Attention bubbles

Stock-pickers have no shortage of buy signals, be they fundamental indicators of earnings and cash flow, or technical trading signals. There is, though, another type of indicator to consider - the frequency with which shares are searched for in Google. New research by Zhi Da and Pengjie Gao of the University of Notre Dame and Joseph Engelberg at the University of North Carolina has found that this can predict returns.

They measured abnormal search volume - the number of searches for a share's ticker symbol in one week, relative to the number in the previous eight weeks. They found that lots of searches led to share prices rising. A one standard deviation rise in search volume led to shares outperforming by an average of 0.3 percentage points in the following two weeks.

The reason for this is simple. A lot of searches for a stock is a sign that investors are paying attention to it - which is often a precursor to them buying it. This is consistent with other evidence that Google search data has predictive power. Hal Varian, Google's chief economist, has found that search volume is correlated with sales of cars and houses. And Google researcher Jeremy Ginsberg shows that searches can be a more timely indicator of 'flu epidemics than official data.

You might think there is a trivial reason for this correlation. If a stock enjoys some good news, you'd expect it to both attract investors' attention and enjoy a price rise.

But this is not the whole story. If this were the case, you'd expect the share to rise to a permanently higher level in response to that good news. But it doesn't. The researchers found that, a month after the peak in search volume, share prices start to fall. This is inconsistent with the idea that genuine news drives both prices and search volume. Instead, it suggests there might be bubbles in investors' attention that push up prices too far.

This explanation is consistent with a better-known anomaly - the initial public offering effect, whereby newly floated shares do well initially after coming to market, only to fall thereafter. This might be a sign of attention bubbles, too; a new flotation grabs investors' attention, which causes them to buy too much.

The message here is perhaps not that we should use Google trends as a buy signal. 0.3 per cent is quite a small effect, and it is (so far) only established for US stocks, not UK ones. Instead, there's a more general warning. We should be wary of buying stocks after they have received lots of media and investor attention. You don't find under-priced stocks in the same place that everyone else is looking for them.

visible-status-Standard story-url-attention_markets_250211.xml