serial correlation

[kobiw]

I am estimating an equation with time series panel data. I have significant explanatory variables but a very low durbin watson. I added AR(1), which came out very significant, the durbin watson came up to around 2, but all of the explanatory variables became very insignificant (very low t). I added many lags to all of them and only then they became significant. Is this ok? do the explanatory variables need to be significant after you add teh AR(1)?
Thank you.

[jps25]

yes they do...

[kobiw]

Thank you.

Is it common that once you add the ar(1) you loose the significance of the variables and only after you add some lags you manage to make them significant?