Urgent help needed! MLE and ar(1)

[lenasong7]

Hi!
I have a question on time series regression.

P = a + b1X1+b2X2+b3D+b4D*X1+Monthly dummies

P, X1, X2 are all I(1). D is time period dummy.
The OLS of the above eqn seems spurious (with low DW statistic).
But its residuals follow I(0) so I believe there should exist some cointegration relationship. Am I right?

The error terms are, however, autocorrelated disturbances, i.e. AR(1).
So in this case, I have read from some textbook that I can reestimate the equation including AR(1) by MLE.
I think for MLE I can keep the OLSE but should produce more robust standard errors. Is this correct?
If correct, how can I obtain these new errors in eviews?
I tried the following code I got from some website, but it does not work at all with a warning sign that says: 'missing value in @LogLi series ...''.

@logl logLi
@temp e
e=p-c(1)-c(2)*d01-c(3)*m-c(4)*d01*m-c(5)*g-c(6)*@seas(2)-c(7)*@seas(3)-c(8)*@seas(4)-c(9)*@seas(5)-c(10)*@seas(6)-c(11)*@seas(7)-c(12)*@seas(8)-c(13)*@seas(9)-c(14)*@seas(10)-c(15)*@seas(11)-c(16)*@seas(12)
logLi = log(1/c(17)*@dnorm(e/c(17)))
@param c(17) 1

Please kindly help me. Thank you in advance.