A precise understanding of lead-lag structures in economic data is important for many economic agents such as policymakers, traders in financial markets, and producers in goods markets. To identify time varying lead-lag relationships across various frequencies in economic time series, recent studies have used phase difference on the basis of a continuous wavelet transform. However, the extant literature includes several conflicting interpretations of phase difference. In this study, we extensively discuss wavelet phase difference, determine its most plausible interpretation, and thus attempt to address gaps in the existing literature. Consequently, this study suggests that some lead-lag results of previous works have been driven by incorrect interpretations of wavelet phase difference. (C) 2016 Elsevier Ltd. All rights reserved.
