Re: st: Moment evaluator programs with K equations

["Brian P. Poi" <bpoi@stata.com>]

On Wed, 17 Feb 2010, Christopher Kodongo wrote:

> Dear statalisters
>
> Something appears to be fundamentally wrong with the following program. Trouble is - I can't spot it. Can you?
>
> program tryagain2, eclass
> version 11
> syntax varlist [if], at(name) depvars(varlist) indvars(varlist)
> quietly {
>     tempvar xb
>     generate double `xb' = 0 `if'
>     local k = 1
>     local i = 1
>     foreach var of varlist `depvars' {
>         foreach var of varlist `indvars' {
>             replace `xb' = `xb' + `at'[`k',`i']*`var' `if'
>             local `++i'
>         }
>         local `++k'
>     }
>     replace `xb' = `xb' + `at'[`k',`i'] `if'   
>     replace `varlist' = `depvars' - `xb' `if'               
> }
> end
>
> gmm tryagain2, nequations(3) parameters(wldemr usaemr _cons) ///
>     depvars(kenemr moremr zaremr) indvars(wldemr usaemr) ///
>     instruments(nigemr empemr zaremr) wmatrix(robust) vce(hac bartlett 2) igmm


Since you specified three parameters in the parameters() option, the `at' 
vector is going to be 1 by 3; even if you have multiple equations, the 
parameters are packed as a row vector, not a matrix.

Also, since you specified nequations(3), the local macro `varlist' has 
three words in it.  Part of your program will end up looking something 
like this:

   local m1 : word 1 of `varlist'
   local m2 : word 2 of `varlist'
   local m3 : word 3 of `varlist'
   local dv1 : word 1 of `depvars'
   local dv2 : word 2 of `depvars'
   local dv3 : word 3 of `depvars'
   replace `m1' = `dv1' - (something that depends on `at' and `indvars')
   replace `m2' = `dv2' - (something that depends on `at' and `indvars')
   replace `m3' = `dv3' - (something that depends on `at' and `indvars')


   -- Brian Poi
   -- bpoi@stata.com