EViews.com

Hi All,

I am a bit confused on when to use VAR or VECM. Is this decision dependent on the unit root test? Let say I have dependent variable Y and independent variable X where each is stationary at I(1), which model should I use?

And when is the right time to use Toda Yamamoto?

Thanks in advance.

Regards,
Confused.

Hi,

You should use VECM if 1) your variables are nonstationary and 2) you find a common trend between the variables (cointegration). If your variables are nonstationary and you transform them to became stationary (diferences or subtracting trends) without doing a cointegration test you might be omitting a relevant variable, the cointegration relationship. So, if you don´t find cointegration in your test you can estimate a VAR with the stationary variables.

Best regards.

Hello,

Now I get. thanks much!