CMStatistics 2017: Start Registration
View Submission - CFE
Title: On the Model Uncertainty of Infinite Hidden Markov Models with Application to Taylor Rule Characterization Authors:  Zhuo Li - University of Melbourne (Australia) [presenting]
Yong Song - University of Melbourne (Australia)
Qiao Yang - ShanghaiTech University (China)
Abstract: This paper investigates model uncertainty of infinite hidden Markov models (IHMM)through stochastic search variable selection (SSVS). We apply the normal-gamma shrinkage prior of Grin and Brown (2010) for variable selection and develop new specification search method guided by economic theory. We compare our approach to Bayesian model averaging (BMA) and static/dynamic pooling methods in an application to investigating the time-varying representation of the Taylor rule by using U.S.macroeconomic vintage data.