xtivprobit

[hunter shi]

Hi folks,

Just wanna ask if anyone knows the command regarding panel fixed effect iv logit estimation? Because I want to use panel data and consider the fixed effect, I couldn't use pooled iv logit.

Cheers,
Hunter.

[Joro Kolev]

Can you give a reference, in which paper/textbook you have seen an IV panel data logit?

[hunter shi]

Can you give a reference, in which paper/textbook you have seen an IV panel data logit?

Hi Joro,

I've never seen this before. So I wonder if IV panel data logit command exists in Stata or not. I am asking this because I am working on panel data (IFLS) and the LHS variable is a dummy (0 or 1). And I need to use IV to deal with the endogenous independent variable. So I have 1.panel data, 2. IV, 3. binary LHS variable.

Cheers,
Hunter.

[Joro Kolev]

This does not exist in Stata to my knowledge.

I do not know what is IFLS.

How many IDs you have in your panel, and how many time periods?

[hunter shi]

This does not exist in Stata to my knowledge.

I do not know what is IFLS.

How many IDs you have in your panel, and how many time periods?

Hi Joro,

I have around 10,000 observations and 4 waves. But the panel is not balanced. Is there a way to do xtivlogit? I know there is xtlogit and ivlogit though...

[Joro Kolev]

I do not think that there is ivlogit (I do not see any reasons as to why not, but I think such an estimator does not exist in Stata), there is -ivprobit-.

Most probably you would have to do manually control function, and use -xtlogit-. I have not seen a paper or a textbook doing or justifying this, but mechanically can be done.

[Joerg Luedicke (StataCorp)]

Stata's xteprobit fits random-effects IV probit models:

https://www.stata.com/manuals/ermeprobit.pdf

[George Ford]

sorry. you're right.

xteprobit fits RE model.

Also see

https://www.stata.com/meeting/germany14/abstracts/materials/de14_drukker_gsem.pdf