A latent variable approach to forecasting the unemployment rate

ABSTRACT A forecasting model for unemployment is constructed that exploits the time series properties of unemployment while satisfying the economic relationships specified by Okun's law and the Phillips curve. In deriving the model, we jointly consider the problem of obtaining estimates of the unobserved potential rate of unemployment consistent with Okun's law and the Phillips curve, and associating the potential rate of unemployment with actual unemployment. The empirical example shows that the model clearly outperforms alternative forecasting procedures typically used to forecast unemployment. Copyright © 2011 John Wiley & Sons, Ltd.