I'm testing for causality between two series. PACF and ACF select lags that, when used, do not yield significance in the causality tests. Playing around with the lag I find that I get causality significance for other lags...What are the problems associated with NOT using the lag selected by ACF/PACF? For the record, the range between ACF/PACF values was -3.07 to -3.05.
Asked
Active
Viewed 158 times
