IV for Negative Binomial Regression

[rashel hasan]

Dear users,

I need your help understanding the following coding suggested by Jeff Wooldridge in his post in 2014.

In this post while explaining the technique using stata code:

1. At first stage regression, z1.... zM are instruments?
2. Why we are taking residual, why not predicted y2?
3. At the second stage regression, should it be y1 instead of y2?
4. why we are aging taking y2 and all z including residual at second stage regression?

I would appreciate your response.
Thanks
Rashel

https://www.statalist.org/forums/for...ative-binomial

[Devra Golbe]

Rashel,

Jeff is suggesting a control function approach. He's written an overiew to the method here: Wooldridge, JM 2015, 'Control Function Methods in Applied Econometrics', Journal of Human Resources, vol. 50, no. 2, pp. 420-445.

Devra

[rashel hasan]

Thanks a lot Golbe for your response. I will read the article and try to understand the approach.