Random effects

[sana.mazhar]

I want to choose between fixed or random effects.
My cross-section random effects are signficant so i am using them, but i want to know whether the period effects should be modelled as random too.

The redundant fixed effects are significant for all countries (cross sections) but the fixed effects are not significant for periods.

When i run the hausman test on random cross section effects, it fails to reject null and i think that means it favours random cross section effects?

So...

When I run the hausman test on period random effects i get

** Warning: estimated period random effects variance is zero.

What does this mean?

Also does this mean there are no period effects at all (obvs no fixed, but random either) and i should just model cross section random effects?

The thing is I do not know how to explain the warning result in my disseration. I saw somewhere on this forum that the warning means that "there is no difference between the GLS and OLS estimates". But what does that mean?

Thank you so much if you reply. My due date is 9th sept and i really need to write my results up! I just want to know how i can explain this result in my dissertation as my supervisor doesn't know how and I just need a tight arguement for choosing whichever model i use to estimate.