Blog - Investing Notes

August 21, 2007 - Using Insider Trading to Find Informed Short Sellers

Conventional wisdom says that both short sellers and corporate insiders are typically better informed than most traders. However, much short selling comes from programmed (uninformed) hedging, and much insider trading is pre-planned diversification of concentrated positions by firm executives. Is there a way to overlay the activities of these two groups to isolate truly informed trading? In their July 2007 draft paper entitled "Shorts and Insiders", Amiyatosh Purnanandam and Nejat Seyhun investigate the combined power of unusual levels of short interest and unusual insider trading to predict stock returns. They test for "unusual" short interest and insider trading by subtracting the historical mean from the current value and dividing this difference by the historical standard deviation on a firm-by-firm basis. Using monthly short interest, insider trading and stock return data for all NYSE/AMEX-listed firms during 9/91-12/03, they find that:

• The one third of firms with the unusually highest (lowest) levels of short interest, rebalanced monthly, earn an average monthly equal-weighted return of -0.29% (0.49%) after adjusting for market return, size, book-to-market and momentum.
• When insider trading is unusually negative, the one third of stocks with unusually highest short interest earn an average risk-adjusted monthly return of -0.49% (-0.45%) on an equal-weighted (value-weighted) basis. For stocks with insiders unusually negative but short sellers unusually positive, average future stock returns are not significantly abnormal.
• When insider trading is unusually positive, the one third of stocks with unusually lowest short interest earn an average risk-adjusted monthly return of 0.73% (0.37%) on an equal-weighted (value-weighted) basis. For stocks with insiders unusually positive but short sellers unusually negative, average future stock returns are not significantly abnormal.
• When insiders are neutral, the predictive power of unusual levels of short interest is weakened, especially for large capitalization stocks.
• A hedging strategy long the third of stocks with unusually positive insiders and short sellers and short the third of stocks with unusually negative insiders and short sellers generates an average monthly return of 1.22% (0.82%) on an equal-weighted (value-weighted) basis, excluding transaction costs.

The following chart, constructed from data in the paper, summarizes raw average monthly returns (before transaction costs) for portfolios based on unusual levels of insider trading and short interest during 1993-2003. "Unusual" is defined above in the introductory paragraph. Portfolio rebalancing is monthly. Results show that combining the two indicators enhances predictive power for future returns.

In summary, unusual trading by insiders helps isolate which short sellers know what they are doing, and vice versa.

The authors assert that the returns for the above hedging strategy "are economically large...and unlikely to be explained away by transaction costs." We would like to see that assertion tested.

For related research, see Blog Synthesis: Short Selling and Short Interest.