Do new technologies offer more powerful and immediate ways to measure investor sentiment? In the March 2009 version of their paper entitled “In Search of Attention”, Zhi Da, Joseph Engelberg and Pengjie Gao investigate the link between investor attention and asset pricing dynamics based on the levels of and changes in the Google Search Volume Index. Using weekly search frequency data for Russell 3000 and Initial Public Offering (IPO) stock ticker symbols over the period January 2004 through June 2008, along with contemporaneous trading, firm characteristics and news data, they conclude that:
- Ticker symbol search frequency relates positively to market capitalization, abnormal turnover, analyst coverage and news flow, but these factors explain only 5% of the variation in of search frequency among ticker symbols.
- The change in ticker symbol search frequency relates significantly to abnormal return, abnormal turnover and news flow, but these factors explain only 3% of the variation in the change in search frequency among ticker symbols.
- Ticker symbol search frequency actually leads extreme returns and news, consistent with a belief that investors anticipate news events.
- There is a strong and direct link between changes in search frequency and trading by individual retail (naive?) investors.
- IPOs that exhibit relatively large increases in search frequency during the week prior to the IPO outperform IPOs that exhibit relatively small increases in search frequency by nearly 7% during the day after the IPO. These same stocks tend to exhibit long-run return reversals.
- Search frequency-related price pressure emerges mostly for small market capitalization stocks among the Russell 3000. Small stocks exhibiting large increases in search frequency outperform those experiencing large decreases in search frequency by an average of about 0.08% per week during the next two weeks (about 4% annualized), before accounting for trading frictions.
- A momentum strategy applied to stocks with high search frequencies outperforms the same strategy applied to stocks with low search frequencies across all holding horizons. The outperformance over a 52-week holding period is almost 2.5%.
In summary, evidence suggests that Google search frequency data may disproportionately measure the attention of naive investors/traders and therefore indicate some temporary price pressure for small capitalization stocks.
