Internet Search and Momentum

Abstract

In a sample of Russell 3000 stocks from 2004 to 2008, we find the momentum effect (Jegadeesh and Titman (1993) and Gutierrez and Kelley (2008)) to be much stronger among those searched more in Google. Our result provides strong support for the overconfidence model as in Daniel, Hirshleifer, and Subrahmanyan (1998) for several reasons: (1) retail investors are more likely to search a stock in internet and are more likely to suffer from overconfidence and other behavioral biases; (2) investors are likely to be even more overconfident in stocks that they search and analyze on the internet; (3) as searches in Google are often led to the same information, the private signals to the investors are in fact highly correlated, thus exacerbate the impact of overconfidence. We confirm that the relation between internet search and momentum effect is not driven by other variables such as news events, size, analyst coverage, and trading volume, which have been documented to be related to the momentum effect. Finally, we find the stronger momentum effect among stocks searched more in Google mainly comes from the winners, consistent with the insight in Barber and Odean (2008) that retail investors on average purchase attention-grabbing stocks.