The syntax of reghdfe

[FitzGerald Blindman]

Hi all.

I use the next command to estimate what called in the economic literature the "child penalty", following Kleven, Landais and Sogaard (AEJ: Applied Economics, 2019):

reg wage ib(6).year_relative_to_first_child_birth_adj i.year i.age, robust

But i heard the the command

reghdfe

is better to use since it's standard errors are more accurate. What id the parallel syntax that I should use?

Thanks,

Fitzgerald

[George Ford]

reghdfe wage ib(6).year_relative_to_first_child_birth_adj , absorb(year age) //cluster(x) or vce(robust)

[Carlo Lazzaro]

I fear that the community-contributed module -reghdfe- does not support -robust- standard errors:

sysuse auto
. reghdfe price weight length, absorb(rep78) robust
options robust not allowed
r(198);

. reghdfe price weight length, absorb(rep78) vce(cluster rep78)
(MWFE estimator converged in 1 iterations)

HDFE Linear regression                            Number of obs   =         69
Absorbing 1 HDFE group                            F(   2,      4) =      21.65
Statistics robust to heteroskedasticity           Prob > F        =     0.0072
                                                  R-squared       =     0.4341
                                                  Adj R-squared   =     0.3793
                                                  Within R-sq.    =     0.4258
Number of clusters (rep78)   =          5         Root MSE        =  2294.5106

                                  (Std. err. adjusted for 5 clusters in rep78)
------------------------------------------------------------------------------
             |               Robust
       price | Coefficient  std. err.      t    P>|t|     [95% conf. interval]
-------------+----------------------------------------------------------------
      weight |   5.478309   .8536918     6.42   0.003     3.108081    7.848538
      length |  -109.5065   25.31469    -4.33   0.012    -179.7914   -39.22167
       _cons |   10154.62   4209.396     2.41   0.073     -1532.54    21841.77
------------------------------------------------------------------------------

Absorbed degrees of freedom:
-----------------------------------------------------+
 Absorbed FE | Categories  - Redundant  = Num. Coefs |
-------------+---------------------------------------|
       rep78 |         5           5           0    *|
-----------------------------------------------------+
* = FE nested within cluster; treated as redundant for DoF computation

.

[Andrew Musau]

I fear that the community-contributed module -reghdfe- does not support -robust- standard errors:

Carlo Lazzaro, for Huber-White standard errors, you want the option -vce(robust)-. Alternatively, you could cluster at the observation level. Consider:

sysuse auto, clear
areg price weight length, absorb(rep78) robust
reghdfe price weight length, absorb(rep78) vce(robust)
reghdfe price weight length, absorb(rep78) vce(cluster make)

Res.:

. areg price weight length, absorb(rep78) robust

Linear regression, absorbing indicators          Number of obs     =        69
Absorbed variable: rep78                         No. of categories =         5
                                                 F(2, 62)          =     17.75
                                                 Prob > F          =    0.0000
                                                 R-squared         =    0.4341
                                                 Adj R-squared     =    0.3793
                                                 Root MSE          = 2294.5106

------------------------------------------------------------------------------
             |               Robust
       price | Coefficient  std. err.      t    P>|t|     [95% conf. interval]
-------------+----------------------------------------------------------------
      weight |   5.478309   1.903131     2.88   0.005     1.674004    9.282614
      length |  -109.5065   60.82641    -1.80   0.077    -231.0968    12.08375
       _cons |   10154.62   6010.797     1.69   0.096    -1860.795    22170.03
------------------------------------------------------------------------------

. 
. reghdfe price weight length, absorb(rep78) vce(robust)
(MWFE estimator converged in 1 iterations)

HDFE Linear regression                            Number of obs   =         69
Absorbing 1 HDFE group                            F(   2,     62) =      17.75
                                                  Prob > F        =     0.0000
                                                  R-squared       =     0.4341
                                                  Adj R-squared   =     0.3793
                                                  Within R-sq.    =     0.4258
                                                  Root MSE        =  2294.5106

------------------------------------------------------------------------------
             |               Robust
       price | Coefficient  std. err.      t    P>|t|     [95% conf. interval]
-------------+----------------------------------------------------------------
      weight |   5.478309   1.903131     2.88   0.005     1.674004    9.282614
      length |  -109.5065   60.82641    -1.80   0.077    -231.0968    12.08375
       _cons |   10154.62   6010.797     1.69   0.096    -1860.795    22170.03
------------------------------------------------------------------------------

Absorbed degrees of freedom:
-----------------------------------------------------+
 Absorbed FE | Categories  - Redundant  = Num. Coefs |
-------------+---------------------------------------|
       rep78 |         5           0           5     |
-----------------------------------------------------+

. 
. reghdfe price weight length, absorb(rep78) vce(cluster make)
(MWFE estimator converged in 1 iterations)

HDFE Linear regression                            Number of obs   =         69
Absorbing 1 HDFE group                            F(   2,     62) =      17.75
Statistics robust to heteroskedasticity           Prob > F        =     0.0000
                                                  R-squared       =     0.4341
                                                  Adj R-squared   =     0.3793
                                                  Within R-sq.    =     0.4258
Number of clusters (make)    =         69         Root MSE        =  2294.5106

                                  (Std. err. adjusted for 69 clusters in make)
------------------------------------------------------------------------------
             |               Robust
       price | Coefficient  std. err.      t    P>|t|     [95% conf. interval]
-------------+----------------------------------------------------------------
      weight |   5.478309   1.903131     2.88   0.005     1.674004    9.282614
      length |  -109.5065   60.82641    -1.80   0.077    -231.0968    12.08375
       _cons |   10154.62   6010.797     1.69   0.096    -1860.795    22170.03
------------------------------------------------------------------------------

Absorbed degrees of freedom:
-----------------------------------------------------+
 Absorbed FE | Categories  - Redundant  = Num. Coefs |
-------------+---------------------------------------|
       rep78 |         5           0           5     |
-----------------------------------------------------+

.

[Carlo Lazzaro]

Thanks, Andrew!