st: Re: hodrick prescott filter

[Kit Baum <baum@bc.edu>]

On Nov 1, 2009, at 2:33 AM, Michael wrote:

> On Oct 29, 2009, at 12:45 PM, Marwan Elkhoury wrote:
>
> > I have a panel data with 180 countries over 10 years (annual data)
> > and I tried to run a hodrick prescott filter over it to smooth data
> > by typing the following command in stata 10:
>
> H-P filtering 12 annual observations seems a highly dubious practice,
> even if it were possible.  Even detrending such a short sample with a
> quadratic or cubic trend seems to be asking quite a lot -- if not too
> much -- from those data.
>
>
> > by cntry: hprescott realgdomcred realgdpgrowth if tin(1997, 2008),
> > stub(HP) smooth(6.25)
> >
> > and I get error 3301 subscript invalid;
> >
> > I've tried to increase memory size to 100M
> > and maxvar size to 10000
> > as well as matsize to 6500 (could not increase it to more)
>
> Kit Baum, the author of -hprescott-, would be the best person to
> answer this question.  Nonetheless, I would not be surprised if the -
> hprescott- code implicitly assumes a much longer time series, for the
> reason stated above.



I agree wholeheartedly with Michael that time series filtering  
techniques like the HP filter make little sense in the context of 10  
observations. However I cannot reproduce the error:

. webuse grunfeld,clear

. by company:hprescott invest kstock if tin(1935,1947), stub(HP) smooth 
(6.25)

does indeed generate results for each company.   Return code 3301 is a  
Mata error, but without an example of how this error can be generated,  
I don't know where it might be generating a problem.

Kit

Kit Baum   |   Boston College Economics & DIW Berlin   |   http://ideas.repec.org/e/pba1.html
                              An Introduction to Stata Programming   
|   http://www.stata-press.com/books/isp.html
   An Introduction to Modern Econometrics Using Stata  |   http://www.stata-press.com/books/imeus.html

*
*   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/