Estimating cointegrating regression

[az20]

Hello,
I would like to know if when you estimate the cointegration regression using OLS, do you just use the longrun model in levels in OLS and how does the results from OLS compares with the results that you get for your beta estimates when you do the cointegration test using eviews Johansen maximum likelihood method. When I estimate the LR equation using OLS gets a very different results, so was wondering whether I can still use the residuals from OLS to estimate the ECM.
Would you also be able to tell me how to use general –to – specific modelling approach to obtain a parsimonious final model, in Eviews.

Appreciate any help. thanks.

[az20]

Can anyone please provide some advice for my previous post? I am still at a loss.