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

RE:more questions st: RE: xtabond


From   "Dimova, Ralitza" <Ralitza.Dimova@econ.kuleuven.ac.be>
To   <statalist@hsphsun2.harvard.edu>
Subject   RE:more questions st: RE: xtabond
Date   Sun, 21 Sep 2003 12:00:46 +0200

Hi again,

I still don't absolutely understand the xtabond syntax and what exactly stata is doing; maybe because my "x" part of the manual disappeared and I am only using the "help".... Sorry if I am asking about the obvious.

First of all, what exactly does "xi" do and is it only used with xtabond? My problem was NOT that I WANTED to include year dummies, but that I DID NOT WANT to include them. Since there was a crisis in a certain year in my time series I wanted to check whether there is a structural change after the crisis. I defined a "CRISIS" variable including all the period after the crisis. However, when I was trying to run xtabond with stata (without xi) stata was automatically including year dummies and my crisis variable was dropping due to collinearity.

Secondly, do I correctly understand that the data that we include in stata before running xtabond should be in levels and that xtabond is automatically differencing whatever we define as predetermined variables and using their lags in both levels and differences as instruments. And if we want to include more instruments, additional to what Arellano and Bond (1991) talk about we need to indclude them in the "inst" option? But then what exacly is the "diff" option doing? In the help they say it "specifies a set of variables that are already differenced to be included as stricly exogenous covariates". I don't understand the "already differenced" part in so far as year and esp industry dummies are concerned. In sum do we difference or infline in levels the original file?

Finally, stata seems to be automatically using as a regressor the first difference of the dependent variable. However, I want to interact this first difference with a variable called "trend". How do I go about it? Do I just do the interaction in advance and then plug it as "predetermined" disregarding the fact that stata will automatically include the uninteracted lag? 

Thank you very much.

Ralitza

-----Original Message-----
From: Mine Zeynep Senses [mailto:msenses@umich.edu]
Sent: Tuesday, September 16, 2003 6:15 PM
To: statalist@hsphsun2.harvard.edu
Subject: Re: st: RE: xtabond 


Hi Ralitza,
Thanks fot the reply... first, your question: you can add year dummies
using the xi command along with xtabond. then you just have to define
diffvars(i.year).
mine

On Tue, 16 Sep 2003, Dimova, Ralitza wrote:

> Hi Mine,
>
> I also had trouble using time dummy interactions with the xtabond
> command. In fact, I couldn't figure out a way whereby the program does
> not automatically include year dummies in the regression. Have you?
>
> Have you tried using the Gauss DPD instead? It turned out easier than
> I expected. And it is very flexible in specifying instruments with the
> "right" lags. If you are interested I can send you a detailed
> algorithm and there are also good descriptions on the Arellano's
> website.
>
> Ralitza
>
> -----Original Message-----
> From: Mine Zeynep Senses [mailto:msenses@umich.edu]
> Sent: Tuesday, September 16, 2003 5:00 AM
> To: statalist@hsphsun2.harvard.edu
> Subject: st: xtabond
>
>
> Hi,
> I sent this e-mail a while ago and I got no reply. I am sort of hoping
> that somebody who used this command joined the list in the meantime. sorry
> for duplicates but I am sort of getting desperate!!
> so any little help will be very appreciated.
> mine
>
> > I have a question regarding the xtabond command. I am using:
> > xi: xtabond  investment_2, lags(1) maxlags(3) pre(sales_k, lag(0,3))
> > pre(cash_k, lag(0,3)) diffvars(i.year i.industry)
> >
> > I want to add an interaction variable:  year*cash_k. Year is a dummy
> > variable and cash_k is predetermined. So what i would like is to
> > instrument this variable with year*cash_k(lag) etc. Defining it
> > predetermined will use (year*cash_k)(lag) which is probably not right.
> >
> > is there a way to do this? I will appreciate any help.
> > mine
>
> *
> *   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/
>
> *
> *   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/
>

*
*   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/

*
*   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 1996–2021 StataCorp LLC   |   Terms of use   |   Privacy   |   Contact us   |   What's new   |   Site index