As an aside, does anyone know of a basic level explanation of the
benefits
to using xtabond2 v. xtabond?
The xtabond2 help files does a pretty good job of it. One can use .2 to
do GMM-SYS (Blundell-Bond) as well as original Arellano-Bond. The
syntax of .2 is more flexible: there is an example in Greene's
Econometrics text which cannot be est by xtabond, but can be est with
xtabond2. And .2 does the "Windmiejer correction", without which
two-stage results are generally junk in terms of inference. Every other
package that does DPD provides Windmeijer.
