研究期間：10208~10307;This proposal contains three sub-projects tries to adopt a new perspective and method, namely the Empirical Mode Decomposition (EMD) proposed by Huang et al. (1998), to reexamine the decomposition, modeling and forecasting of a non-stationary and potentially nonlinear macroeconomic time series. Besides some preliminary interesting results from our previous papers, the proposal shall concentrate on (1) comparing the trend-cycle decomposition of the EMD approach with the HP and BP filters with some remarks on the Beveridge-Nelson and unobservable component decompositions of the log post-war quarterly US real GDP and the issue of dating business turning points. (2) International business cycle synchronization:, we are in particular interested in which frequencies and how are those business cycle across borders being linked since our dynamic IMFs capture all international and intertemporal cross-correlation among the observed real log GDP. Sychronization test of the business cycle phases raised by Harding and Pagan will be examined. (3) Discuss the dynamic sychronization between series both in the time domain as well as in the frequency domain with the idea of sychronization between stationary series brought up by Liu (1990) and Granger and Liu (1994) and then generalize our discussions into the case of nonstationary time series with the framework under the EMD or Ensamble EMD decomposed series.
