Best form of time series regression

noetsi

Fortran must die
#1
If you want to know if an X influences a Y at some point in the future, or several points, which is the best type of time series regression to run. I have looked at many none of which seem to work. I want to focus on the best one.

All I really want to know is 1) which predictors have the most impact 2) how they influence the forecast (that is improving estimates of spending if you know a related predictor).

not much. :) Our data is commonly, seasonal and non-stationary.