Statalist


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

Re: st: Re: save coeff estimators of Arima model with foreach


From   "Hua Pan" <[email protected]>
To   [email protected]
Subject   Re: st: Re: save coeff estimators of Arima model with foreach
Date   Thu, 12 Mar 2009 13:22:58 +0100

Hello Marin, Hello Statalisters,

Thank you very much for your quick response and quite a few advices, Martin. It’s really very kind of you.

Following your advice, I picked out some gvkey randomly and run the foreach command several times, but it appears the same result, all of the AR remain missing.

I’ve tried statsby, which you advice me
- tsset gvkey qtl
-statsby, by(gvkey): arima eps, arima(1,0,0) sarima(0,1,0,4)
but an error appears:
-sample may not include multiple panels
-an error occurred when statsby executed arima
-r(459);

I’ve checked rolling, estimates store, eststo and postfile, but they don’t seem to help a lot. I just want to generate the new variable “_b[]” in the old datasheet. But thank you all the same for your advice.

I really have no idea what’s wrong with it, it seems to be correct like your sample, but it doesn’t work!?

Maybe some more details may help us to find out the key:
I tied this one for just one gvkey and it works well:
- tsset qtl
- arima eps, arima(1,0,0) sarima(0,1,0,4)
- generate AR = _b[ARMA:L1.ar]

and for many gvkey(panel data) without loop:
- tsset gvkey qtl
- by gvkey, sort : arima eps, arima(1,0,0) sarima(0,1,0,4)
tables of estimation of all gvkeys appear on the screen, it works well. 
But wenn I typed:
- generate AR = _b[ARMA:L1.ar]
The value of ARs for all gvkeys is the value of AR for the last gvkey. So I want to try with foreach to get AR (AR-Coef.) for each gvkey and save them als new variable in datasheet.
Thany you very much in advance.

Best Regards

Hua 

-------- Original-Nachricht --------
> Datum: Thu, 12 Mar 2009 01:21:42 +0100
> Von: "Martin Weiss" <[email protected]>
> An: [email protected]
> Betreff: st: Re: save coeff estimators of Arima model with foreach

> <>
> 
> I cannot see what is inherently wrong in your code. If others want to
> check 
> my reasoning, it is pretty much equivalent to
> 
> *******
> sysuse auto, clear
> g coeffs=.
> qui levelsof rep78, loc(levels)
> qui foreach l of local levels {
>      reg pr we f tr if rep == `l'
>      replace coeffs= _b[tr] if rep == `l'
> }
> sort rep
> l coef rep78, noo sepby(rep)
> *******
> 
> which does work, and is itself equivalent to the easier
> 
> *******
> statsby _b[tr], by(rep78) clear : reg pr we f tr
> *******
> 
> though that -after your confirmation- destroys your dataset in memory.
> 
> 
> So you say that you have checked you can run these commands on the entire 
> sample. That need not mean much. Rather, pick out some level of your gvkey
> randomly and run the thing with the if-statements present. Try that for a 
> couple of these levels to get a flavor of what is going on and report the 
> results back to the list...
> 
> HTH
> Martin
> _______________________
> ----- Original Message ----- 
> From: "Hua Pan" <[email protected]>
> To: <[email protected]>
> Sent: Wednesday, March 11, 2009 11:42 PM
> Subject: st: save coeff estimators of Arima model with foreach
> 
> 
> > Hi statalisters,
> >
> > I’m a beginner for Stata, and I have just a problem with it. I want to
> do 
> > a cross-sectional time series analysis with command “arima” and
> imput the 
> > estimator of “AR L1. Coefficient” of every panel data into the
> datasheet 
> > with the help of “foreach” and “levelsof”.
> >
> > My Data looks like this:
> > gvkey      qtl         eps
> > 1 2003q1        ….
> > 1       2003q2        ….
> > 1       2003q3        ….
> > ...       …….
> > 2       2003q1
> > 2       2003q2
> > 2       2003q3
> > …
> > gvkey:  identify-Nr. of firms (panel variable, long)
> > qtl:     quarter (time variable, float)
> > eps:    earnings per share (dependent variable, float)
> >
> > So I’ve tried:
> >
> > - generate AR =.
> > tsset gvkey qtl
> > levelsof gvkey, local(levels)
> > foreach l of local levels {
> > arima eps if gvkey == `l', arima(1,0,0) sarima(0,1,0,4)
> > replace AR = _b[ARMA:L1.ar] if gvkey == `l'
> > }
> >
> > But it doesn’t seem to work. Time series analysis (Arima) hasn’t
> been done 
> > and all of AR are still missing value.
> > I’m sure this is correct:
> > - arima eps, arima(1,0,0) sarima(0,1,0,4)
> > - replace AR = _b[ARMA:L1.ar]
> > I’ve tried it for data with one Observation (one gvkey)
> >
> > I guess, there may be something wrong with foreach or levelsof, but I
> can’t 
> > find the error. Can anyone here give me any advice?
> > Any help would be greatly appreciated. Thanks in advance for your time.
> > I’m using Stata version 10.1 SE (Windows).
> >
> > Best Regards
> >
> > Hua Pan
> > University of Cologne
> >
> >
> > -- 
> > Psssst! Schon vom neuen GMX MultiMessenger gehört? Der kann`s mit
> allen: 
> > http://www.gmx.net/de/go/multimessenger01
> > *
> > *   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/

-- 
Psssst! Schon vom neuen GMX MultiMessenger gehört? Der kann`s mit allen: http://www.gmx.net/de/go/multimessenger01
*
*   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/



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