An investigation on mixed housing-cycle structures and asymmetric tail dependences

Abstract

This paper designs a Markov-switching mixture Copula-based model with a mixed Markov-transition mechanism to investigate the mixed housing-cycle structures and asymmetric tail dependences for the Pacific and Mountain divisions of American regional housing markets. The empirical results demonstrate four interesting findings. First, the Markov-switching process can capture the housing cycle of each housing market. Second, the evidence of the mixed Markov-switching specification indicates that the joint housing-cycle behaviors are related not only to the total dependence mechanism, but also to the independent mechanism. Specifically, each housing market has its own characteristics, and these characteristics play relatively more important roles in determining the joint housing-cycle pattern than do common factors related to the total dependence framework. Third, the two housing markets have asymmetric tail dependences. Tail dependence exists when two markets experience the same housing-cycle modes, but does not occur when two markets experience distinct housing-cycle modes. In addition, the intensity of tail dependence is stronger when two markets remain in recession mode, as opposed to when they remain in recovery mode. This finding suggests that downward price rigidity does not exist in regional housing markets. Fourth, the spillover effects between housing returns are asymmetric.