I am working on a time series data and I want to determine the order of integration for the various variables.
When looking at the graph I observed a potential break point in the some series...so to account for this in my unit root test, I used "breakpoint unit root test" on Eviews.
My Problem: when should I use an innovative model versus additive outlier model?
Would it suffice to look at the graph, pin-point the suspected break date, and see if the series after the break date changes rapidly (then use AO) or change gradually (then use IO). Or is there another way?