Traditional measures of dependence in time series are typically based on correlations or periodograms. These are adequate in many circumstances but, in others, especially when trying to assess market linkages (e.g., financial contagion), might be inappropriate. In the present paper we propose...