Bookmark and Share

Notice: On April 23, 2014, Statalist moved from an email list to a forum, based at

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

st: Use ivregress or xtivreg2

From   FredVer <>
Subject   st: Use ivregress or xtivreg2
Date   Wed, 16 May 2012 05:19:04 -0700 (PDT)

Dear all,

Im stuck in deciding between the two regression models. Let me explain. I
have a panel dataset with unique id's (banks) on 23 countries over time (10
years). Its a unbalanced dataset WITH gaps in years...

I have on model in where the dep variabel is averaged for banks, indep
variables averaged for banks and some country variables averaged on country

This model with: ivregress gmm dep indep (end=istr instr), vce(cluster
councode) works fine. I've tested for heteroskedasticity which is present
and therefore the gmm estimation is used. The instruments also are relevant
and valid. And beneficially the variables used are most of the time highly

But! Now I also have a important variable estimated on the country level
over time, its the Boone-indicator (Bjt). As i want to keep this for country
and years, i cannot use the ivregress command anymore.

Therefore the xtivreg2 for panel data is a good solution, but now some
instruments arent valid anymore...this as some of these are time-invariant.

I now have two regression models, the ivregress i understand, but the second
problem is that i do not quite understand the output of xtivreg2...(and the
problem now I cannot copy paste the output from stata :S)

Please can someone help me out on this...what to do. I have to decide asap
as I have a deadline.

View this message in context:
Sent from the Statalist mailing list archive at
*   For searches and help try:

© Copyright 1996–2018 StataCorp LLC   |   Terms of use   |   Privacy   |   Contact us   |   Site index