Nowcasting of the U.S. unemployment rate using Google Trends

Shintaro Nagao, Fumiko Takeda, Riku Tanaka

Research output: Contribution to journalArticlepeer-review

19 Citations (Scopus)

Abstract

This study examines whether and how the search intensity data obtained from Google Trends contributes to nowcasting of the U.S. unemployment rate compared to the conventional AR model. Our assessment is motivated by two issues that may affect the validity of the forecast model using the Google search intensity. The first issue is the change in Google Trends specification that limits the period during which the search intensity data can be retrieved on weekly basis. The second issue is the potential change in the endpoint value of seasonally-adjusted series based on the timing of seasonal adjustment, which may generate a problem when running a real-time forecast. Our results show that the usage of Google Trends doesn't necessarily contribute to improving the accuracy of forecasts under some preconditions, suggesting that there is a limit to the method of adding the search intensity of single keyword to the forecast model.

Original languageEnglish
Pages (from-to)103-109
Number of pages7
JournalFinance Research Letters
Volume30
DOIs
Publication statusPublished - 2019 Sep
Externally publishedYes

Keywords

  • Forecasting
  • Google Trends
  • Nowcasting
  • Unemployment rate

ASJC Scopus subject areas

  • Finance

Fingerprint

Dive into the research topics of 'Nowcasting of the U.S. unemployment rate using Google Trends'. Together they form a unique fingerprint.

Cite this