Clearing the Fog: The Predictive Power of Weather for Employment Reports and Their Asset Price Responses

Abstract

This paper exploits vast granular data—with over one million county-month observations—to estimate a dynamic panel data model of weather's local employment effects. The fitted county model is then aggregated and used to generate in-sample and rolling out-of-sample (nowcast) estimates of the weather effect on national monthly employment. These nowcasts, which use only employment and weather data available prior to a given employment report, are significantly predictive not only of the surprise component of employment reports but also of stock and bond market returns on the days of employment reports.