The Ins and Outs of Labor Market Forecasting

Abstract

Barnichon and Nekarda (2012) have brought attention to forecasting the labor market using the bathtub model. In addition Diebold and Yilmaz (2009) have made substantial advances to the spillover literature by introducing connectedness tables and plots. We blend the two works to answer two interrelated questions: Are measures of steady-state relationships as implied by gross flows data "connected" to their stock-based counterparts? And second, if the measures are related, how can we produce reliable forecasts with gross flows data? We propose a new method to exploit information embedded in labor flows data to forecast the unemployment rate, labor force participation rate and employment-to-population ratio. The forecast evaluation suggests that understanding flow-based methods pays off. In particular, we find forecasting performance to be superior at longer-horizons and conclude that some measures of the labor market become more connected at longer horizons.

Keywords: labor market, forecasting, gross flows, transition rates