Yes, me too. It is quite easy to run this test in Eviews, just follow the manual.
Thank you for all your reply. Basically,I do know how to use Eviews , but the problem is that when I run Johansen model, I always get a problem with the sign of the coefficients. could you help with that?
Hello!
With Eviews if you do the Johansen, during interpretation you'll have to consider the opposite sign. So, if you get a (-)ve sign in Eviews, you'll have to interpret the result as (+)ve sign. If that doesn't solve your problem, then I will give you the detail description. But hope this much will help you to start your analysis.
when we use Johansen approach in E-Views, we do not need to specify dependent variable. first you have to check the stationarity of all the variables. if all variables are integrated of order 1, then we can apply Johansen approach. for this we have to construct VAR by using appropriate lag length. here we write all variables without distinction of independent and dependent variables. afterwards apply cointegration test, this gives cointegrating vectors and error correcting coefficients.
Thank you Dear Rabbi for your explanation which is really important.
Actually I watched several educational video to learn how to apply Johansen approach, my concern is, after applying Johansen approach then I do apply VECM to look for the short run relationship, where shall I start to interpret the result.?
shall I rely on VECM for interpreting the long run and short run relationship among the variables? or I do rely on the johansen for the long run relationship and on the VECM for the short run relationship? During interpretation does VECM also taking the opposite sign too ?
Thank you in advance
Dear Tehreem, thank you for your answer.
Please could I have your email to get more help about how to we have to construct VAR by using appropriate lag length? may I need to know how to apply VECM step by step, and needing help to interpret the result for VECM and how to get cointegrating vectors and error correcting coefficients.
regards
To choose an appropriate lag length, you first have to construct VAR, estimate by taking following lag lengths: 4 for quarterly data, 12 for monthly data. now when the VAR is constructed, here you will go to view, lag length criteria and using AIC or SIC, you can choose a lag length. AIC and SIC because they are more popular ones. now using the lag length specified by these criteria again estimate VAR. N ow you have obtained VAR results. Now go to view of VAR and choose cointegration test. finally you will get the results of Johansen along with cointegrating equations, error correcting coefficients and cointegrating coefficients. you will have to sumproduct error correcting coefficients and cointegrating coefficients. you will get a single value. if its sign is negative, it means the shock that took place in previous period will be corrected in current period.
If you still have any ambiguity, I would like to refer you book of Auteriou and Hall. hope your work gets completed successfully. here is my email address
Dear Tehreem,
Thank you very much for your help and providing your email address, I really do appreciated.
I would like to email you and give some explanation on my work, lets see what do you think about it.
many thanks
Make sure the variables are relating with one another before considering running the co-integration tests otherwise you will get a misleading results. E.g if you are considering two markets ( price series) they must to be trading among themselves and separated by less than a critical distance. First test for stationarity using the augmented dickey-fuller procedure . The test will give you the order at which the variables are stationary. Run the Con-integration test .
Yes You can use Eviews for Co-integration. It is a long-run relationship, so it should be run in levels of variables. You can find out the number of co-integrating vectors and then use them to generate error correction term.
- Badges
- Science topic