| This paper reports on an empirical investigation of the relationship between long-term stock returns and the percentage of institutional stock ownership for the period from July 1982 to December 1994. It is shown that, on average, the biggest high-IO stocks outperform their low-IO size and book-to-market matches by roughly 60 basis points per month. Because in IO-sorted groups average contemporaneous IO change is always larger for low-IO stocks than for high-IO stocks, and because the positive relation between returns and contemporaneous IO changes is stronger for relatively smaller stocks, the IO effect is substantially attenuated in portfolios that go long on relatively smaller high-IO stocks and short on their low-IO size and book-to-market matches. If IO change is controlled for, the IO effect is strong for all the sample stocks from the top four NYSE size quintiles. This suggests that a successful forecasting model for IO changes could improve trading strategies aimed at profiting from the IO effect.;The paper analyzes whether the IO effect could have attenuated the size effect starting in the 80's. When IO is controlled for, the size effect is pronounced within the group of the smallest stocks but nonexistent within the top four NYSE size quintiles. This suggests that the IO effect cannot fully account for the reversal of the size effect during the sample period.;The paper attempts to uncover the nature of the IO effect and shows that the documented anomaly cannot be explained by price pressure, momentum, stock turnover, or the S&P500 effect. In addition, the paper presents the evidence in favor of an explanation of the S&P500 effect based on a combination of the IO effect and increases in the index liquidity. |
