Dear all,

I have 2 variables x and y. y is the dependent variable.

I found that they are cointegrated. So I need to define the Error correction model (ECM).

So I run the simples ECM

reg d.y l.resid_1 d.x

But I find that the residuals from this regression are correlated. How can I eliminate serial residual correlation? I know that I have to add lags of d.y and d.x to the ECM, but is there any stata procedure that that allows me to end up with a final augmented ECM?

Best,
John