Page 1 of 1
ECM
Posted: Mon Dec 16, 2013 5:48 am
by kafff
Hi everybody
want to create spot and futures returns ecm.
I know that in ordinary case it will be like d(s) d(f) u(-1)
But if I'll take f(-1), so it will be d(s) d(f) d(f(-1)) u(-1)?
or I have to estimate lead-lag model?
Re: ECM
Posted: Mon Dec 16, 2013 9:26 am
by EViews Gareth
I'm afraid I don't follow the question.
Re: ECM
Posted: Mon Dec 16, 2013 9:53 am
by kafff
Re: ECM
by EViews Gareth on Mon Dec 16, 2013 10:26 am
I'm afraid I don't follow the question.
Ok, sorry. Let me explain. I want to create the Error Correction Model in EViews. I have the spot and futures returns. It's already tested for unit roots and had cointegration. So, ECM can be created. But my problem is that i also need to create ecm for spot and lagged futures f(-1). I'm confused which model need I use - the general ECM or lead-lag model for spot and futures returns? And if general ECM - how i guess, so, i need to use the command like d(s) d(f) d(f(-1)) u(-1)? where d(s) - differenced spot and so on?
Re: ECM
Posted: Wed Dec 18, 2013 5:45 am
by eobumneke
Hello, What you need to do is to have one over-parametrized ECM model, from which you have to delete the insignificant variables using general to specific method until you arrive at a parsimonious model.