Title: Multivariate direct filter analysis for co-integrated processes
Authors: Tucker McElroy - Census Bureau (United States) [presenting]
Marc Wildi - Zurich University (Switzerland)
Abstract: Real-time signal extraction for multivariate time series extracts trends, cycles, and seasonal patterns by utilizing information from related series. Optimal real-time signal extraction filters can be obtained through the multivariate direct filter analysis (MDFA) methodology. Two advances to MDFA are discussed: (i) integration constraints and (ii) co-integration constraints. A new approach is used, whereby integration/co-integration constraints are incorporated into the optimization problem by transforming the real-time filters. The new techniques are demonstrated in construction and employment data.