Duration Dependence in Korean Business Cycles

Abstract

The hypothesis of business cycle duration dependence is tested by estimating the Hamilton regime-switching model with duration dependence using the Gibbs sampler. Data are two versions of the index of coincident indicators; (linear) dynamic factor index of Stock and Watson (1989) and (nonlinear) dynamic Markov switching factor index of Diebold and Rudebusch (1994). When the Gibbs sampler is applied to the duration dependent regime-switching model using quarterly Korean business cycle indices for the 1977:1 1994:4 period, this paper finds that the probability of a transition into an recession increases as the expansion ages, and somewhat weaker evidence for the reverse. Example of out-of-sample forecast for business cycle turning points is also provided.