Title: The predictive power of Google searches in forecasting US unemployment
Authors: Juri Marcucci - Bank of Italy (Italy) [presenting]
Francesco D Amuri - Bank of Italy (Italy)
Abstract: The use of an index of Google job-search intensity is suggested as the best leading indicator to predict the monthly US unemployment rate. We perform a deep out-of-sample forecasting comparison among models thatadopt the Google Index, the more standard initial claims or alternativeindicators based on consumers' and employers' surveys, selecting the best model specification in sample using the BIC. Google-based model soutperform traditional ones, with their relative performance improving with the forecast horizon. Furthermore, quarterly predictions constructed using Google-based models provide more accurate forecasts than the Survey of Professional Forecasters, models based on labor force flows or standard nonlinear models. Google-based models seem to predict particularly well at the turning point taking place at the beginning of the Great Recession, while their relative predictive ability stabilizes afterwards.