Dear all,
I have a series which is clearly trending. It looks to be stationary around a linear deterministic trend and shocks does seem only to affect the series temporarily.
So my guess is that this is a trend stationary series - i.e stationary about a deterministic trend. However, it could have a stochastic trend (unit root) as well.
I am asked to test the hypothesis of the series beeing non-stationary against the alternative hypothesis:
1) the series being stationary
2) the series being stationary around a trend
I am also asked to coment on the results.
My question is how to formally test for this.
Could I run an ADF test on the series specified first including a trend term and then without a trend term and conclude from that?
And how do you normally decide how to set up the ADF regression? I mean no. of lags, deterministic terms and so on.
Thank you very much for your help on this matter. I feel stuck reading for an exam.
Chris