Last fall, Google’s philanthropic division released Flu Trends, a tool that purportedly predicts regional influenza outbreaks 7-10 days faster than traditional methods. The bio-surveillance tool relies on the fact that people use Google to search flu-related terms well before calling their physicians.
Now, a study of similar methodologies appears to show that the Mountain View-based company’s omnipotence extends to the prediction of economic trends as well.
Hal Varian, an economics professor at UC Berkeley who moonlights as Google’s chief economist, and Hyunyoung Choi, a Google employee tested the hypothesis that variations in search frequency for certain phrases improves the accuracy of econometric models used to forecast retail and home sales, among other things.
Such data are available to the public through Google Trends, which enables interested parties to access reports on search volumes for particular categories and terms. The reports are updated daily.
The scientists found that addition of such information improves the predictive value of the standard models used to forecast car and truck sales by 18%.
Similarly, search volume on terms like Hong Kong and other ports of call carried out in Australia, India, the UK and the US can foretell bumps in tourist volume to these locations.
The tool still needs refining, however. The scientists showed for example that searches for real estate agents are better predictors of future home sales than those for home financing.