wrong estimate for constant

[strypste]

Hello,

I would like to estimate an AR(1) model y_t = constant + phi*y_t-1 + e where the constant and phi are equal to 5 and 0.4, respectively (I simulated the data a 100 times).

When I estimate this models with the command equation ar.ls y c ar(1), then I get the correct estimate for phi but the estimate for c is wrong.. I get something close to 8,33, which is equal to the long run average of y..

When I estimate the model with the command equation ar.ls y c y(-1), then phi and the constant are correctly estimated.

Why is that?

Regards
Steven

[EViews Gareth]

http://forums.eviews.com/viewtopic.php?f=7&t=465