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

st: RE: Bootstrapping from panel data


From   "Nick Cox" <n.j.cox@durham.ac.uk>
To   <statalist@hsphsun2.harvard.edu>
Subject   st: RE: Bootstrapping from panel data
Date   Sun, 13 Oct 2002 22:46:54 +0100

Patrice Yee Chong San
>
> I have a balanced panel data of N=1130 firms over T=36 time
> periods and
> 15 explanatory variables. What I plan to do now is some sort
> of 'bootstrapping.
>
> I want to create a random sample of n<N firms with
> replacement in such
> a way that i also pick their entire corresponding time
> series at the
> same time. And also, i want to ensure that a particular
> firm can be in
> the sample more than once.
>
> The bstrp command in Stata only  picks individuals/firms i from a
> particular period but i want to preserve the time structure
> of my data.
> I have been through the manuals and FAQ but cannot find anything on
> bootstrapping from panel data. The goal is to do my
> analyses on this
> sample and to repeat this over and over and then to compare results.
>
> Is there any way to do this? The solution is eluding me.

I don't know whether what you are proposing
is a valid procedure, but if the Stata documentation
doesn't answer the question, then the wider literature
may help.

As for a Stata way of doing this, this may help: your data
are currently in a long shape, so you could do whatever
you do for all panels and store each result of interest in a variable
with the same value within each panel. Then -reshape-
to wide so that each panel is mapped into one observation
and take samples with replacement from that data set.

Nick
n.j.cox@durham.ac.uk

*
*   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 1996–2014 StataCorp LP   |   Terms of use   |   Privacy   |   Contact us   |   What's new   |   Site index