Handling AutoCorrelation in Fixed Effect Model

[syed.hassanjamil]

Hi,
Is there any available reference regarding use of AR(1) term or its restriction in Fixed Effect model? In my recent study I used AR(1) term in FE model to control Durbin Watson value for autocorrelation problem. By inclusion of AR(1) term the D.W value has been improved from 0.96 to 1.97, but I am now questioned that if it is valid inclusion and would not bias the estimate?

Any idea, help please.