Google search activity and labor reallocation

Abstract

This study looked into the Reallocation Hypothesis — that is, the notion that shocks that have asymmetrical effects on sectors of the economy, known as reallocation shocks, induce unfavorable unemployment conditions — using web search data as an unconventional indicator of unemployment conditions. In a panel data regression framework for the US during 2003-16, I specify FE regressions where search terms replace the unemployment rate as an indicator of unemployment on the LHS and acts as the explanatory variable. I find that selected terms showcase a positive and statistically significant correlation with the dispersion index even after the addition of control variables. Such a finding weakens any criticism that attributes the tenability of the RH to the selection of data that measure unemployment. Furthermore, this study adds to a burgeoning literature that advocates that non-traditional economic data — such as those utilized here — may contain useful information for economic research.