Re: st: RE: Bootstrap error

[Carl Nelson <chnelson@uiuc.edu>]

At 03:16 PM 5/4/2006, you wrote:
> The first error is a Statalist
> mistake: you appear to be using Stata 8 and not
> telling us that.
My apologies. I am using Stata 9, but I am using the Stata 8 bootstrap syntax


> More seriously, in your program [not "do file"]
>
> program kuznets, rclass
>         args lnSEC lnGDPC lnGDPC2
>         tsset countrycode year
>         xtabond2 lnSEC l1.lnSEC lnGDPC lnGDPC2, gmm(l1.lnSEC) ///
>       iv(lnGDPC lnGDPC2)
>         gen kuz = exp(-_coef[lnGDPC]/(2*_coef[lnGDPC2]))
>         return scalar x1 = kuz
> end
>
> You do nothing with the results of -args-, so that can go.
>
> I doubt you need to -tsset- every time, but you'd need to
> check that.
>
> There is no need to put a constant in a variable. It's poor style.
>
> So, this looks more like
>
> program kuznets, rclass
>         tsset countrycode year
>         xtabond2 lnSEC l1.lnSEC lnGDPC lnGDPC2, gmm(l1.lnSEC) ///
>       iv(lnGDPC lnGDPC2)
>         return scalar x1 = exp(-_coef[lnGDPC]/(2*_coef[lnGDPC2]))
> end
>
> Now what -bootstrap- should pick up is r(x1):
>
> So your call should be more like
>
> bootstrap "kuznets" r(x1), reps(25) cluster(countrycode)
> idcluster(countryrs)
I tried the suggested change and obtained the following result:

 do leekuznets

. tsset countrycode year
       panel variable:  countrycode, 1 to 66
        time variable:  year, 1960 to 1990

. program kuznets, rclass
  1.         args lnSEC lnGDPC lnGDPC2
  2.         xtabond2 lnSEC l1.lnSEC lnGDPC lnGDPC2, gmm(l1.lnSEC) 
iv(lnGDPC lnGDPC2)
  3.         return scalar x1 = exp(-_coef[lnGDPC]/(2*_coef[lnGDPC2]))
  4. end

.
. bootstrap "kuznets" r(x1), reps(200) cluster(countrycode) idcluster(newid)

command:      kuznets
statistic:    _bs_1      = r(x1)

Bootstrap statistics                              Number of obs    =      1980
                                                  N of clusters    =        66
                                                  Replications     =       200

------------------------------------------------------------------------------
Variable     |  Reps  Observed      Bias  Std. Err. [95% Conf. Interval]
-------------+----------------------------------------------------------------
       _bs_1 |     0  42843.02         .         .          .          .   (N)
             |                                              .          .   (P)
             |                                              .          .  (BC)
------------------------------------------------------------------------------
Note:  N   = normal
       P   = percentile
       BC  = bias-corrected

insufficient observations to compute bootstrap standard errors
no results will be saved
r(2000);

which is no change from my previous result.

And the direct call to bootstrap produces the following, which is the 
reason that I wrote the intermediate program.

bootstrap x1=exp(-_coef[lnGDPC]/(2*_coef[lnGDPC2])), reps(200) 
cluster(countrycode) idclu
> ster(newid) : xtabond2 lnSEC l1.lnSEC lnGDPC lnGDPC2, 
gmm(l1.lnSEC) iv(lnGDPC lnGDPC2)
(running xtabond2 on estimation sample)

Bootstrap replications (200)
----+--- 1 ---+--- 2 ---+--- 3 ---+--- 4 ---+--- 5
repeated time values within panel
the most likely cause for this error is misspecifying the cluster(), 
idcluster(), or
group() option

The low number of replications is just a test to get the code working.

Carl Nelson 
*
*   For searches and help try:
*   http://www.stata.com/support/faqs/res/findit.html
*   http://www.stata.com/support/statalist/faq
*   http://www.ats.ucla.edu/stat/stata/