Re: st: Re: st: How do I suppress computation of standard errors in stata gmm function?

[Michael Barker <mdb96statalist@gmail.com>]

Hi Howard and others,

Thanks for clarifying the time requirement for standard error
calculations. I haven't used GMM very much. The problems I work on
spend a lot of time in iterations and not much in standard error
calculations.

Maybe as a work-around, you could use the vce option to calculate some
faster, possibly incorrect standard errors. For example, you could
try:

vce(unadjusted, independent)

or

vce(bootstrap, reps(2) size(50))

Mike

On Thu, Aug 1, 2013 at 5:20 AM, Howard Smith
<howard.smith@economics.ox.ac.uk> wrote:
> Hi Mike
>
>
>
> Thank for your reply.
>
>
>
>> When you run the estimator once, how much time elapses between the
>
>> final iteration and displaying results? That would be an upper bound
>
>> on the time savings from eliminating the standard error calculation.
>
>
>
> I did this upper bound check, and in fact that's why I wanted to turn off the standard errors.
>
>
>
>
> Quite a lot of time elapses after the last iteration and the computation of standard errors: several minutes. Even though I supply analytical gradients.
> I am not sure why it takes so long,
> perhaps because I have a lot of parameters (50) and a very large dataset.
>
>
>
>
> It's a pity they can't be switched off.
>
>
>
> Howard
>
>
>
> PS I hope I have responded to this reply correctly. I am a new statalist person, and I am not sure how to reply to people, given that I signed up to statalist in "digest" mode.
>
>
>
>
> On Wed, Jul 31, 2013 at 9:40 AM, Michael Barker
>
> <mdb96statalist@gmail.com> wrote:
>
>> Hi Howard,
>
>>
>
>> I don't know of any way to switch off the standard error calculation,
>
>> but I would guess that doing so would not speed up the estimation very
>
>> much. If the GMM estimation is taking a long time, it is probably
>
>> spent in the minimization routine, finding the parameter estimates.
>
>> Once the minimization is complete, all of  the components to calculate
>
>> standard errors (gradients, etc.) have already been calculated.
>
>>
>
>> When you run the estimator once, how much time elapses between the
>
>> final iteration and displaying results? That would be an upper bound
>
>> on the time savings from eliminating the standard error calculation.
>
>>
>
>> Mike
>
>>
>
>>
>
>>
>
>> On Tue, Jul 30, 2013 at 5:44 AM, Howard Smith
>
>> <howard.smith@economics.ox.ac.uk> wrote:
>
>>> I am using the gmm() function interactively in stata, with analytical gradients and two equations.
>
>>>
>
>>> I would like to obtain estimated parameters but not standard errors!
>
>>>
>
>>> This is to speed things up (for bootstrapping for example, but also just to try a few alternative specifications). I guess that computing standard errors takes time.
>
>>>
>
>>> My question then: Is it possible to "switch off" the computation of standard errors? If so, how? I have searched quite a lot to figure this out but can see no reference to it in documentation or in past statalist discussions.
>
>>>
>
>>> Many thanks for any assistance you can offer.
>
>>>
>
>>> Howard Smith
>
>
>
>
>
>
>
>
> *
> *   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/