Autocorrelation in a system of equations

[PoachedWonk]

Hello,

I am writing a system of three equations and have a major issue with autocorrelation (a DW of 0.2 for one equation) when I run the SUR regression

To resolve the problem, I decided to do a VECM, but the third equation has a log lhs but an exponential in the rhs, so differencing the exponential would be tricky for this regression). So that went out the window.

I was wondering instead, if I can use a 3sls or GMM approach, using the lagged variables as instruments to deal with the autocorrelation issue. Im not sure it makes a lot of sense, but was wondering if anyone had come across this or other ways to resolve the issue of ACs.

I am using Eviews 8.

Any help would be greatly appreciated!

[PoachedWonk]

Hi guys,

I managed to fix the problem by including AR(1) terms into the model. you can do this by doing something along the lines of the following:

cs = c(1) + c(2)*gdp + [ar(1)=c(3), ar(2)=c(4)]

Cheers