Google Inc. (NASDAQ:GOOG) search terms are usually analyzed by companies looking for personal—or at least demographic—information, but a team of researchers at the C.D. Howe Institute claim that search terms can be used to predict a recession faster than traditional financial data because it can be done in real time, unlike stats like the unemployment rate and GDP growth, which aren’t available until months after the fact, according to a Canadian Press report.
Google search terms predicting recession
Greg Tkacz looked at the prevalence of the search terms ‘recession’ and ‘jobs’ ahead of the 2008-2009 financial crisis, and found that there was enough of a correlation that he could have predicted the recession three months before its official onset. He warns that Google Trends only provides information on searches going back to 2004, so he doesn’t have enough data to say anything conclusive.
The Children's Investment Fund Management LLP is a London-based hedge fund firm better known by its acronym TCI. Founded by Sir Chris Hohn in 2003, the fund has a global mandate and supports the Children's Investment Fund Foundation (CIFF). Q3 2021 hedge fund letters, conferences and more The CIFF was established in 2002 by Hohn Read More
While the idea is certainly promising, there is an open question about what exactly these search terms are measuring. Even if there is a consistent, predictable correlation between certain search terms and the financial crisis, when the recession became official it was already old news. Most people had already accepted that the economy was in serious trouble, the only question was whether it would go flat or actually contract.
With people getting laid off and news programs covering the unfolding crisis, it would probably be more surprising if those particular search terms weren’t trending.
Google trends vs traditional methods
Other researchers have had mixed results working with Google Trends. A team of journalists who tried to predict the outcomes of U.S. elections using Google Inc. (NASDAQ:GOOG) search terms found that it was less reliable than polling and other traditional methods. Again, questions about what is really being measured by search terms and the ways that populations skew (different demographics use the internet in different amounts and in different ways), and making the assumption that searching for a candidate implies support, or that the search term ‘jobs’ implies a recent firing, is somewhat bold.
Until these issues can be dealt with, Google Trends will probably remain a curiosity used for after-the-fact analysis instead of legitimate policy tools that form the basis of important decisions.