0

I have a model consisting 4 variables, two of which are stationary at level and two that are I1 stationary. There seems to be a cointegration relationship between the two I1 variables after running the Johansen test.

Do I use a VAR model but taking the first differences of these I1 variables and ignore cointegration, or I proceed with a VECM model? If is the latter how do I proceed?

Karolis Koncevičius
  • 4,282
  • 7
  • 30
  • 47
Constantinos Rousos
  • 61
  • 2
  • https://stats.stackexchange.com/questions/148994/var-or-vecm-for-a-mix-of-stationary-and-nonstationary-variables Does this not answer your question? – Matt P Apr 18 '19 at 17:59
  • @MattP do you know how can i construct such a model in R? Do I need to create the VECM with the two cointegrated variables and then use the error correction term somehow in the VAR model containing all 4 parameters? How do I do this in practice? – Constantinos Rousos Apr 18 '19 at 18:09
  • I would appreciate some help on these in terms of R code, if anybody can help me it would be great!! Thanks in advance – Constantinos Rousos Apr 19 '19 at 06:44

0 Answers0