Omit Consant

[ancraeg52]

Hi,

I have been running a standard OLS regression and all my variables are statistically significant. It has passed all necessary diagnostic tests. However, my constant is not statistically significant and I was wondering does this compromise the results of this model. Should I omit it from the regression all together?

Any advice would be greatly appreciated.
J.

[startz]

Leave it in. The constant rarely has a substantive interpretation anyhow.

[ancraeg52]

Hi Startz,

Thanks for your reply and apologies, I probably should have mentioned that all variables (both LHS and RHS) are first differenced - Would this be a reason to omit the constant?

Thanks,
J

[startz]

That might make it more likely that the constant is really zero, but I'd leave it in anyway.

[ancraeg52]

Thanks for your help Startz,
Regards,