I'm working on temporal disaggregation of economic time series (expecting to use the R package tempdisagg) and would like to use indicator series, but I have not been able to find out how to identify suitable indicator series. I have a list of possible candidates for indicator series, but I do not know how to narrow it down or choose one over the other.
Do you just choose indicator series based on the (theoretically justified) assumption that they might adequately describe higher frequency movement of the lower frequency time series?
Do you aggregate the higher frequency time series to the same frequency as the lower frequency one and run tests for correlation/synchronicity?
Do you model the disaggregation with a number of different candidate indicator series and use some measure of quality or descriptive power to choose the best models? Is there a way to compare different methods of temporal disaggregation, which might all come with their own specific, non-comparable quality measures?
Can anybody help me identify the best indicator series?