Page 1 of 1

Intercept terms for Random Effects model

Posted: Thu Sep 12, 2013 2:09 am
by Dave_the_Forecaster
I've estimated several equations using panel regression with random effects.

Each of these equations contain some "cross-section specific" variables (there's probably a more technical name for these!), i.e. variables which vary across the cross-sections as well as varying through time.

When I click on View->Fixed/Random Effects->Cross-section Effects, I notice that some equations have all zero terms, and some equations have non-zero terms here.

My interpretation of this is that if these cross-section effects are zero, then the equation is able to describe all of the cross-sectional variance using the "cross-section specific" variables. On the other hand, if the effects are non-zero, then that means that my cross-section specific variables are not fully describing the variance between cross-sections.

Is this interpretation sensible or have I mis-understood something?


Cheers,
Dave

Re: Intercept terms for Random Effects model

Posted: Wed Oct 02, 2013 8:45 am
by Dave_the_Forecaster
Anyone got any thoughts on this one?