NBER-NSF Time Series Conference; University of California at Davis, USA

Abstract for "Dating U.S. Business Cycles with Macro Factors" by Sebastian Fossati

The Business Cycle Dating Committee of the NBER determined that a peak in economic activity (beginning of a recession) occurred in the U.S. economy in December 2007. This announcement, however, did not come out until December 1, 2008. Given this delay in NBER announcements (12 months on average), researchers have focused on generating recession indicators that approximate NBER dates. This paper uses Gibbs sampling to estimate a self-exciting threshold autoregressive probit model for dating business cycle turning points. The dependent variable is the state of the economy (recession/expansion) as defined by the NBER. Explanatory variables are a set of static latent common factors estimated by the method of asymptotic principal components from a large data set of 131 monthly macroeconomic time series. Results show that the proposed model generates recession probabilities that almost perfectly reproduce NBER dates. As a consequence, the model shows an excellent performance as a dating algorithm. Furthermore, these results are achieved with a very simple model specification and with no need to allow for structural breaks or time-varying parameters.