Predicting Initial Claims for Unemployment Benefits
July 22nd, 2009 | Published in Google Research
One of the strongest leading indicators of economic activity is the number of people who file for unemployment benefits. Macroeconomists Robert Gordon and James Hamilton have recently examined the historical evidence. According to Hamilton's summary: "...in each of the last six recessions, the recovery began within 8 weeks of the peak in new unemployment claims."
In an earlier blog post, we suggested that Google Trends/Search Insights data could be useful in short term predictions of economic variables. Given the importance of initial claims as a macroeconomic predictor, we thought it would be useful to try to forecast this economic metric. The initial claims data is available from the Department of Labor, while the Google Trends data for relevant categories is available here.
We applied the methodology outlined in our earlier paper, building a model to forecast initial claims using the past values of the time series, and then added the Google Trends variables to see how much they improved the forecast. We found a 15.74% reduction in mean absolute error for one-week ahead out, of sample forecasts. Most economists would consider this to be a significant boost. Details of our analysis may be found in this paper.
The bottom line is that initial claims have been generally declining from their peak and that, so far at least, the Google query data is forecasting further short term declines. It would be good news indeed if this particular Google trend continues.