Re: st: problem with negative binomial fixed effect model

[Nahla Betelmal <nahlaib@gmail.com>]

Thanks for reply, but the issue is not about the mata size, even if I
increase the size , the number of indicator variables is too large
which will take huge computation time and will greatly affect the
degree of freedom. so increasing the mata will not solve the issue
statistically wise.

The authors mention that LIMDEP on  SAS can efficiently do that, and I
wonder if there is something like that in Stata.

There is another solution using the hybrid model , which I will use if
I could not figure out

see the article below for the same author with brief description about
the problem and solutions of negative binomial fixed effect model


http://www.statisticalhorizons.com/fe-nbreg


I wonder if anyone has experienced this issue before , I highly
appreciate your advice.

Thanks

Nahla Betelmal

On 10 March 2014 23:20, Jorge Eduardo Pérez Pérez <jorge_perez@brown.edu> wrote:
> You should try increasing the matsize, see -help matsize-
> --------------------------------------------
> Jorge Eduardo Pérez Pérez
> Graduate Student
> Department of Economics
> Brown University
>
>
> On Mon, Mar 10, 2014 at 6:39 PM, Nahla Betelmal <nahlaib@gmail.com> wrote:
>> Dear Statalist,
>>
>> I want to apply fixed effect negative binomial model on my panel data.
>> the panel variable is firm and I  control for year effect by year
>> indicator variable as follow:
>>
>> xtnbreg dependent-variable independent-variables i.year, fe
>>
>> However, I found this paper by Paul D. Allison and Richard Waterman,
>> "FIXED-EFFECTS NEGATIVE BINOMIAL REGRESSION MODELS " which argues that
>> the above command does not actually produce the correct fixed-effect
>> negative binomial model.
>>
>> http://www.ssc.upenn.edu/~allison/FENB.pdf
>>
>> Instead the authors present the case to include indicator variable
>> (dummy variable) to get the fixed-effect estimates, so the command
>> should be as follow
>>
>> nbreg dependent-variable independent-variables i.year i.firm
>>
>>
>> In my case there is a problem in using this command as I have large
>> number of firms, 1,200 firms for years from 1993-2010. I got the error
>>
>> matsize too small, r(908);
>>
>> Is there a way in Stata to get the correct fixed-effect estimates as
>> suggested by the above authors ? I have 12.1 version, MP edition
>>
>> Thank you for all the help
>>
>> Best Regards
>>
>> Nahla Betelmal
>> *
>> *   For searches and help try:
>> *   http://www.stata.com/help.cgi?search
>> *   http://www.stata.com/support/faqs/resources/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/faqs/resources/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/faqs/resources/statalist-faq/
*   http://www.ats.ucla.edu/stat/stata/