Regardless of the software used, it seems that your variables do not contain unit root, so they are probably not cointegrated.
The question whether the variables are I(0) or I(1) can be tested and it is better if you do (if you expect that they are cointegrated) because relying just on one test might not be advisable. Testing by a variety of unit root tests is better.
But, maybe you can also see ARDL approach proposed by Pesaran, which does not necessarily require unit root to be identified.
If both variables are I(0), you do not need cointegration, you can use OLS. Contegration is useful in cases when both variables are I(1), and you therefore cannot use OLS.
If the variables are stationary at level difference, it indicates existence of long -run relationship between the variables. Hence it is not necessary to use Johansen Test.
Johansen test is only used when variables have unit roots. OLS stay a good method in your case
You can use OLS or VAR if all variables are I(0). If you have a mix of I(0) and I(1) variables then use Peseran et. al (2001) ARDL Bounds testing. If you are certain all variables are I(0) write it in ARDL- error correction form and test the level relationship.
If two macroeconomic series are I(0) it means stationary at level. Here simple static relationship that is Yt=a+Bxt+et using OLS is best keeping in view the nature of series. the estimated coefficient of B can be interpreted as long run coefficient. Second part, the long run relationship is verified by estimating simple static model and then checking weather the residuals of the model are I(0). if they are the we conclude that there exist long run relationship between variables.
- Badges
- Science topic
- Similar topics
- Engineering
- Modeling