Stata The Stata listserver
[Date Prev][Date Next][Thread Prev][Thread Next][Date index][Thread index]

st: RE: Already Differenced Variables in xtabond2


From   "Salvati, Jean" <JSalvati@imf.org>
To   <statalist@hsphsun2.harvard.edu>
Subject   st: RE: Already Differenced Variables in xtabond2
Date   Tue, 8 Nov 2005 12:43:53 -0500

> My regression in a xtabond command line:
> xtabond y x1 x2 x3, nocons diff(x4 x5)
> 
> How do I translate this into xtabond2?
> According to the xtabond2 help and some suggestions here in 
> Statalist Archive I tried
> 
> xtabond2 y l.y x1 x2 x3 x4 x5, iv(x1 x2 x3) iv(x4 x5, 
> passthru) gmm(l.y) noleveleq


The source of the discrepancy in the coefficients produced by your two
commands is the difference between diff() and iv():

- diff(x4 x5) prevents x4 and x5 from being first-differenced in the
equation as well as in the instruments matrix.

- iv(x4 x5, passthru) only prevents x4 and x5 from being
first-differenced in the instruments matrix; x4 and x5 will still be
first-differenced in the equation.

There is no option in the current version of xtabond2 to prevent
regressors from being first-differenced in the first-difference
equation. 

Apart from that, xtabond and xtabond2 give slightly different results
for the standard deviations and the J-stat. I have no idea why. I've
never looked closely at the code for xtabond. 

I can send you a custom version of xtabond2 that will give you the same
coefficients as -xtabond, diff()-, but you will still observe small
discrepancies in the J-stats and standard deviations.

Jean Salvati

*
*   For searches and help try:
*   http://www.stata.com/support/faqs/res/findit.html
*   http://www.stata.com/support/statalist/faq
*   http://www.ats.ucla.edu/stat/stata/



© Copyright 1996–2014 StataCorp LP   |   Terms of use   |   Privacy   |   Contact us   |   What's new   |   Site index