Cointegration Estimation with Stationary Components
Posted: Wed Jan 20, 2010 12:00 pm
I have a model Yt = b1X1 +b2X2 + b3t + Dt + et
both Yt and X1 are I(1) variables but X2 is I(0). t is an included time trend and Dt is a dummy variable to indicate a specific event (0 up until the event, 1 afterward)
Running ADF tests on the residuals from the above equation show that the residuals are in fact stationary, which would suggest that Yt and X1 are cointegrated and results arent spurious
Is it possible to use the Engle-Granger 2 step method here even though not all of the variable are I(1)? If so, how should this be set up in Eviews?
if not, what would be the best strategy to estimate this relationship
both Yt and X1 are I(1) variables but X2 is I(0). t is an included time trend and Dt is a dummy variable to indicate a specific event (0 up until the event, 1 afterward)
Running ADF tests on the residuals from the above equation show that the residuals are in fact stationary, which would suggest that Yt and X1 are cointegrated and results arent spurious
Is it possible to use the Engle-Granger 2 step method here even though not all of the variable are I(1)? If so, how should this be set up in Eviews?
if not, what would be the best strategy to estimate this relationship