Hi list,
It appears that the dynamic panel data estimator -xtabond- is different
in Stata8 and Stata9. However, this change is not reflected in
the "What's new" website for panel data (see
http://www.stata.com/stata9/more.html#paneldata ). I noticed the
following change and can only partially make sense of it (and there
might be other changes too):
xtabond var1, pre(var2, lag(prelags , premaxlags )
run in Stata8 gives the same result as
xtabond var1, pre(var2, lag(prelags , premaxlags+1 )
run in stata9, as long as prelags is set to 0. However, if we use some
value other than 0, it is not clear to me how to get the same results
in the two stata versions. Looking at the help for xtabond in the two
versions, the help has not changed at all.
So when no further lags of var2 are included in the model, the number
of lags of var2's levels that Stata8 uses as instruments is (T_i -
premaxlags), and the number of lagged levels Stata9 uses as instruments
is (T_i - premaxlags - 1). But this difference, if I am correct, is not
captured in the respective help viewers. In any case, as soon as we
include lags of var2 in the equation, I don't see how Stata8 and
Stata9's xtabonds can be reconciled. It would be great if someone could
comment on this if they have more insights into how xtabond is working
in the two versions.
Tewodaj
~.~.~.~.~.~.~.~.~.~.~.~.~.~.~.~.~.~.~.~.~.~.~
Tewodaj Mogues
Dept. of Agricultural and Applied Economics
University of Wisconsin - Madison
427 Lorch St. #317, Taylor Hall
Madison, WI 53706
*
* 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/
