Google eyes trends as economic indicators

Share

Move over, Statistics Canada and U.S. Bureau of Labour Statistics. There's a new kid on the block.

Google Inc. has a research project under way to use its mammoth database to start a "Google Price Index," a real-time measure of inflation.

The concept was discussed by the company's chief economist Hal Varian at a weekend conference of the U.S. National Association for Business Economics.

It comes as Google is expanding its analysis to examine how trends on the world's largest search engine might also be economic indicators.

The drawback of Google's inflation tracking is that it can only measure price changes of purchases made over the Internet. So shampoo, parking tickets and rent are unlikely to show up in its data.

The advantage, though, is timeliness - Google's data is available continuously, while official agencies lag by weeks. Statistics Canada, for example, typically releases its consumer price index in the third week of the month, based on prices collected by 80 people in stores a month earlier.

The new price index is unlikely to replace official statistics any time soon, but may become a useful supplement in an era when real-time news, and trading, is becoming more prevalent.

"If the information is useful, then why not?" said Kam Yu, associate professor at Lakehead University whose PhD is on inflation. "The Toronto Stock Exchange publishes every few minutes, and that's basically a price index."

Charts comparing the U.S. CPI and Google's "GPI" show a correlation among items that are frequently bought online, such as cameras, baby toys and watches. Other items bought less frequently on the Web - such as footwear and car parts - differ between the two indexes, Mr. Varian said in his presentation.

Deflation, meanwhile, is a "clear" trend in the United States, Google's price index shows.

The company is exploring other ways Google's search data can be used in economic forecasting. Google dubs its "predicting the present" as "nowcasting," and says real-time data can help improve the accuracy of forecasts.

In a paper last year, Mr. Varian said Google's trends data can help improve forecasts of the current level of activity for several different economic time series, including automobile sales, home sales, retail sales, and travel behaviour. And searches for "unemployment insurance" might help predict unemployment insurance claims, or the unemployment rate.

Mr. Varian joined Google in 2002 and is also an emeritus professor at the University of California, Berkeley.

Statistics Canada declined to comment on Google's move Tuesday, saying it doesn't typically comment on other entity's products.