[FernandoRios]

All of them
but at the very least all time varying ones
time invariant can still be included

[Charles Saba]

Hi Everyone, please what is the STATA code for running GMM quantile regression

[Charles Saba]

@ Dear Prof Joao Santos Silva and others, please could you with the above question.


Regards,

[FernandoRios]

Well there are many approaches
mmqreg and Xtqreg do that manually
there is also sivqr which is very explicit about the gem too
do you have a explicit question you are trying to answer?

[Charles Saba]

@FernandoRios thanks so much. I was thinking there should be a single code that takes care of GMM quantile regression and not mmqreg and Xtqreg.

Please how do i use these two codes mmqreg and Xtqreg for GMM quantile regression. Can you help with the steps please.

[FernandoRios]

again, they do so, the same way as qreg

qreg y x1 x2 x3, q(50)

This two use GMM. (generalized method of moments) but do it under different assumptions.
mmqreg y x1 x2 x3, q(50)
sivqr y x1 x2 x3, q(50)

I do feel you are trying to do something else here tho. Perhaps it depends on your research question or modeling purpose

[Charles Saba]

@FernandoRios thanks so much for the above steps. Do you have a material that could be of help for GMM quantile regression. I will really appreciate this.

[FernandoRios]

Look in the help files and the reference papers

[Joao Santos Silva]

Dear Charles Saba,

FernandoRios already provided great advice. If I understand you correctly, the main reference is

Machado, J.A.F. and Santos Silva, J.M.C. (2019), Quantiles via Moments, Journal of Econometrics, 213(1), pp. 145–173.

In the paper we treat the case of panel models with fixed effects and models with endogeneity (but no fixed effects). the commands xtqreg and mmqreg implement the fixed effects estimator, and ivqreg2 implements the estimator for models with endogeneity. The command sivqr that Fernando mentions also estimates modes with endogeneity but uses a very different approach (I would not call it a method of moments estimator). We may be able to help more if you tell us more about what you are doing.

Best wishes,

Joao

[Charles Saba]

@ Prof Joao Santos Silva thanks so much for your contribution. It has helped.
@FernandoRios thanks also.

[Gabriel Cruz]

Dear FernandoRios,
I have a question regarding "mmqreg".
I am adding an individual fixed effect to the quantile regression, where the fixed effect is based on two observations per individual. My problem is that the regression only reports the coefficients (see below).
Do you know what might have caused this issue?

mmqreg Puntaje_std nivel2_mindset_p i.asignatura Nota_std i.nivel_expectativas_prof cprof_sexo i.cprof_titulo cprof_exp_colegio cprof_exp_docente , absorb(mrun) q(70) cluster(codigocurso)
MM-qreg Estimator
Number of obs = 585960
Quantile:  70
                                    (Std. err. adjusted for 12,318 clusters in codigocurso)
-------------------------------------------------------------------------------------------
                          |               Robust
              Puntaje_std | Coefficient  std. err.      z    P>|z|     [95% conf. interval]
--------------------------+----------------------------------------------------------------
location                  |
         nivel2_mindset_p |   .0169536          .        .       .            .           .
                          |
               asignatura |
                       2  |  -.0002984          .        .       .            .           .
                       3  |   .0014447          .        .       .            .           .
                       4  |          0  (omitted)
                          |
                 Nota_std |   .1661539          .        .       .            .           .
1.nivel_expectativas_prof |   .0184953          .        .       .            .           .
               cprof_sexo |   .0061708          .        .       .            .           .
                          |
             cprof_titulo |
          Escuela Normal  |   .0304696          .        .       .            .           .
             Universidad  |   .0386623          .        .       .            .           .
                IP o CFT  |   .0168532          .        .       .            .           .
                          |
        cprof_exp_colegio |   .0010298          .        .       .            .           .
        cprof_exp_docente |  -.0010683          .        .       .            .           .
                    _cons |  -.0688461          .        .       .            .           .
--------------------------+----------------------------------------------------------------
scale                     |
         nivel2_mindset_p |          0  (omitted)
                          |
               asignatura |
                       2  |          0  (omitted)
                       3  |          0  (omitted)
                       4  |          0  (omitted)
                          |
                 Nota_std |          0  (omitted)
1.nivel_expectativas_prof |          0  (omitted)
               cprof_sexo |          0  (omitted)
                          |
             cprof_titulo |
          Escuela Normal  |          0  (omitted)
             Universidad  |          0  (omitted)
                IP o CFT  |          0  (omitted)
                          |
        cprof_exp_colegio |          0  (omitted)
        cprof_exp_docente |          0  (omitted)
                    _cons |          0  (omitted)
--------------------------+----------------------------------------------------------------
qtile                     |
         nivel2_mindset_p |   .0169536          .        .       .            .           .
                          |
               asignatura |
                       2  |  -.0002984          .        .       .            .           .
                       3  |   .0014447          .        .       .            .           .
                       4  |          0  (omitted)
                          |
                 Nota_std |   .1661539          .        .       .            .           .
1.nivel_expectativas_prof |   .0184953          .        .       .            .           .
               cprof_sexo |   .0061708          .        .       .            .           .
                          |
             cprof_titulo |
          Escuela Normal  |   .0304696          .        .       .            .           .
             Universidad  |   .0386623          .        .       .            .           .
                IP o CFT  |   .0168532          .        .       .            .           .
                          |
        cprof_exp_colegio |   .0010298          .        .       .            .           .
        cprof_exp_docente |  -.0010683          .        .       .            .           .
                    _cons |  -.0688461          .        .       .            .           .
-------------------------------------------------------------------------------------------

[Joao Santos Silva]

Dear Gabriel Cruz,

The estimator is only valid for large T, with T = 2, the scale parameters are not identified.

Best wishes,

Joao

[Panagiotis Palaios]

Dear FernandoRios

Could you please explain if there is a command (like predict) in mmqreg to estimate the fitted values for each quantile?

Regards,
P.Palaios

[FernandoRios]

No there isnt. But i may try adding an option for future updates.
F

[Panagiotis Palaios]

FernandoRios
However, if I run "predict" command after estimation I get fitted values. So, if I run one quantile each time, using "nols" as an option,I get the fitted values of that quantile?