Empowering Investor Knowledge: Learning from the Sophisticates

What can be learned from short sellers? There are two primary reasons for selling a stock short.  Either the short seller is betting that the stock is substantially overvalued, or the short seller is engaging in some hedging- or arbitrage-related activity, such as a merger or a convertibles arbitrage.  In the latter case, the short seller is not betting on the overvaluation of the shorted stock per se, but instead is attempting to exploit relative mis-valuation of two securities.  Thus, the information content of the two sources is different.  However, it is difficult to distinguish between the two, as the data are only available on an aggregate basis. In their recent paper titled “On Distinguishing Between Valuation and Arbitrage Motivated Short Selling,” SMU Cox Professors Hemang Desai and Kumar Venkataraman and co-author Srinivasan Krishnamurthy present a model that disentangles the two sources of short interest.

Background
The short interest data reported in the United States aggregate short selling that is driven by a pessimistic opinion on firm valuation (valuation shorts) and by various arbitrage or hedging strategies (arbitrage shorts). While valuation shorts are motivated by a bearish view on the stock, arbitrage shorts convey little information about the perceived overvaluation of the stock. Thus, the information content of the two sources of short selling is quite different. Disentangling the components of short interest was not as critical for past research because the incidence of arbitrage shorts was relatively modest.

In recent years, short sales motivated by arbitrage strategies have increased significantly, reflecting the explosive growth in hedge funds and institutions that engage in various “market neutral” investment strategies. Confirming this trend, recent research documents that the percentage of shares held short increased dramatically from 1988 to 2002.  More importantly, this research shows that the association between the level of short interest and future returns has weakened in recent years. Given the continuing growth in institutions that engage in arbitrage strategies, this problem is expected to get more acute over time, suggesting that distinguishing between valuation shorts and arbitrage shorts is important.

The Predictive Model
By analyzing firms identified as valuation shorts, the authors develop an empirical model that helps differentiate valuation shorts from arbitrage shorts.  The authors also present new evidence on the information set of short sellers. Although previous theoretical and empirical research has characterized short sellers as informed investors, the empirical evidence on their information set is relatively sparse. This study documents the incremental information in valuation multiples, earnings quality, and other firm characteristics for short sellers. The authors argue that these findings improve our understanding of the information arbitrage process, i.e., identification of mis-priced securities by informed traders. Desai commented, “The two types of short selling—for valuation (speculative) and arbitrage reasons—have different motivations. Valuation shorts occur because the trader believes the stock is over-priced and wishes to short it to benefit from the mis-pricing. Arbitrage shorts, which hedge funds utilize to take advantage of relative over-valuation, may occur, for example, during a merger announcement. The hedge fund trading desk may short the acquiring firm’s stock and long (hold) the target firm’s stock to benefit from mis-pricing during a window of opportunity.” Desai noted that clearly the information content in these two types of short selling is very different.

Recent empirical work supports the notion that short sellers facilitate the flow of private information into stock prices. Thus, this article presents new evidence on the information set and trading behavior of an important group of information arbitragers. The authors also show that valuation short selling is consistently sensitive to information contained in price multiples and earnings quality. Companies with a high price-to-book value tend to under-perform or those with high book-to-market values over-perform; other research has shown that firms with high earnings accruals have lower returns and short sellers target these firms. Desai mentioned that these lower returns exist in the data but that there was very little evidence suggesting that investors were exploiting these mis-pricings. This research gets at that precisely (as did past research by Desai et al. on short sellers and earnings quality in note 1) and goes beyond by separating the two distinctly different types of short selling. Desai states, “Short selling is sustained by differing opinions on the value of a firms’ stock, and the ability to capture the anomaly and benefit from it within a window of time.”

Applications and Implications
Desai mentions that their model is a relatively inexpensive tool that fund managers could use to develop cost-effective portfolio management strategies. The model can also identify over-valued stocks in seasoned equity offerings. For example, conventional wisdom says that firms will issue their stock (not an IPO) when it is over-priced. Subsequently, the market reacts negatively. But if that stock is being issued to fund a positive net present value project, how can this be distinguished before the fact? According to Desai, since their model can identify over-priced stocks, it can be applied to such a situation. And, it can be applied across many different settings in which over-pricings may be an issue. At a minimum, if a stock arises as over-priced in the list generated by the model, investors could unload that stock and minimize their losses. For momentum strategy traders that invest in 100 momentum stocks for example, consider that potentially 70 stocks will maintain their momentum and 30 will not exhibit continuing superior performance. Using the model, the 30 stocks could be identified as over-priced and sold before they actually go down.  Desai stated, “Certainly our model can be used to help refine a number of trading strategies. Or at the least, stock trading strategies can be refined in a more measured manner. And, we can understand and learn from the decision process of some of the most sophisticated investors who add to pricing efficiency in the market.”

The paper “On Distinguishing Between Valuation and Arbitrage Motivated Short Selling,” written by Hemang Desai and Kumar Venkataraman of the SMU Cox School of Business and Srinivasan Krishnamurthy of SUNY-Binghamton University, is currently under review.

(1) See also ¨ Desai, H., Krishnamurthy, S., Venkataraman, K., 2006, “Do Short Sellers Target Firms with Poor Earnings Quality? Evidence from Earnings Restatements,” Review of Accounting Studies, Vol. 11, pg. 71-90. http://kvenkataraman.cox.smu.edu/papers/dkv1.pdf

Summary by Jennifer Warren