I am running the Student Lite version 9:
For an estimated two-way fixed effects model I feel like there is an a inconsistent presentation for the individual and time effects.
(Those effects that can be accessed by View -> Fixed/Random Effects -> Cross-section Effects|PeriodEffects.)
I know those effects are in deviation from the mean. I implemented the calculating myself. I get identical results for the cross-section effects. The period effects are off by a constant and that constant amounts to the difference between the mean of the period effects and the weighted mean of the period effects.
Thus, it seems to me EViews uses the mean of the cross-section effects while for the period effects it uses the weighted mean of the period effects. If so, this feels inconsistent. Please notice, that this only arises in the two-way _unbalanced_ case. Could you please check?
I have not checked, but it could well be the case that the intercept in those models is effected as well?
EDIT:
I should have added: in the one-way cases balanced (unbalanced), it is always the (weighted) mean, i.e. for cross-section as well as period effects.
I would highly appriciate if you could comment on the rational of using the weighted mean. There does not seem to be documentation about the effects at all?
Maybe there is a way in Eviews to get the effects in levels?
cross-sec and time effects in two-way unbalanced FE model
Moderators: EViews Gareth, EViews Moderator
Who is online
Users browsing this forum: No registered users and 13 guests