Dynamic panel quantile regression with fixed effects

[Ritika Khurana]

Hi all,

I am studying the impact of quality of institutions on carbon emissions. I would like to know the level of carbon emissions at different levels of carbon emissions conditional on entire distribution. For this, the best choice of the model would be dynamic panel quantile regression with fixed effects.

My dataset runs for 34 years from 1984 to 2016 for 130 countries.

I read this thread https://www.statalist.org/forums/for...mic-panel-data and found out that the most appropriate methodology would be the one introduced by Dr. Antonio Galvao. (Link to the paper).

https://reader.elsevier.com/reader/sd/pii/S0304407611000443?token=5D37920192D147349496A998B6 D773E4BA81024B50EA05316BA134B1999416B9EE31D7E1ABC0 C8578DE107E33A934526

I want to know if

qregpd

is the right command for this?

If no, then which is the right command for methodology proposed by the paper mentioned above.

Thanks,

Ritika

[Joao Santos Silva]

Dear Ritika Khurana,

No, that is not the code for it; I do not think the Galvão estimator is implemented in Stata, but I may be wrong.

Best wishes,

Joao

[Ritika Khurana]

Dr. Joao Santos Silva,

Thank you for your reply.

My question now pertains to

qregpd

. How do we decide we need MCMC optimization or Nelder-Mead optimization for estimating vector b.
For more clarity, the variables are as follows:
dlrgdp = Log of real per capita GDP, dlrgdp2 = squared dlrgdp, dfrleg = institutional integrity, dfrleg2 = squared dfrleg, dlpopden = log of population density, dlenguse = log of energy use, and GDP, dlpccarb = log of per capita carbon emissions

'd' in front of every variable is to show first-difference as T=30.

 qregpd dlpccarb dlrgdp dlrgdp2 dlenguse dlpopden dfrleg dfrleg2 , quantile(0.25) id(cty) fix(year) optimize(mcmc) noisy draws(1000) burn(100) arate(.5)

Adaptive MCMC optimization
...edited...

Quantile Regression for Panel Data (QRPD)
     Number of obs:              3231
     Number of groups:            120
     Min obs per group:             3
     Max obs per group:            30
------------------------------------------------------------------------------
    dlpccarb |      Coef.   Std. Err.      z    P>|z|     [95% Conf. Interval]
-------------+----------------------------------------------------------------
      dlrgdp |   .7053674    .017316    40.73   0.000     .6714286    .7393062
     dlrgdp2 |  -.0251297   .0010471   -24.00   0.000     -.027182   -.0230774
    dlenguse |   .8264385   .0027036   305.68   0.000     .8211395    .8317375
    dlpopden |  -.3522818   .0054659   -64.45   0.000    -.3629948   -.3415688
      dfrleg |  -.5466352   .0039406  -138.72   0.000    -.5543586   -.5389118
     dfrleg2 |   .3849055   .0050248    76.60   0.000      .375057     .394754
------------------------------------------------------------------------------
No excluded instruments - standard QRPD estimation.


MCMC diagonstics:
     Mean acceptance rate:      0.304
     Total draws:                1000
     Burn-in draws:               100
     Draws retained:              900
     Value of objective function:   
             Mean:             -77.0044
             Min:              -82.6782
             Max:              -74.2085
MCMC notes:
     *Point estimates correspond to mean of draws.
     *Standard errors are derived from variance of draws.

 

. qregpd dlpccarb dlrgdp dlrgdp2 dlenguse dlpopden dfrleg dfrleg2 , quantile(0.25) id(cty) fix(year)
Nelder-Mead optimization
...edited...

Quantile Regression for Panel Data (QRPD)
     Number of obs:              3231
     Number of groups:            120
     Min obs per group:             3
     Max obs per group:            30
------------------------------------------------------------------------------
    dlpccarb |      Coef.   Std. Err.      z    P>|z|     [95% Conf. Interval]
-------------+----------------------------------------------------------------
      dlrgdp |   .8300392   .5573196     1.49   0.136    -.2622871    1.922365
     dlrgdp2 |  -.0337071   .0289122    -1.17   0.244     -.090374    .0229598
    dlenguse |   .8550453   .0732594    11.67   0.000     .7114594    .9986311
    dlpopden |   .2359039    .125433     1.88   0.060    -.0099403    .4817481
      dfrleg |  -.6102081   .2429038    -2.51   0.012    -1.086291   -.1341254
     dfrleg2 |   .4625645    .204273     2.26   0.024     .0621968    .8629321
------------------------------------------------------------------------------
No excluded instruments - standard QRPD estimation.

Looking forward to hearing from you.

Thanks,

Ritika

[Joao Santos Silva]

Sorry, I am not the author of this command I cannot comment on it; maybe Matthew J. Baker can help. In any case, please note that qregpd does not estimate a model with fixed effects.

[Ritika Khurana]

Dr. Joao Santos Silva thank you so much for your reply.

Dr. Matthew J. Baker it will be of great help if you could help me with the difference between the two.

Thank you.

Ritika