Instrument variable regression- when endogenous variable is categorical and dependent variable is a count data

[Kirti Gaur]

Dear Statalisters,

Thanks for adding me to the forum to discuss my issues. I would like to inform you that I wanted to instrument variable regression as I am facing the problem of endogeneity in the model. So my two queries are:
1. As my dependent variable is a count so can I Poisson regression in second stage?
2. My endogenous variable is a categorical in nature categorized as "0", "1" and "2". Please let me know whether endogenous variable can be categorical or not. So that when I use the predicted value of the endogenous variable in the second stage I get the co-efficient against each category of endogenous variable.

It would be great if you could guide me through this and suggest me some articles in which the authors have done a similar thing.

Many thanks in anticipation.

Waiting to hear from you soon.

[Kirti Gaur]

Dear Statalister,

Could you please help me out of this as soon as possible. I am waiting for your reply and hope to get a positive response.

Many regards

Kirti

[Manish Srivastava]

Kirti, You may want to search Statalist archives first. But here is one paper by Miranda and colleagues that discusses this. http://www.stata-journal.com/sjpdf.h...iclenum=st0107 Manish

[Kirti Gaur]

Hey Manish,

Thanks for the information but I am not able to download the article can you please send me the pdf version of the same.

Many thanks and regards

Kirti

[Dimitriy V. Masterov]

Also, -cmp- might be worth a look if you're willing to treat your outcome as continuous.

[Manish Srivastava]

Links works fine at my end, but here is a pdf copy of the article.

-Manish

[Kirti Gaur]

Thanks Manish

[Ruchi Vangani]

Dear Kirti,

Could you please share how you managed the endogenous regression of categorical variable using poisson regression.
The stata journal example is for dummy variable, what I have is a categorical variable with categories so am a bit confused so as to how to apply the ssm.

Please help.

Kind Regards,
Ruchi