Non-cointegrated variables and OLS estimates
Posted: Tue Feb 11, 2014 8:23 am
Hi,
I am new to econometrics, I am just trying to estimate the linear regression using OLS method.
I used Quarterly time-series data for 30years (120 data points).
d(Y)=B0+B1*d(X1)+B2*d(X2)+B3*d(X3)+E.
All I have used are I(1) variables as I(0) are non-stationery.
I used Engle–Granger two-step method and found no co-integration. (Not-co integrated)
Can someone explain me on my following doubts
1. Whether I can proceed with estimation of the above regression using OLS?
2. Whether I can I use lag-distribution models (based on OLS)?
3. Granger causality test?
One of my friends told me when variables in the regression are non-cointegrated, (means no long run relationship between variables) and no point of go ahead and estimate coefficients of the regression. Is this true?
Thank you in advance.
I am new to econometrics, I am just trying to estimate the linear regression using OLS method.
I used Quarterly time-series data for 30years (120 data points).
d(Y)=B0+B1*d(X1)+B2*d(X2)+B3*d(X3)+E.
All I have used are I(1) variables as I(0) are non-stationery.
I used Engle–Granger two-step method and found no co-integration. (Not-co integrated)
Can someone explain me on my following doubts
1. Whether I can proceed with estimation of the above regression using OLS?
2. Whether I can I use lag-distribution models (based on OLS)?
3. Granger causality test?
One of my friends told me when variables in the regression are non-cointegrated, (means no long run relationship between variables) and no point of go ahead and estimate coefficients of the regression. Is this true?
Thank you in advance.