A number of bloggers seem perplexed that Larry Summers is using Google Trends to get a feel for economic behavior. Ed Morrissey says:

I wonder whether a Google search metric falls within the Keynesian or Friedman schools of economic thought. If I had to guess, which is apparently what Summers is doing when he�s not busy counting Google searches, I�d say it falls within the Marx school of economic thought �. Groucho Marx.

But this really isn't news. Hal Varian and Hyunyoung Choi find that compared to time series analysis, including the results of Google Trends or Insight will improve predictions of the retail activity. In short, expectations matter, and people engaging in searches for goods will use Google. The research was mentioned in The Economist in April.

And Summers is right, but you'll note searches drop every summer. (Graph idea from here.) My argument? It's students researching papers for classes, which they don't take in the summer. But that's a SWAG, not something I would say at the White House.
The question is whether you could use that graph to help predict what happens in overall economic activity. On eyeball, it doesn't look so. But some undergraduate now has a nice paper to write.

Labels: economics