Masaru CHIBA
This paper proposes a multivariate bandpass filter based on the trend, cycle, and seasonality decomposition model. Cycle shifts for individual time series are incorporated as part of the multivariate model. The inclusion of leading, coincident, and lagging variables for the measurement of the business cycle is therefore possible without prior analysis of lead-lag relationships between economic variables. In addition, we explore a stacking approach that is specified by means of a low-frequency time index. In this approach, no artificial missing values are needed and also no information is lost about the high-frequency series and their dynamic features. Our approach is illustrated in detail for business cycle tracking of Thailand.
Business Cycle; Bandpass Filter; State-Space Model;