Weak Instrument Test for Panel Data

[Yangxi Zhuguan]

Dear Statalist,

I'm trying to conduct a weak instrument test in my panel data but have little clue how to get started.

The panel is strongly balanced and contains several thousand observations. In addition to a few exogenous independent variables, I have exactly one endogenous explanatory variable of interest (NO2) and one instrumental variable (Ivs_low) for it. So for instance, a 2SLS regression with entity fixed effect will look like:

xtivreg FatalCrash PRCP SNWD AWND TAVG (NO2 = Ivs_low),fe

Please correct me if I'm wrong, is the Kleibergen-Paap Wald rk F statistic the appropriate weak instrument test statistic for panel data?

If so, what is the procedure or command to have Stata calculate it for me?

Does the decision rule of

the instrumental variable is weak if the test statistic is less than or equal to 10

also apply to a panel data?

Please let me know if further information is needed and thank you for your time!

Best,
Yangxi

[Phil Bromiley]

You might look at user written xtivreg2 - it has many of the tests built in.

[Yangxi Zhuguan]

You might look at user written xtivreg2 - it has many of the tests built in.

Thank you, Phil! Will definitely take a look.