Hi ALL,
I say Merry Christmas for those who celebrate Christmas.
I have two questions as usual.
1) Can we use random effect estimation when the some of the coefficients of
the between and fixed effects estimators differ (xtreg, be and xtreg, fe)?
2) If hausman specification test rejected the fixed effects model, does this
mean that all regressors in the model are not correlated with unobserved
effects? Or is there a possibility for one or two regressor to be correlated
even if hausman test favour random effect.
Thanks
menale
_________________________________________________________________
Last ned MSN Messenger gratis http://www.msn.no/computing/messenger - Den
raskeste veien mellom deg og dine venner