Re: st: Standard Errors using Bootstrapping with Panel Data

[Benjamin Volland <volland@econ.mpg.de>]

Hey Nil,

there is no bootstrap option for the user-written (G. Frechette, 2001) 
-reoprob- command. Thus you need to follow the instructions from the 
FAQ: http://www.stata.com/support/faqs/stat/xt_boot.html

Best, Ben

On 22/11/2011 10:39, nil sen wrote:
> Dear Statalist Users,
>
> I have a question regarding bootstrapped standard errors for panel
> data models. Any help would be much appreciated.
>
> I am doing a two stage estimation with endogenous explanatory
> variables (in a two equations simultaneous model setup).
>
> I would like to know, how to write bootstrap commands in order to
> obtain standard errors after xtreg, xtprobit and reoprob (random
> effect ordered probit).
>
> I have 2 sets of equations.........................
>
> The 1st set contains a binary outcome variable (Y) and a continuous
> outcome variable (X). For which, I am using xtprobit and xtreg as
> following:
>
> First stage:
> ----------------
> For the binary outcome model (where Y is binary outcome variable and X
> is continuous explanatory variable, which is endogenous):
>
> xtprobit Y $xvarsall
> predict Y_hat, xb
>
> For the continuous outcome model (where X is continuous outcome
> variable and here Y is binary explanatory variable and is
> endogenous.):
>
> xtreg X $xvarsall
> predict X_hat, xb
>
> **$xvarsall: contains all exogenous variables in the model.
>
> Second stage:
> ---------------------
> For the binary outcome model:
>
> xtprobit Y X_hat $xvars1, vce(bootstrap)
>
> **$xvars1: exogenous variables specific to the binary model.
>
> For the continuous outcome model,
>
> xtreg X Y_hat $xvars2, vce(bootstrap)
>
> **$xvars2: exog. vars. specific to the continuous model.
>
>
> *******My question is: Am I using the vce(bootstrap) option correctly
> with these models? If not what is the correct way of finding
> bootstrapped standard errors for these models?
>
>
> My 2nd set of equations contains an ordered outcome variable (K) and a
> binary outcome variable (L). For which, I am using reoprob (random
> effect ordered probit) and xtprobit and again using a two stage
> estimation technique, similar to the one mentioned above. Here, when I
> am using vce() option with reoprob, it is giving error that "vce()
> option cannot be used with reoprob".
>
> *******My question is: How can I find bootstrapped standard errors
> using the reoprob command?
>
> Any help would be sincerely appreciated.
>
> Thank you very much.
>
> Regards,
>
> Nil Sen
> *
> *   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/

*
*   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/