Hi,
I have an unbalanced panel data (1000 firms, for the period 2002-2009).
I am trying to estimate the probability of survival using probit model.
I have three options:
1)xtprobit, re
xtprobit x_status x1 x2 x3 timedummy i.sector ,re
2)xtprobit,pa
xtprobit x_status x1 x2 x3 timedummy i.sector ,pe
3) fixed effect by adding firm dummies.
probit x_status x1 x2 x3 timedummy i.sector i.firmid
The estimation results are NOT sensitive to the number of quadrature points, therefore I have eliminated first option. 3rd suffers from incidental parameter problem, so this goes as well.
The manual tells that I can use pa estimates.
I can also use pooled probit with only sector dummies (probit x_status x1 x2 x3 timedummy i.sector )
I do not know which one to choose.
I need some guidance.
Thanks
Nazlı
I have an unbalanced panel data (1000 firms, for the period 2002-2009).
I am trying to estimate the probability of survival using probit model.
I have three options:
1)xtprobit, re
xtprobit x_status x1 x2 x3 timedummy i.sector ,re
2)xtprobit,pa
xtprobit x_status x1 x2 x3 timedummy i.sector ,pe
3) fixed effect by adding firm dummies.
probit x_status x1 x2 x3 timedummy i.sector i.firmid
The estimation results are NOT sensitive to the number of quadrature points, therefore I have eliminated first option. 3rd suffers from incidental parameter problem, so this goes as well.
The manual tells that I can use pa estimates.
I can also use pooled probit with only sector dummies (probit x_status x1 x2 x3 timedummy i.sector )
I do not know which one to choose.
I need some guidance.
Thanks
Nazlı
