Re: st: Stata loop execution, failing to take into consideration all variables

[Nick Cox <njcoxstata@gmail.com>]

In addition, as I understand it, your various samples overlap. I can't
see that would create your problem, but I think it won't make it
easier.

On Wed, Mar 2, 2011 at 1:28 PM, Nick Cox <njcoxstata@gmail.com> wrote:
> I don't have your data, but in any case the bigger issue is how to
> debug a problem like this. I would take your loops and turn them into
> code that spits out what data are being identified. Looking at the
> first few funds should help to identify what is being looked at, say
>
> forvalues f= 1/10 {
>        forvalues y=1998(1)2010{
>                        forvalues m=1(1)12{
>                               list reg exret mktrf hml smb if
> fundid==`f' & (date>=ym(`y'-3,`m') & date<=ym(`y',`m'))
>                }
>        }
>  }
>
> or (less output, and also less diagnostic, but may reveal a problem)
>
> forvalues f= 1/10 {
>        forvalues y=1998(1)2010{
>                        forvalues m=1(1)12{
>                               count if fundid==`f' &
> (date>=ym(`y'-3,`m') & date<=ym(`y',`m'))
>                }
>        }
>  }
>
> On Wed, Mar 2, 2011 at 1:11 PM, S.A.J.van Vijfeijken
> <S.A.J.vanVijfeijken@uvt.nl> wrote:
>
>
>> As a means of return comparisment in my mutual fund research I am using the Fama-French factors as a measurement of excess return.
>>
>> However, when I want to regress the Fama-French factors for every fund in my database Stata only calculates the values for the first fund number.
>>
>> The data used is from the CRSP mutual fund data base, monthly mutual fund returns and the monthly Fama-French factors from the website of Kenneth R.  French for the 1995-2010 period. I have used the panel data command tsset to let Stata know I’m using panel data.
>>
>> The part of the code that does not work is as follows:
>>
>> gen exret = mret - rf
>> sort crsp_fundno
>> egen fundid = group(crsp_fundno)
>> gen hmlexp = .
>> gen smbexp = .
>> gen performance = .
>>
>> forvalues f= 1(1)11281{
>>        forvalues y=1998(1)2010{
>>                        forvalues m=1(1)12{
>>                        quietly reg exret mktrf hml smb if fundid==`f' & (date>=ym(`y'-3,`m') & date<=ym(`y',`m'))
>>                        replace performance = _b[_cons] if e(sample)
>>                        replace hmlexp = _b[hml] if e(sample)
>>                        replace smbexp = _b[smb] if e(sample)
>>                }
>>        }
>> }
>>
>> As you can see I want the data for the 1998-2010 period and I have 11281 mutual funds in my data currently. However Stata only calculates the coefficients for the first fund, and leaves all the others as missing(.), the way they are generated. I looked at the Statalist database and the FAQ, however the solutions offered there did not help.
>> http://www.stata.com/support/faqs/data/foreach.html
>> I cannot figure out why Stata stops after the first fund and can’t find the error in the code (if there is one).
>>
>

*
*   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/