Intercept terms for Random Effects model
Posted: Thu Sep 12, 2013 2:09 am
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
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