Are there exploitable differences in returns for stocks with heavy versus light institutional trading activity? In his March 2008 paper entitled “Trader Composition and the Cross-Section of Stock Returns”, Tao Shu analyzes the impact of institutional trading activity on the returns of individual stocks and on the strength of the momentum effect, post earnings-announcement drift (PEAD), the value premium and the investment effect. He calculates institutional trading activity at a quarterly frequency by dividing the aggregate absolute change in reported institutional holdings of a stock by the contemporaneous total quarterly trading volume for the stock. Using holdings data as reported via SEC Form 13F and associated stock trading volume and return data for the period 1980-2005, he concludes that:
- Institutions account for about half of all trading volume.
- Institutional trading activity is significantly different from institutional ownership, with a correlation of only 0.41.
- Both institutional ownership and firm characteristics affect institutional trading activity, but the most important determinant is historical trading activity. Persistence in institutional trading activity is stronger for stocks with light analyst coverage.
- Stocks with low institutional trading activity underperform stocks with high institutional trading activity by 0.25% to 0.53% per month depending on risk adjustments applied. The difference in performance is most pronounced for liquid stocks (1.03% per month for the most liquid but practically zero for the least liquid).
- Major return anomalies are much stronger in stocks with low institutional trading activity. Controlling for institutional ownership and other factors, stocks in the lowest third of institutional trading in the prior quarter compared to those in the top third (re-ranked monthly) exhibit:
- Momentum (six-month formation period, one-month skip period and six-month holding period) stronger by 0.40% per month.
- PEAD (from two days before announcement through one day after) stronger by 0.46% per month.
- A value (high book-to-market minus low) premium larger by 0.97% per month.
- An investment effect (low capital investment minus high) bigger by 0.57% per month.
- These findings suggest a positive relationship between (presumably informed and sophisticated) institutional trading activity and market efficiency.
In summary, evidence suggests that stocks with low institutional trading activity (distinct from institutional ownership) tend to be overpriced, with amplified return anomalies.
