short interest; sell-short stocks that evoke strongly unfavorable analyst recommendations and a high level of short interest.

Sensible though the approach seemed, its gains proved meager, a mere 0.36% abnormal returns per month.

Then, a contrarian intuition led the researchers to turn the strategy on its head and to combine the insights of short-sellers with the very opposite of what analysts recommended. In other words, buy if short interest is low and the analyst consensus is to sell; sell short if short interest is high and the analyst consensus is to buy.

Now the results proved anything but meager -- a robust 1.11% average abnormal return per month over the course of 13 years, amounting to an average abnormal yearly return of more than 14%

In other words, returns surpassed those achieved from investing in similar firms by roughly 14 percentage points a year.

In the words of the study, published in the current issue of the American Accounting Association journal The Accounting Review, "Monthly abnormal returns are insignificant for portfolios containing stocks about which analysts and short sellers strongly concur...The most profitable investment strategy is when an investor trades in firms about which analysts and short sellers strongly disagree, and the investor takes the buy or sell position indicated by the short interest signal."

The findings lead to the conclusion that, in the study's words, "short sellers are under-represented in price formation whenever they disagree with analysts, regardless of whether they are the optimists or the pessimists." Indeed, alluding to restrictions imposed on short-sellers during the financial crisis of 2008, the report warns against such rules' becoming permanent, noting that "an important implication of our study is that regulations that restrict or increase the cost of short selling run the risk of limiting a potentially important source of information for investors about future equity values."

Comments Edward P. Swanson of the Mays Business School at Texas A&M University, who carried out the research with his Mays colleague Lynn Rees and Michael S. Drake of the Fisher School at Ohio State University, "It seems ironic that information about short interest, which our study shows to be of considerable value to investors, is so much scarcer than data about analyst recommendations, which our study finds somewhat less reliable as a investment guide than tossing a coin would be. Yet, any number of Internet sites provide consensus recommendations for thousands of companies, while short interest data is much harder for the average investor to find."

Prof. Rees adds: "Scholarly research on the information value of short sales is much sparser than research on analysts' recommendations. Although our study is hardly unique in raising doubts about those recommendations, our findings are particularly dramatic because we're the first to assess them in combination with short interest; in so doing, we discovered to our amazement that investing opposite to what analysts recommend actually enhances returns."

The findings derive from databases of analyst recommendations, short-interest information, and stock returns for the years 1994 through 2006. For each of the 156 months in that period, the large sample of companies was divided into quintiles in two ways, each based on a different factor -- the favorability of analyst recommendations and the amount of short interest. The strategy each month involved devoting equal resources to two kinds of trades -- 1) buying stocks that were in the lowest quintile of short interest (suggesting that they were unlikely to fall) but also in the lowest quintile of analyst favorability; and 2) selling short stocks that were in the highest quintile of short interest (suggesting they were likely to fall) but also in the highest quintile of favorability. Every month the quintiles were re-sorted, based on current data. All positions, whether involving buying or selling short, were for six months, at which point they would close out, meaning that every month one buy position and one sell-short position would close out, five of each would continue, and one new one of each would be inaugurated based on latest data.

Based on data from 564,101 firm-months over 13 years, the professors report an average monthly abnormal return of 0.4% for the buy strategy and of 0.71% for the short strategy, for a combined total of 1.11% per month and about 14.2% per year. (Abnormal in this context means in comparison to companies of similar size, book-to-market ratio, and stock momentum.) In other tests, they find that the strategy would have yielded high returns in three contrasting sub-periods -- in the strong bull market of 1994-98 (1.1% monthly abnormal return), during the precipitous NASDAQ decline and major regulatory initiatives of 1999-2003 (1.3% monthly abnormal return), and in the post-Enron years of 2004-06 (0.71% monthly abnormal return).

What accounts for the fact that short interest proves so decidedly superior to analyst recommendations as an investment guide? The study shows that shorts make better use of accounting information that researchers have found to be predictive of future returns and that shorts are suspicious of high growth. Prof. Swanson adds: "Sophisticated investors don't pay as much attention to analysts' recommendations as they do to other aspects of their output, such as earnings forecasts or their narrative discussions of companies. In fact there is a fair amount of research among accounting scholars which questions whether analysts' recommendations are even consistent with their earnings estimates and which suggests that special incentives come into play in the issuing of recommendations."

Still, the professor expresses surprise that analysts' recommendations prove to be not just inaccurate but so consistently wrong that money is to be made by trading against them. In the words of the study (emphases included), "Abnormal returns generally decrease as the recommendation level increases. When we calculate returns from investing long in firms classified in the most favorable recommendation quintile and taking an offsetting short position in firms in the least favorable recommendation quintile, we obtain a modestly negative return of 26 monthly basis points."

In other words, a slight loss.

The finding that investing contrary to analysts' advice pays off is "as rock solid as anything you'll ever see," Prof. Swanson says. While the exact strategy described in the paper is beyond the capacity of most individual investors, because of the sheer volume of trading involved, the professor sees lessons useful to that group. "If a stock looks attractive, perhaps for fundamental reason, but the consensus recommendation is poor to middling, take an additional step and check out the amount of short interest. If it's about 5% or less, consider the unfavorable recommendation a green light. Yes, it's natural to wish for a favorable recommendation, but that would merely suggest that all the good news is already in the price of the stock. Analysts tend to recommend glamour stocks, so being receptive to lower recommendations is likely to represent a move toward value."

The study, entitled "Should Investors Follow the Prophets or the Bears? Evidence on the Use of Public Information by Analysts and Short Sellers," is in the January/February issue of The Accounting Review, published six times a year by the American Accounting Association, a worldwide organization devoted to excellence in accounting education, research, and practice. Other journals published by the AAA or its specialty sections include Accounting Horizons, Issues in Accounting Education, AUDITING: A Journal of Practice and Theory, Behavioral Research in Accounting, The Journal of the American Taxation Association, and The Journal of Management Accounting Research.


© 2010 American Accounting Association All rights reserved.