Quantile regression

[Lucia Gracia]

Dear Joao Santos Silva ,

I have already asked about this, but I still having problems with the regression. So, I would appreciate your help. I createad this quantile regression following your advice, but I have been told that it is not correct.

xtset cnt_enc

xtqreg ln_manEMP lnPop lnPop_sq lngdppc lngdppc_sq fdi ln_fdi remitt ln_remitt foreignaid ln_foreignaid i.year, ls q(0.25, 0.5, 0.75)

Can you help me? You also advice me to add the bootsrap command, but I do not know how to do it.

Thanks in advance,
Best
Lucia

[Joao Santos Silva]

Dear Lucia Gracia,

Why is it not correct? On the bootstrap, please check the example here (you can ignore the bit about c>=10).

Best wishes,

Joao

[Lucia Gracia]

Thank yoiu Joao Santos Silva,


My professor told me the tables I obtained were wrong. I have a panel data of 31 countries and 29 years, 899 observations in total. I want to add fixed effects for countries and years.

I added the bootstrap like in your example:

xtset cnt_enc year

bs, cl(id) r(50) id(cnt_enc): xtqreg manEMP lnPop lnPop_sq lngdppc lngdppc_sq fdi ln_fdi remitt ln_remitt foreignaid ln_foreignaid i.year i.cnt_enc, q(0.25, 0.5, 0.75)

but, I got this:

WARNING: 1.89099% of the fitted values of the scale function are not positive

Bootstrap replications (50): xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx done
x: Error occurred when bootstrap executed xtqreg.
insufficient observations to compute bootstrap standard errors
no results will be saved
r(2000);

I do not know what I can do, can you help me?

Best,
Lucía

[Joao Santos Silva]

Dear Lucia Gracia,

Do not worry about the warning. You are not using the command as in the example.
1) xtset only by cnt_enc
2) Keep in mind that xtqreg automatically included the cnt_enc fixed effects, so to not include these in the model.
3) Make sure you the bootstrap as in the example.

Best wishes,

Joao

[Lucia Gracia]

Dear Joao Santos Silva ,

Thanks for your help, it worked. I just have one last question if you do not mind.

My results with a previous linear regression were mostly significant, but with this quantile regression I obtain insignificant results. Why could be so?

Best regards,

Lucía

[Joao Santos Silva]

Your earlier regression probably did not include fixed effects?

[Lucia Gracia]

I also included year and country fixed effects for the previous regression.

[Joao Santos Silva]

Use the option ls in xtqreg and see if the results match. If the estimates are different, the models are different, If it is only the SE that are different, maybe you did not use robust standard errors? If this does not answer your question, please post the two sets of results (using the right delimiters).

[Lucia Gracia]

Thanks!

I did this at the end. I also wanted to use the quantile regression for subdivisions of my sample, but I think xtqreg is not useful for less than 10 countries sample, right?

. eststo s1: bs, cluster(cnt_enc) rep(500): xtqreg manEMP lnPop lnPop_sq lngdppc 
> lngdppc_sq fdi1 remitt foreignaid1 i.year, ls q(0.25)
(running xtqreg on estimation sample)
                                                           Location parameters
                                                              Scale parameters
WARNING: 2.3359288% of the fitted values of the scale function are not positive

Bootstrap replications (500): .........10.........20.........30.........40.......
> ..50.........60.........70.........80.........90.........100.........110.......
> ..120.........130.........140.........150.........160.........170.........180..
> .......190.........200.........210.........220.........230.........240.........
> 250.........260.........270.........280.........290.........300.........310....
> .....320.........330.........340.........350.........360.........370.........38
> 0.........390.........400.........410.........420.........430.........440......
> ...450.........460.........470.........480.........490.........500 done

Bootstrap results                                          Number of obs = 899
                                                           Replications  = 500

