I am comparing a number of pairs of time series, let us call them A and B, with yearly data stretching over about 60 to 80 years each, to test whether an increase of A causes a decrease of B, and a decrease of A causes an increase of B, as has been claimed by other researchers. Visual interpretation of time series graphs clearly suggests that this is not the case, but I want to make a statistical analysis to strengthen my argument. The results of a cross-correlation analysis are a bit difficult to interpret because each time series is also highly autocorrelated. How can I go about to do a cross-correlation analysis that also takes into account the auto-correlation?