Re: st: Blundell & Robin (1999) estimator

[Jorge Eduardo Pérez Pérez <perez.jorge@ur.edu.co>]

That is one of the alternatives I suggested in the other thread.

"* Estimate the model with -nlsur- including the residuals from the
first stage regression in the share equations. Again, adjustment of
the standard errors is needed. I am not aware of the adjustment in
this case. Bootstrapping would be slower."

To be clear, to make this similar to the Blundell and Robin estimator,
you would have to bootstrap the two steps, not just the -nlsur- part.
Do not iterate on the variance-covariance matrix in -nlsur- if you're
going to do this: It will only slow down each bootstrap iteration, and
even without iterating the coefficients you obtain are consistent.


_______________________
Jorge Eduardo Pérez Pérez


On Tue, Oct 11, 2011 at 12:04 PM, Ivica Rubil <irubil@eizg.hr> wrote:
> My question is:
> Is it wrong to do the following:
> 1. regress total expenditure on disposable income as an instrument and on other explanatory variables from the share equations
> 2. include the residuals from the above first stage in share equations and estimate the system using -nlsur-, bootstraping std errors
>
> Thanks.
>
>
> --
> Ivica Rubil
> Ekonomski institut / The Institute of Economics, Zagreb
> Trg J. F. Kennedyja 7, 10 000 Zagreb, Croatia
> tel. +385-1-2362-269
> fax. +385-1-2335-165
> irubil@eizg.hr
> www.eizg.hr
>
>
> -----Original Message-----
> From: owner-statalist@hsphsun2.harvard.edu [mailto:owner-statalist@hsphsun2.harvard.edu] On Behalf Of Jorge Eduardo Pérez Pérez
> Sent: 11. listopad 2011 17:37
> To: statalist@hsphsun2.harvard.edu
> Subject: Re: st: Blundell & Robin (1999) estimator
>
> Ivica wrote to me privately, I am sharing the conversation with the list.
>
> What makes the Blundell and Robin estimator easi for pratical purposes
> its precisely that each regression is linear on each iteration. Doing
> a complete non linear estimation on each iteration kind of kills the
> purpose. Also, I am not aware of what the correction for standard
> errors might be if you use -nlsur qaids- in an iterative procedure.
>
> Your alternatives are summarized in this thread:
>
> http://www.stata.com/statalist/archive/2011-05/msg01514.html
>
> To program the Blundell and Robin estimator, you can follow this code,
> which uses the iterative estimator on another demand system (The EASI
> demand system). It also shows you how to use -reg3- inside the
> iterations to deal with endogeneity :
>
> http://www.sfu.ca/~pendakur/iterated_3sls_without_pz,py,zy.do
>
> Regards,
>
> Jorge Pérez.
> -----
>
> Dear Jorge Eduardo,
>
> I have recently googled how to estimate a demand system and I came
> across your name a number of times. I have a question and hope you can
> help me. I am trying to estimate QUAIDS for a number of food
> categories using data from Croatian Household Budget Survey. I wish to
> account for zero-expenditures and deal with potential endogeneity of
> total expenditures. As I have seen so far, there is a number of
> different approaches to do that. I intend to use the Shonkwiler and
> Yen (1999) approach to deal with zero-expenditures and the Blundell
> and Robin
> (1999) approach to deal with endogeneity of total expenditure. Now,
> the problem is that I am not good at Stata programming, and therefore
> I can hardly write a program for Bludell and Robin's Iterated Least
> Squares Estimator. My question is: Do you think it would be right to
> use the procedure for dealing with endogeneity used in Blundell and
> Robin within nonlinear SUR framework, i.e. using Poi's (2008) -nlsur
> quaids- command in Stata. It seems to me that Blundel and Robin's
> procedure for dealing with endogeneity need not necessarily be used
> only together with their iterated LS estimator. Am I right?
>
> Sorry for bothering you this way.
>
> Thanks.
>
> Best regards,
>
> Ivica Rubil
>
> --
> Ivica Rubil
> Ekonomski institut / The Institute of Economics, Zagreb Trg J. F.
> Kennedyja 7, 10 000 Zagreb, Croatia tel. +385-1-2362-269 fax.
> +385-1-2335-165 irubil@eizg.hr www.eizg.hr
>
>
>
> _______________________
> Jorge Eduardo Pérez Pérez
>
>
>
>
> On Mon, Oct 10, 2011 at 5:49 AM, Nick Cox <njcoxstata@gmail.com> wrote:
>> In such cases, you should always try -findit- or Googling first. That
>> should certainly find inbuilt commands. Did you do that?
>>
>> Naturally, it is always possible that someone is sitting on something
>> they have written but not made public, or has done this under another
>> name, but if -findit- or Google finds nothing, you probably need to
>> write a program.
>>
>> Nick
>>
>> On Mon, Oct 10, 2011 at 10:29 AM, Ivica Rubil <irubil@eizg.hr> wrote:
>>
>>> Does anybody knows how can I implement using Stata the Blundell & Robin
>>> (1999) procedure (see full reference below)? Do I have to program
>>> something extensively on my own or there is an in-built procedure that
>>> implements the iterated estimator proposed in the paper. Thanks.
>>>
>>> Blundell, Richard & Robin, Jean Marc, 1999. "Estimation in Large and
>>> Disaggregated Demand Systems: An Estimator for Conditionally Linear
>>> Systems," Journal of Applied Econometrics, John Wiley & Sons, Ltd., vol.
>>> 14(3), pages 209-32, May-June
>> *
>> *   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/
>
> *
> *   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/