Multivariate detrending under common trend restrictions: Implications for business cycle research

Abstract

This paper outlines a methodology to detrend multiple time series under common trend restrictions. The same filters used to construct the estimated trend in univariate exercises are shown to be appropriate in multivariate studies with a single common trend. However, to estimate the common trend in the multivariate case, the filter is applied to a linear combination of series rather than to each series individually. An empirical example and simulation exercises illustrate the implications of common trend detrending for measurement of business cycle properties.