In the dynamic factor model, a single unobserved factor common to some macroeconomic variables is defined as a composite index to measure business cycles. This model has recently been developed by combining with the regime switching model so that the mean growth of the index may shift depending on whether the economy is in a boom regime or in a recession regime. An advantage of this dynamic Markov switching factor model is that estimating the model by a Bayesian method produces the posterior probabilities that the economy is in the recession regime, which can be used to date the business cycle turning points. This article estimates the dynamic Markov switching factor model using some macroeconomic variables in Japan. The model comparison using the Bayes factor does not provide strong evidence that the mean growth of the index shifts, but the dynamic Markov switching factor model is found to produce the estimates of turning points close to the reference dates of the Economic and Social Research Institute in the Cabinet Office, unless only weakly correlated variables are used.
Keywords: Business cycles; Factor model; Gibbs sampling; Marginal likelihood; Markov switching; Particle filter
Views expressed in the paper are those of the authors and do not necessarily reflect those of the Bank of Japan or Institute for Monetary and Economic Studies.