Working Paper: NBER ID: w20010
Authors: Dolan Antenucci; Michael Cafarella; Margaret C. Levenstein; Christopher R. Levenstein; Matthew D. Shapiro
Abstract: Social media enable promising new approaches to measuring economic activity and analyzing economic behavior at high frequency and in real time using information independent from standard survey and administrative sources. This paper uses data from Twitter to create indexes of job loss, job search, and job posting. Signals are derived by counting job-related phrases in Tweets such as "lost my job." The social media indexes are constructed from the principal components of these signals. The University of Michigan Social Media Job Loss Index tracks initial claims for unemployment insurance at medium and high frequencies and predicts 15 to 20 percent of the variance of the prediction error of the consensus forecast for initial claims. The social media indexes provide real-time indicators of events such as Hurricane Sandy and the 2013 government shutdown. Comparing the job loss index with the search and posting indexes indicates that the Beveridge Curve has been shifting inward since 2011. \n\nThe University of Michigan Social Media Job Loss index is update weekly and is available at\nhttp://econprediction.eecs.umich.edu/.
Keywords: Social Media; Labor Market; Job Loss; Job Search; Job Posting
JEL Codes: C81; C82; E24; J60
Edges that are evidenced by causal inference methods are in orange, and the rest are in light blue.
| Cause | Effect |
|---|---|
| hurricane Sandy (H84) | social media job loss index (J63) |
| 2013 government shutdown (H12) | social media job loss index (J63) |
| social media job loss index (J63) | job loss signals (J63) |
| social media job loss index (J63) | initial claims for unemployment insurance (UI) (J65) |
| social media job loss index (J63) | prediction error of the consensus forecast for initial claims (C53) |