ON THE PERIODIC STRUCTURE OF THE BUSINESS-CYCLE

This article tests whether an economic recovery is equally likely to o ccur in any month of the year. I test this hypothesis with a homogeneo us Markov switching-regime model and use the turning-point dates for b usiness cycles marked by the National Bureau of Economic Research and Romer. I tend to find unequal probabilities to switch from recession t o expansions. In particular, the spring and the month of December appe ar to have higher incidence of economic recovery. My results also impl y that recessions and expansions are, on average, longer or shorter, d epending on what time of the year they start. This suggests the presen ce of a seasonal pattern in business-cycle durations, though such a no tion of seasonality is quite different from the common one based on un observed component linear time series models. I investigate issues tha t go beyond linear dependence between seasonality and business cycles.