                                (Replications based on 31 clusters in cnt_enc)
------------------------------------------------------------------------------
             |   Observed   Bootstrap                         Normal-based
      manEMP | coefficient  std. err.      z    P>|z|     [95% conf. interval]
-------------+----------------------------------------------------------------
       lnPop |  -.0927677   .1596205    -0.58   0.561     -.405618    .2200827
    lnPop_sq |   .0031597   .0046362     0.68   0.496     -.005927    .0122464
     lngdppc |   .3247265    .089008     3.65   0.000      .150274     .499179
  lngdppc_sq |   -.020143   .0056077    -3.59   0.000    -.0311339   -.0091521
        fdi1 |  -.1140955   .0666219    -1.71   0.087    -.2446719    .0164809
      remitt |  -.0251568   .0924306    -0.27   0.785    -.2063175    .1560038
 foreignaid1 |   -.182364   .1038098    -1.76   0.079    -.3858274    .0210994
             |
        year |
       1991  |   .0024906    .002174     1.15   0.252    -.0017704    .0067516
       1992  |   .0059991   .0039114     1.53   0.125    -.0016671    .0136652
       1993  |   .0075807   .0052863     1.43   0.152    -.0027802    .0179416
       1994  |   .0116205   .0069138     1.68   0.093    -.0019303    .0251713
       1995  |   .0111865   .0081258     1.38   0.169    -.0047397    .0271128
       1996  |   .0123331   .0090359     1.36   0.172     -.005377    .0300432
       1997  |   .0124426   .0097484     1.28   0.202     -.006664    .0315492
       1998  |    .009756   .0104444     0.93   0.350    -.0107147    .0302266
       1999  |   .0087197   .0113886     0.77   0.444    -.0136016     .031041
       2000  |   .0093158   .0125195     0.74   0.457     -.015222    .0338537
       2001  |   .0105706   .0134823     0.78   0.433    -.0158541    .0369954
       2002  |   .0114262   .0146315     0.78   0.435     -.017251    .0401034
       2003  |    .010635   .0153861     0.69   0.489    -.0195212    .0407913
       2004  |   .0139318   .0161469     0.86   0.388    -.0177155    .0455791
       2005  |   .0148417   .0172553     0.86   0.390    -.0189782    .0486615
       2006  |   .0170212    .018455     0.92   0.356    -.0191499    .0531923
       2007  |   .0157797   .0192936     0.82   0.413     -.022035    .0535945
       2008  |   .0142864   .0201606     0.71   0.479    -.0252277    .0538004
       2009  |   .0092623   .0209186     0.44   0.658    -.0317375     .050262
       2010  |   .0111292   .0219996     0.51   0.613    -.0319891    .0542475
       2011  |   .0137383    .023124     0.59   0.552     -.031584    .0590605
       2012  |    .013993   .0243992     0.57   0.566    -.0338285    .0618144
       2013  |    .015331   .0255658     0.60   0.549     -.034777     .065439
       2014  |   .0143813    .026027     0.55   0.581    -.0366307    .0653932
       2015  |   .0186217   .0266522     0.70   0.485    -.0336156    .0708591
       2016  |   .0174581   .0270788     0.64   0.519    -.0356153    .0705315
       2017  |   .0192564   .0278411     0.69   0.489     -.035311    .0738239
       2018  |   .0195152    .028845     0.68   0.499    -.0370199    .0760504
------------------------------------------------------------------------------

. 
. eststo s11: bs, cluster(cnt_enc) rep(500): xtqreg manEMP lnPop lnPop_sq lngdppc
>  lngdppc_sq fdi1 remitt foreignaid1 i.year, ls q(0.5)
(running xtqreg on estimation sample)
                                                           Location parameters
                                                              Scale parameters
WARNING: 2.3359288% of the fitted values of the scale function are not positive

Bootstrap replications (500): .........10.........20.........30.........40.......
> ..50.........60.........70.........80.........90.........100.........110.......
> ..120.........130.........140.........150.........160.........170.........180..
> .......190.........200.........210.........220.........230.........240.........
> 250.........260.........270.........280.........290.........300.........310....
> .....320.........330.........340.........350.........360.........370.........38
> 0.........390.........400.........410.........420.........430.........440......
> ...450.........460.........470.........480.........490.........500 done

Bootstrap results                                          Number of obs = 899
                                                           Replications  = 500

