Markov stochastic switching regime models indicate that business cycle
phase transitions appear to exhibit uneven switching propensities thr
oughout the calendar year. Instead of studying regime switching data,
we investigate business cycle durations obtained from turning point ch
ronologies to investigate this question. We test for the presence of s
easonal features in duration data via nonparametric rank-based tests.
This approach has some advantages over the parametric Markov stochasti
c switching regime models, since it only involves mild regularity cond
itions regarding the hazard model for business cycle turning points. W
e examine two alternative chronologies (NBER and Romer) to address the
question of uneven propensity of switching.