AW: st: AW: creating subsamples in panel data

["Martin Weiss" <martin.weiss1@gmx.de>]

<> 

This is a dopey example that shows you valuable techniques in this area:

*************
clear*
set obs 10
gen byte bank=_n

//expand to # of time periods
expand 4

//get time period 
bys bank: gen byte time=_n

//xtset the thing
xtset bank time
gen assets =rnormal(1000000,200000)

l, noo sepby(bank)

//get average assets
bys bank: egen avassets=mean(assets)
egen tag=tag(bank)
egen rank=rank(avassets) if tag

gen x=rnormal()

//regressions based on average assets
reg avassets x if inrange(rank,1,4)
reg avassets x if inrange(rank,5,7)
reg avassets x if inrange(rank,8,10)
*************



HTH
Martin


-----Ursprüngliche Nachricht-----
Von: owner-statalist@hsphsun2.harvard.edu
[mailto:owner-statalist@hsphsun2.harvard.edu] Im Auftrag von David Kehrl
Gesendet: Freitag, 6. August 2010 18:00
An: statalist@hsphsun2.harvard.edu
Betreff: Re: st: AW: creating subsamples in panel data

Yes i want to use the average asset size over the whole sample period.
But my main problem is that once I have e.g. ranked the asset variabel
i dont know who I can connect the ranking to the other variables
without mixing  observations from different banks.
For example i used collapse but the I only get the asset variabel in
order and not the other ones.
But obviously I need to order the whole dataset by asset size.
But maybe I am just confused and missing the point?

Thanks so much,

David

2010/8/6 Martin Weiss <martin.weiss1@gmx.de>:
>
> <>
>
>
> Some function in -help egen- will be the solution, I guess, e.g. the
> -pctile- function, or -rank-. Do you want to base the decision on average
> assets of the bank over the entire sample period, or on those in a
specific
> year?
>
>
>
> HTH
> Martin
>
> -----Ursprüngliche Nachricht-----
> Von: owner-statalist@hsphsun2.harvard.edu
> [mailto:owner-statalist@hsphsun2.harvard.edu] Im Auftrag von David Kehrl
> Gesendet: Freitag, 6. August 2010 17:43
> An: statalist@hsphsun2.harvard.edu
> Betreff: st: creating subsamples in panel data
>
> Hi,
>
> I have have a large panel dataset containing bank data. I have a time
> variabel and a panel variable that identifies every bank uniquely.
> Now I want to divide my data in three groups( by percentiles) by their
> asset size  to run regression for each individual group.
> I tried evey function I know about but nothing seems to work properly.
> I would be extremely grateful for every help I can get on this.If you
> wonder this is for my Bachelor Thesis and kind of new to STATA.
>
> Kindly,
>
> David Kehrl
> *
> *   For searches and help try:
> *   http://www.stata.com/help.cgi?search
> *   http://www.stata.com/support/statalist/faq
> *   http://www.ats.ucla.edu/stat/stata/
>
>
> *
> *   For searches and help try:
> *   http://www.stata.com/help.cgi?search
> *   http://www.stata.com/support/statalist/faq
> *   http://www.ats.ucla.edu/stat/stata/
>

*
*   For searches and help try:
*   http://www.stata.com/help.cgi?search
*   http://www.stata.com/support/statalist/faq
*   http://www.ats.ucla.edu/stat/stata/


*
*   For searches and help try:
*   http://www.stata.com/help.cgi?search
*   http://www.stata.com/support/statalist/faq
*   http://www.ats.ucla.edu/stat/stata/