                                (Replications based on 31 clusters in cnt_enc)
------------------------------------------------------------------------------
             |   Observed   Bootstrap                         Normal-based
      manEMP | coefficient  std. err.      z    P>|z|     [95% conf. interval]
-------------+----------------------------------------------------------------
       lnPop |  -.0455358   .1646567    -0.28   0.782     -.368257    .2771853
    lnPop_sq |   .0016424    .004917     0.33   0.738    -.0079946    .0112795
     lngdppc |   .3542114   .0873258     4.06   0.000      .183056    .5253668
  lngdppc_sq |  -.0223611   .0055757    -4.01   0.000    -.0332893   -.0114328
        fdi1 |  -.0858002   .0595399    -1.44   0.150    -.2024962    .0308959
      remitt |  -.0303432    .086144    -0.35   0.725    -.1991824     .138496
 foreignaid1 |  -.1865193   .0874872    -2.13   0.033     -.357991   -.0150476
             |
        year |
       1991  |   .0006587   .0017966     0.37   0.714    -.0028626      .00418
       1992  |   .0013588   .0030184     0.45   0.653    -.0045572    .0072748
       1993  |   .0006869   .0042751     0.16   0.872    -.0076922     .009066
       1994  |   .0031872   .0058454     0.55   0.586    -.0082696    .0146441
       1995  |   .0016991   .0072799     0.23   0.815    -.0125692    .0159675
       1996  |   .0025381   .0085415     0.30   0.766    -.0142028    .0192791
       1997  |   .0030955   .0093016     0.33   0.739    -.0151352    .0213262
       1998  |   .0008436   .0099971     0.08   0.933    -.0187504    .0204375
       1999  |  -.0000759   .0109475    -0.01   0.994    -.0215327    .0213809
       2000  |   .0003432   .0120093     0.03   0.977    -.0231946    .0238809
       2001  |   .0005197   .0128075     0.04   0.968    -.0245824    .0256219
       2002  |   .0010178   .0138981     0.07   0.942    -.0262219    .0282575
       2003  |   .0008507   .0148075     0.06   0.954    -.0281714    .0298728
       2004  |   .0036356   .0158792     0.23   0.819     -.027487    .0347582
       2005  |   .0053226   .0170383     0.31   0.755    -.0280718     .038717
       2006  |   .0081196   .0182353     0.45   0.656    -.0276209    .0438601
       2007  |   .0082401   .0190837     0.43   0.666    -.0291633    .0456435
       2008  |   .0079327   .0199619     0.40   0.691    -.0311919    .0470573
       2009  |   .0034394   .0208204     0.17   0.869    -.0373679    .0442467
       2010  |   .0058616   .0219328     0.27   0.789    -.0371259    .0488491
       2011  |   .0089516   .0230901     0.39   0.698    -.0363041    .0542072
       2012  |   .0111877   .0244464     0.46   0.647    -.0367264    .0591019
       2013  |   .0143546   .0255866     0.56   0.575    -.0357942    .0645034
       2014  |   .0144421   .0265047     0.54   0.586    -.0375062    .0663903
       2015  |   .0173116    .027401     0.63   0.528    -.0363934    .0710166
       2016  |   .0163881    .028212     0.58   0.561    -.0389065    .0716827
       2017  |   .0184948   .0292225     0.63   0.527    -.0387803    .0757699
       2018  |   .0197102   .0301877     0.65   0.514    -.0394566     .078877
------------------------------------------------------------------------------

. 
. eststo s111: bs, cluster(cnt_enc) rep(500): xtqreg manEMP lnPop lnPop_sq lngdpp
> c lngdppc_sq fdi1 remitt foreignaid1 i.year, ls q(0.75)
(running xtqreg on estimation sample)
                                                           Location parameters
                                                              Scale parameters
WARNING: 2.3359288% of the fitted values of the scale function are not positive

Bootstrap replications (500): .........10.........20.........30.........40.......
> ..50.........60.........70.........80.........90.........100.........110.......
> ..120.........130.........140.........150.........160.........170.........180..
> .......190.........200.........210.........220.........230.........240.........
> 250.........260.........270.........280.........290.........300.........310....
> .....320.........330.........340.........350.........360.........370.........38
> 0.........390.........400.........410.........420.........430.........440......
> ...450.........460.........470.........480.........490.........500 done

