robust standard error for IV panel regression

[Saleh-albrmaki Mousa]

Hi
I have been running panel data random effect 2sls (G2SLS) and i want to robust to heteroskedasticity and autocorrelation any advice on how to perform it in Stata as i know we robust to heteroskedasticity and autocorrelation using cluster(id) in panel fixed or random effect but this choice is not available for G2sls

[Joro Kolev]

You have to be more specific regarding which command you are using, and which Stata you are using.

What you are saying might have been true in old Stata like Stata 7 say.

But as of now in Stata 15,

xtivreg, re

does accept robust and cluster options.

[Joro Kolev]

Here:

.  webuse nlswork, clear
(National Longitudinal Survey.  Young Women 14-26 years of age in 1968)

. xtivreg ln_w age c.age#c.age not_smsa 2.race (tenure = union birth south), re vce(cluster idcode)

G2SLS random-effects IV regression              Number of obs     =     19,007
Group variable: idcode                          Number of groups  =      4,134

R-sq:                                           Obs per group:
     within  = 0.0664                                         min =          1
     between = 0.2098                                         avg =        4.6
     overall = 0.1463                                         max =         12


                                                Wald chi2(5)      =     643.00
corr(u_i, X)       = 0 (assumed)                Prob > chi2       =     0.0000

                             (Std. Err. adjusted for 4,711 clusters in idcode)
------------------------------------------------------------------------------
             |               Robust
     ln_wage |      Coef.   Std. Err.      z    P>|z|     [95% Conf. Interval]
-------------+----------------------------------------------------------------
      tenure |   .1391798   .0104381    13.33   0.000     .1187216     .159638
         age |   .0279649   .0080686     3.47   0.001     .0121507     .043779
             |
 c.age#c.age |  -.0008357   .0001415    -5.91   0.000     -.001113   -.0005584
             |
    not_smsa |  -.2235103   .0146421   -15.26   0.000    -.2522084   -.1948123
             |
        race |
      black  |  -.2078613   .0165504   -12.56   0.000    -.2402995   -.1754231
       _cons |   1.337684   .1173549    11.40   0.000     1.107673    1.567695
-------------+----------------------------------------------------------------
     sigma_u |  .36582493
     sigma_e |  .63031479
         rho |  .25197078   (fraction of variance due to u_i)
------------------------------------------------------------------------------
Instrumented:   tenure
Instruments:    age c.age#c.age not_smsa 2.race union birth_yr south
------------------------------------------------------------------------------

.

[Saleh-albrmaki Mousa]

dear Joro Kolev thank you for responding i'm using Stata 13 and the command is xtivreg y x1 x2 x3 (z= I1 I2), re

[Joro Kolev]

dear Joro Kolev thank you for responding i'm using Stata 13 and the command is xtivreg y x1 x2 x3 (z= I1 I2), re

OK, Stata 13 does not allow the -xtivreg, re vce(cluster)- option.

Just bootstrap the standard errors, like that:

. webuse nlswork, clear
(National Longitudinal Survey.  Young Women 14-26 years of age in 1968)

. xtivreg ln_w age c.age#c.age not_smsa 2.race (tenure = union birth south), re vce(bootstrap, reps(1000) nodots)

G2SLS random-effects IV regression              Number of obs      =     19007
Group variable: idcode                          Number of groups   =      4134

R-sq:  within  = 0.0664                         Obs per group: min =         1
       between = 0.2098                                        avg =       4.6
       overall = 0.1463                                        max =        12


                                                Wald chi2(5)       =    277.74
corr(u_i, X)       = 0 (assumed)                Prob > chi2        =    0.0000

                               (Replications based on 4134 clusters in idcode)
------------------------------------------------------------------------------
             |   Observed   Bootstrap                         Normal-based
     ln_wage |      Coef.   Std. Err.      z    P>|z|     [95% Conf. Interval]
-------------+----------------------------------------------------------------
      tenure |   .1391798   .0114555    12.15   0.000     .1167274    .1616323
         age |   .0279649   .0079696     3.51   0.000     .0123447    .0435851
             |
 c.age#c.age |  -.0008357   .0001417    -5.90   0.000    -.0011134   -.0005579
             |
    not_smsa |  -.2235103   .0233905    -9.56   0.000    -.2693548   -.1776659
             |
        race |
      black  |  -.2078613   .0169946   -12.23   0.000      -.24117   -.1745526
       _cons |   1.337684   .1235172    10.83   0.000     1.095595    1.579773
-------------+----------------------------------------------------------------
     sigma_u |  .36582493
     sigma_e |  .63031479
         rho |  .25197078   (fraction of variance due to u_i)
------------------------------------------------------------------------------
Instrumented:   tenure
Instruments:    age c.age#c.age not_smsa 2.race union birth_yr south
------------------------------------------------------------------------------

Be warned that this might take some time, depending on how large and complicated is your dataset.

[Carlo Lazzaro]

dear Joro Kolev thank you for responding i'm using Stata 13 and the command is xtivreg y x1 x2 x3 (z= I1 I2), re

Please note https://www.statalist.org/forums/help#version. Thanks.
Joro's helpful advice could have reached you earlier had you specified from your first post of this thread that you are using a (quite) out-of-date Stata release.