Source: Da, Z., Engelberg, J. & Gao, P. (2011) · Journal of Finance 66(5), 1461–1499 · DOI: 10.1111/j.1540-6261.2011.01679.x
TL;DR
Proposes Google Search Volume Index (SVI) as a direct measure of investor attention — chiefly retail attention — instead of inferring it from extreme returns, volume, news, or advertising. In Russell 3000 stocks (2004–2008), a one-standard-deviation rise in abnormal SVI (ASVI) predicts a positive price change of more than 30 bps over the next two weeks, fully reversed within a year (−28.9 bps long-run), and high pre-IPO search predicts larger first-day IPO returns and subsequent underperformance. This confirms Barber & Odean's (2008) price-pressure hypothesis with a clean, timely proxy.
The idea
Attention is a scarce cognitive resource, so prices reflect information only to the extent investors attend to it. Prior attention proxies (extreme returns, turnover, news, ads, price limits) are indirect — they assume an event drew attention. Searching for a ticker is revealed attention: if you Google a stock, you are unambiguously paying attention to it. SVI (weekly searches scaled by their time-series average, from Google Trends, back to Jan 2004) therefore measures attention directly and in real time, and skews toward retail investors.
Evidence
Sample: Russell 3000 stocks, January 2004 – June 2008; IPOs January 2004 – December 2007 (Thomson/Reuters SDC).
Timeliness & who: ASVI is correlated with but distinct from existing proxies, leads (rather than lags) news- and volume-based measures, and behaves like retail attention (e.g., a 1% rise in SVI maps to a far larger response than in institutional measures).
Price pressure → reversal: a one-SD ASVI increase predicts a significant positive return of +30 bps over two weeks (week-by-week ≈ 18.7 bps, then 14.9 bps, fading to 3.85 bps and −1.6 bps), with a long-run reversal of −28.9 bps — initial pressure is fully unwound.
IPOs: high ASVI in the IPO week predicts larger first-day returns (a difference of ≈9.11% between high- and low-ASVI groups) followed by long-run underperformance — consistent with retail-driven price pressure. Media-coverage attention does not produce the same reversal.
Why it matters
A landmark in alternative-data finance: it showed that web/search data is a measurable, predictive, near-real-time attention signal, and gave behavioral attention theories (Merton 1987; Barber & Odean 2008) a direct empirical handle. It seeded the now-standard use of search, social-media, app-usage, and other digital-footprint proxies in quantitative research (e.g., the authors' later FEARS index).
Caveats
The effect is short-horizon and reverses; transaction costs and capacity limit exploitation.
Ticker ambiguity (tickers that are also common words) and Google Trends sampling noise/revisions complicate clean construction — results strengthen where Google's sampling-error SD is low.
It captures retail attention specifically, not institutional information processing.
Key references
Da, Z., Engelberg, J. & Gao, P. (2011) — In Search of Attention — Journal of Finance
Barber, B. & Odean, T. (2008) — All That Glitters: The Effect of Attention on the Buying Behavior of Individual and Institutional Investors — Review of Financial Studies
Merton, R. (1987) — A Simple Model of Capital Market Equilibrium with Incomplete Information — Journal of Finance
Da, Z., Engelberg, J. & Gao, P. (2015) — The Sum of All FEARS: Investor Sentiment and Asset Prices — Review of Financial Studies
Provenance: verified/generated from the paper's full text.
