Arthur Lewbel - ivreg2h

[Hovhannes Nahapetyan]

Hi all,

I hope I can get some help on this issue. Arthur Lewbel has two articles on internally generated instruments in cases where outside instruments are not available: Constructing Instruments for Regressions With Measurement Error When no Additional Data are Available, with An Application to Patents and R&D (1997) and Using heteroscedasticity to identify and estimate mismeasured and endogenous regressor models (2012). If I understand correctly ivreg2h is meant estimate models based on the 2012 peper. I was wondering:

1: is there is Stata command to run regression based on the 1997 paper.
2: is ivreg2h only for cross-sectional regression
3: what is the main difference between the two papers and their estimation methods.

Thank you in advance for your reply,

[Phil Bromiley]

You can search Statalist for Lewbel or google Lewbel Stata to see if anyone has written such a procedure. I don't know of such a a procedure but there are a great many procedures I don't know. In many cases you can move from cross-section to panel fixed effects by just including i.panel among the regressors (assuming you number of panels doesn't result in the procedure exceeding the maximum matrix size). I have no idea if this works for Lewbel's estimators.

As for the difference between the two papers, you should be able to figure that yourself. If you can't, then your methodological skills are such that I would suggest you not mess with somewhat esoteric estimators.