Bootstrap results                                          Number of obs = 899
                                                           Replications  = 500

                                (Replications based on 31 clusters in cnt_enc)
------------------------------------------------------------------------------
             |   Observed   Bootstrap                         Normal-based
      manEMP | coefficient  std. err.      z    P>|z|     [95% conf. interval]
-------------+----------------------------------------------------------------
       lnPop |   .0112382   .1800316     0.06   0.950    -.3416172    .3640936
    lnPop_sq |  -.0001813   .0053281    -0.03   0.973    -.0106242    .0102615
     lngdppc |   .3896532   .0881091     4.42   0.000     .2169624    .5623439
  lngdppc_sq |  -.0250272   .0056318    -4.44   0.000    -.0360653   -.0139891
        fdi1 |  -.0517883   .0587578    -0.88   0.378    -.1669515    .0633749
      remitt |  -.0365773   .0681632    -0.54   0.592    -.1701747    .0970201
 foreignaid1 |   -.191514   .0829704    -2.31   0.021     -.354133   -.0288951
             |
        year |
       1991  |  -.0015433   .0020817    -0.74   0.458    -.0056233    .0025367
       1992  |  -.0042189   .0028015    -1.51   0.132    -.0097097    .0012718
       1993  |  -.0075997   .0038342    -1.98   0.047    -.0151146   -.0000847
       1994  |  -.0069498   .0053089    -1.31   0.191     -.017355    .0034554
       1995  |   -.009705   .0066223    -1.47   0.143    -.0226845    .0032746
       1996  |  -.0092357   .0080229    -1.15   0.250    -.0249603    .0064889
       1997  |    -.00814   .0090607    -0.90   0.369    -.0258986    .0096185
       1998  |  -.0098694    .009663    -1.02   0.307    -.0288085    .0090697
       1999  |  -.0106484   .0108337    -0.98   0.326    -.0318821    .0105852
       2000  |  -.0104423   .0117156    -0.89   0.373    -.0334044    .0125199
       2001  |  -.0115618   .0125127    -0.92   0.355    -.0360862    .0129627
       2002  |  -.0114934   .0133976    -0.86   0.391    -.0377522    .0147654
       2003  |  -.0109104   .0147949    -0.74   0.461     -.039908    .0180871
       2004  |  -.0087407   .0161488    -0.54   0.588    -.0403918    .0229103
       2005  |  -.0061197   .0173752    -0.35   0.725    -.0401745    .0279351
       2006  |  -.0025804    .018688    -0.14   0.890    -.0392081    .0340474
       2007  |  -.0008227   .0196497    -0.04   0.967    -.0393354      .03769
       2008  |   .0002953   .0206698     0.01   0.989    -.0402167    .0408073
       2009  |  -.0035598   .0218082    -0.16   0.870    -.0463031    .0391835
       2010  |  -.0004702   .0230172    -0.02   0.984    -.0455832    .0446427
       2011  |   .0031977    .024459     0.13   0.896     -.044741    .0511365
       2012  |   .0078157   .0261992     0.30   0.765    -.0435338    .0591652
       2013  |   .0131809   .0276254     0.48   0.633    -.0409638    .0673255
       2014  |   .0145151   .0289914     0.50   0.617     -.042307    .0713373
       2015  |   .0157368    .030151     0.52   0.602    -.0433581    .0748317
       2016  |   .0151019   .0315415     0.48   0.632    -.0467183    .0769221
       2017  |   .0175793   .0328248     0.54   0.592    -.0467561    .0819148
       2018  |   .0199446    .033791     0.59   0.555    -.0462845    .0861736
------------------------------------------------------------------------------

. 
. 
. 
. esttab s1 s11 s111, se starlevels(* 0.10 ** 0.05 *** 0.01) stat(N r2, fmt(0 3))
> , using qregDRIV.rtf, keep(lnPop lnPop_sq lngdppc lngdppc_sq fdi1 remitt foreig
> naid1) b(3) se(3) nogaps replace
(output written to qregDRIV.rtf)

Best,
Lucía