Co-movement selective detection filter to identify time series co-movement indicator or to filter out symmetric economic shocks

The paper deals with designing a mask suitable for a selective filtering of data. The design of the mask is performed in the time-frequency domain and the selection is based on the co-movement measure of time series. We propose two approaches for the mask construction: i) hard thresholding based on chi(2) testing; ii) adaptive based thresholding. The proposed mask can be used for time series filtering in which we obtain either the adjusted time series or the construction of the time series containing only the co moved parts. Further, after computing an inverse transform we can obtain time series with/without the co-moved area applicable for consequent econometric analyses. The paper provides recommendations concerning the selection of a particular approach in a given situation. The proposed methodology is demonstrated on the adjustment of industrial production index of Euro Area and selected G8 countries about co-movement with the US. (C) 2021 Elsevier Inc. All rights reserved.