Statalist


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

st: RE: xtdpd vs. xtabond2 discrepancy


From   "David Roodman ([email protected])" <[email protected]>
To   [email protected]
Subject   st: RE: xtdpd vs. xtabond2 discrepancy
Date   Mon, 18 Jan 2010 20:13:26 -0500

To recap, jreif's main issue in getting xtdpd and xtabond2 to match was
needing to add an h(2) option to the xtabond2 command line. (There is an
example of this in the How to Do xtabond2 paper, at least the version on
cgdev.org.) But even after adding this, there remained a discrepancy in
the standard errors.

The problem was indeed in xtabond2, and has now been fixed. It only
affected one-step, non-robust, System GMM, which is rare. It was a
vestige of an erroneous formula in DPD for Ox for the variance of the
errors, which I copied years ago. In 2007, David Drukker pointed this
out and I fixed it--but incompletely. "ssc install xtabond2, replace"
should install the latest version. Then these two will match exactly:

webuse abdata
xtdpd n L.n, dgmm(n) lgmm(n)
xtabond2 n l.n, gmm(l.n) h(2) small

--David




*
*   For searches and help try:
*   http://www.stata.com/help.cgi?search
*   http://www.stata.com/support/statalist/faq
*   http://www.ats.ucla.edu/stat/stata/



© Copyright 1996–2024 StataCorp LLC   |   Terms of use   |   Privacy   |   Contact us   |   What's new   |   Site index