Page 1 of 1

censoring and IV

Posted: Tue Jun 12, 2018 5:54 am
by Luca Tiberti
thanks for the interesting presentation. I have a few comments I would like to share with you:

- equation (3) is censored (I guess that there is a large density at zero) so OLS would be biased. I suggest you use a Tobit model.
- I am not convinced about the validity of your IVs: (1) you need to use historical series of migration (and remittances) (see paper by McKenzie and Rapoport 2007); (2) you should use other sources (census?), especially because I guess that in your survey you have not so many observations per villages.


Re: censoring and IV

Posted: Tue Jun 12, 2018 9:42 am
by chanhang
Hello Luca,
Thank you very much for your comments. I will take them all.