Page 1 of 1

Emergency!!! Time series problem

Posted: Tue Aug 11, 2009 2:02 pm
by kqy0101
Hi guys,

I am doing my thesis now, but I encountered some problems.

Some of the variables follow I(1), some follow I(2), according to the ADF unit root test.
Does that means I can not contact Johanson cointegration test? In my point, I take the first difference of those variables following I(2), then all the varibles become I(1). However, the cointegration test shows the coefficents are 1200, How can I explain them( they are log(xt) )?

Another simple question is how can I know that my variables are stationary after cointegrated?

Thanks,

Re: Emergency!!! Time series problem

Posted: Tue Aug 11, 2009 10:25 pm
by kqy0101
Is there anyone who can help me? Appreciate it!

Re: Emergency!!! Time series problem

Posted: Wed Aug 12, 2009 10:27 pm
by random_access
my advice to you is to use other unit root test which are more reliable than ADF......use PP or KKPS to support ur reasoning.
on another note make sure that u have taken the natural logs of all the variables u r using......sometimes using different unit of measurements creates a problem

hope i helped
all the best