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

RE: st: Ctd.: Problem with Bootstrapping panel data


From   "jyotsna puri" <pjyotsna@hotmail.com>
To   statalist@hsphsun2.harvard.edu
Subject   RE: st: Ctd.: Problem with Bootstrapping panel data
Date   Tue, 24 May 2005 21:38:27 +0000

Tried dropping the missing observations before running the bootstrap?
This ordinarily  shouldn't be a problem and I have run bootstrap
with missing observations (it just means fewer replications than
I'd intended it to). But it may help in your case.
See documentation for bootstrap (I am sure you have
already), but particularly for missing observations.

So suggest that you constructing the variable for time, which shows
it to be continuous. Then, drop observations for which  your
variables are missing.

Also, have you tried running your commands for only a balanced subset
of your dataset?

FWIW.
-Jo


Jyotsna (Jo) Puri
pjyotsna@hotmail.com




----Original Message Follows----
From: "Wrede" <wrede@wiso.uni-koeln.de>
Reply-To: statalist@hsphsun2.harvard.edu
To: <statalist@hsphsun2.harvard.edu>
Subject: st: Ctd.: Problem with Bootstrapping panel data
Date: Tue, 24 May 2005 22:43:47 +0200

Hi there,
I posted the same problem 2 weeks ago, but without any final success.
The problem: I try to run bootstrap command together with xtabond estimation
command. But according to the trace output, the program complains that
A) my time variable is not tsset and
B) that there are too many missing observations

The command is:
bootstrap "xtabond vardep varexp, robust" _b[ld.vardep] _b[d.varexp]
_b[_cons], trace reps(100) cluster(id)

There were two suggestions from Cédric and Jo (thank you both again for your
help):
A) introducing a new time variable to bridge time gaps and avoid the missing
observation problem.
B) Some kind of forcing STATA to tsset the data properly by rewriting the
ado file.

However, I dont believe that there should be an easier way, since the
xtabond estimation command has no problem with the panel data. So why should
there be a problem be using bootstrapping with it?

Any suggestions are highly appreciated since I need to find a solution quite
urgent :(

Niko
---------------------------------






The trace gives back:
...
local tmax = r(tmax)
capture local id "`r(panelvar)'"
=capture local id ""
if "`id'"=="" {
= if ""=="" {
di as err "must tsset data and specify panelvar"
must tsset data and specify panelvar

However, I did tsset my data again and again. Any suggestions?

Thanks in advance
Niko




-----Ursprüngliche Nachricht-----
Von: owner-statalist@hsphsun2.harvard.edu
[mailto:owner-statalist@hsphsun2.harvard.edu] Im Auftrag von jyotsna puri
Gesendet: Montag, 16. Mai 2005 23:40
An: statalist@hsphsun2.harvard.edu
Betreff: RE: st: Bootstrapping panel data


Niko,
Are your panel unequally spaced or are there gaps in years?
If so see the (copied and pasted) exchange I had
a while ago.
Hth.
-Jo




----------------------------------------------------------------------------
---------
Hi Jo,
I had that problem once.  What you have to do is generate a new time
variable to avoid r(2000) (no observations error) and tsset using this new
one. For example: (notice that d73 to d98 are time dummies)

gen time2 =. ;

replace time2 = 1 if d73 == 1 ;
replace time2 = 2 if d78 == 1 ;
replace time2 = 3 if d83 == 1 ;
replace time2 = 4 if d88 == 1 ;
replace time2 = 5 if d93 == 1 ;
replace time2 = 6 if d98 == 1 ;

tsset panchayat time2 ;

Hope it works for you.

Best,


Monica L.


-----Original Message-----
From: owner-statalist@hsphsun2.harvard.edu
[mailto:owner-statalist@hsphsun2.harvard.edu] On Behalf Of jyotsna puri
Sent: Monday, December 06, 2004 7:32 AM
To: statalist@hsphsun2.harvard.edu
Subject: RE: st: xtabond and 'no observations'

Hi all,

I am obviously not seeing something.

I have a dataset for a land variable (sr1tara), water(agwsc2), credit
(crbadu1). There are 'id' variables for villages and unbalanced data is
present for 6 years (maximum). I have tried to use the xtabond command and I
get 'no observations'. There is some thing that I am obviously carelessly
missing. Can someone help me out?

At the minimum I do the following (output below).
My questions:
1. Why am I getting 'no observations'?

2. Is this anything to do with the fact that my data is not balanced?
(Shouldnt be. The abdata.dta - the example dataset provided by stata for
xtabond is also unbalanced). I have obviously used other variables in other
commands but it dosent work for even this minimal list...

3. This may have something to do with the note after 'tsset' which says I
have data from 86 to 96, but with gaps.

I have two types of 'missingness'. This is an unbalanced
data - so all villages are not present for
all years. I have tried xtabond for villages that are only present for all 6
years. No avail.

There is also 'item non-response' i.e the village may be present but data
for sr1tara is missing. I have tried two options. In the first, filling up
with tsfill, full. In the second throwing out these observations completely,
and then filling up with tsfill, full.

I have also tried replacing 'missing' values
of sr1tara with zeros and then using the command.
No avail.

-Jo




Jyotsna (Jo) Puri
pjyotsna@hotmail.com




----Original Message Follows----
From: "Wrede" <wrede@wiso.uni-koeln.de>
Reply-To: statalist@hsphsun2.harvard.edu
To: <statalist@hsphsun2.harvard.edu>
Subject: st: Bootstrapping panel data
Date: Sat, 21 May 2005 23:26:07 +0200

Hi statalisters,
I am using panel data, which are tsset. I am using some xt commands like
xtabond etc. Now I want to bootstrap the sample and re-run the tests.

My problem:
STATA complains that my data are not tsset.

My code is:
bootstrap "xtabond vardep varexp, robust" _b[ld.vardep] _b[d.varexp]
_b[_cons], trace reps(100) cluster(id)

The trace gives back:
...
local tmax = r(tmax)
capture local id "`r(panelvar)'"
=capture local id ""
if "`id'"=="" {
= if ""=="" {
di as err "must tsset data and specify panelvar"
must tsset data and specify panelvar

However, I did tsset my data again and again. Any suggestions?

Thanks in advance
Niko

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