Trend component in d(log...) possible?
Posted: Wed Oct 19, 2011 2:34 pm
Hi to all!
variable: Currency in circ.
- I run a Unit root test on Cur_sa, Not sig in level BUT sig in first diff. --> (dlog(cur_sa))
- In order to capture some outliers I introduce 3 dummy variables
- If I now add some trend components (@trend @trend^2) the residual seems to get better.
dlog(cur_sa) c @trend @trend^2 v1 v2 v3
How is this possible? I thought that if I run a model in first diff. I can not apply any trend component?! :?
Kind regards
variable: Currency in circ.
- I run a Unit root test on Cur_sa, Not sig in level BUT sig in first diff. --> (dlog(cur_sa))
- In order to capture some outliers I introduce 3 dummy variables
- If I now add some trend components (@trend @trend^2) the residual seems to get better.
dlog(cur_sa) c @trend @trend^2 v1 v2 v3
How is this possible? I thought that if I run a model in first diff. I can not apply any trend component?! :?
Kind regards
