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

st: RE: xtabond, xtabond2 & missing values

From   "Martin Weiss" <[email protected]>
To   <[email protected]>
Subject   st: RE: xtabond, xtabond2 & missing values
Date   Thu, 21 Jan 2010 22:41:56 +0100


Just for the record: -xtabond- is Stata`s own command, while -ssc d
xtabond2- is a user-written command. There is no guarantee _whatsoever_ that
these two will get along well and throw up the same results.
may be of interest to you.

Also note this recent thread with a similar topic:


-----Original Message-----
From: [email protected]
[mailto:[email protected]] On Behalf Of Margit Averdijk
Sent: Donnerstag, 21. Januar 2010 22:05
To: [email protected]
Subject: st: xtabond, xtabond2 & missing values

Dear Statalisters,

I am estimating Arellano-Bond models on an unbalanced dataset, and compared
output from xtabond and xtabond2. I used the following commands:

xtabond  Y            time, twostep robust nocons lag(1)
xtabond2 Y            l.Y time, twostep robust gmmstyle(Y, laglimits(2 .))
iv(time) noleveleq

I expected these two commands to give the same results, but they differ
somewhat. I understood that xtabond2 replaces missings with zero's, while
xtabond does not, and that this may produce the difference in results. Is
there any way to prevent xtabond2 from replacing the missings by zero's?
Moreover, from a Statalist-post in 2006 I understood that xtabond2, but not
xtabond, might have problems with gaps in panel data. Does anyone know
whether this is indeed the case?

Many thanks in advance,

Best wishes,
Margit Averdijk
Hotmail: Trusted email with Microsoft?s powerful SPAM protection.
*   For searches and help try:

*   For searches and help try:

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