2

Apart from the Granger test for Granger causality, which tests whether one time series contains information to help predict the other time series, are there any methods to test if on time series lags/leads another most of the time?

kjetil b halvorsen
  • 77,844
cs0815
  • 2,225
  • 1
    One graphical approach is to pre-whiten both series then examine the cross-correlation function; one can test these cross-correlations, but one must keep in mind that with long series and many lags you're very likely to pick something up even if essentially nothing of any importance is going on. If attention is restricted to a modest number of lags the multiple testing issue is less of a problem but the long-series problem of significance of very small effects would remain. – Glen_b Oct 15 '17 at 20:17
  • 2
    Look for coherence – Aksakal Oct 15 '17 at 20:22

0 Answers0