Forecasting national recessions using state-level data

A large literature studies the information contained in national-level economic indicators, such as nancial and aggregate economic activity variables, for forecasting U.S. business cycle phases (expansions and recessions.) In this paper, we investigate whether there is additional information regarding business cycle phases contained in subnational measures of economic activity. Using a probit model to predict the NBER expansion and recession classification, we assess the forecasting benets of adding state-level employment growth to a common list of national-level predictors. As state-level data adds a large number of variables to the model, we employ a Bayesian model averaging procedure to construct forecasts. Based on a variety of forecast evaluation metrics, we find that including state-level employment growth substantially improves short-horizon forecasts of the business cycle phase. The gains in forecast accuracy are concentrated during months of national recession. Posterior inclusion probabilities indicate substantial uncertainty regarding which states belong in the model, highlighting the importance of the Bayesian model averaging approach.