A0755
Title: Bayesian estimation of dynamic relationship between GDP and economic indicators for analyzing business cycles
Authors: Koki Kyo - Obihiro University of Agriculture and Veterinary Medicine (Japan) [presenting]
Hideo Noda - Tokyo University of Science (Japan)
Abstract: A traditional approach for analyzing business cycles is that uses a diffusion index, which is constructed based on several selected economic indicators. A problem of this approach is lack of analysis of the lead-lag relation and the difference of importance between the economic indicators. We analyze the lead-lag relation and the difference of importance between the economic indicators by taking GDP as a basis. We extract a stationary component from time series for GDP and each economic indicator using a set of state space models. The extracted stationary components are regarded as a kind of signal for business cycle analysis, then we estimate dynamic relationship between the stationary components in each economic indicator and GDP using Bayesian modeling methods. Thus, the lead-lag relation and the difference of importance between the economic indicators can be analyzed based on the estimates. It can also be aimed at estimating monthly series of GDP by extending the newly proposed approach.
