The duration of business cycle expansions and contractions: are there change-points in duration dependence?

Abstract

The issue of whether the likelihood of an expansion or contraction ending is dependent on its age, i.e whether they are duration dependent, is widely addressed in the business cycles literature and evidence of positive duration dependence is found in several studies. However, there is an important issue that has not been explored in this literature yet: the presence of change-points in duration dependence. All the studies in this field depart from the assumption that the magnitude of duration dependence is the same over time. However, we conjecture that the degree of likeliness of an expansion or contraction ending as it gets older might change after a specific duration. To test for that possibility, this paper will allow for the presence of a change-point in the analysis of the duration of expansions and contractions for a group of 13 European and Non-European industrial countries over the period 1948-2009. The evidence provided by the estimation of a continuous-time Weibull duration model shows strong support for the presence of positive duration dependence, which is stronger for contractions than for expansions. Results also show that contractions have become longer over time and that their length is negatively affected by the length of the previous expansion. Most importantly, this paper provides quite interesting evidence for the presence of a change-point in duration dependence for expansions, but not for contractions. Results show that the magnitude of the duration dependence parameter decreases significantly when an expansion surpasses 10 years of duration. In particular, evidence of positive duration dependence is no longer found when an expansion surpasses that threshold.