Page 1 of 1

IV

Posted: Tue Jun 12, 2018 7:52 am
by Luca Tiberti
Thanks for your presentation! I have a few comments I would like to share with you and your team:

Regarding your instruments:
- which is the source of your regional unemployment rate? it is always better not to use the same source as the main one (used for your analysis)
- at which geographical level is your unemployment rate variable defined? in principle, for identification, the smaller the better
- Lewbel IV: for personal experience, reviewers might be skeptical about this relatively new method. To convince them, it is really crucial to explain very well the error terms and what drives the error covariance across equations, what explains the heteroskedasticity of the first-stage error. Also, I should you should also refer to Lewbel 2018 (the binary case).

Also, do you discuss/control for possible labour demand-side constraints. I had the impression that the discussion and the analysis focussed uniquely on labour supply contraints.

Thanks