Cointegration

[tomasza]

Hi all,

I hope someone can help me with a problem in eviews 7. I have two time series, one containing stock data, the other performance data. I'd like to build a VAR model with these two time series, however, both are I(1). Hence I have to test for cointegration. I know there are several options to do this in Eviews. I read somewhere I have to open the two series as a group and then Eviews had a build-in Engle-Granger cointegration test. Is this correct in my case? And do I need to perform a cointegration test on the original nonstationary time series or on the stationary first differences if I use this build-in Engle-Granger test? Or do I need to use Johansen's test for cointegration?

Thanks in advance

[mrrox]

if both variables are I(1) then, open them as a group and then go to "View", from the menu click "Cointegration Test...". Then you will have to specify the test by choosing from 5 options for inclusion of trend, constant or none of them. In the right hand side of the specification window, you will see a default lag structure shown as "1 4", you can chenge the lag length as you wish. But if your data is in quarterly frequency leave it as default. But if your data is monthly, change 4 to 12.

Hope this helps