ivreg2 controls for second stage only

[Ivan Popov]

Dear all,

I want to run a iv-2sls using ivreg2, and i have 3 sets of covariates: intended for first stage only, for second stage only, and for both stages(fixed effect). As far as i'm concerned, ivreg2 allows to include controls only in both stages. Is there a way to go around that issue? My outcome variable is z_el_maths, Lunch is covariate for second stage, fully_vaccinated is endogenous regressor, Lottery and Celebrity are instruments, teaching_exp is covariate for first stage, and School2 School3 is covariate for both stages

I run the following command:

ivreg2 z_el_maths Lunch (fully_vaccinated = Lottery Celebrity teaching_exp) School2 School3, cluster(Sr)


------------------------------------------------------------------------------
             |               Robust
fully_vacc~d |      Coef.   Std. Err.      t    P>|t|     [95% Conf. Interval]
-------------+----------------------------------------------------------------
     Lottery |  -.0878575   .0525038    -1.67   0.094     -.190772    .0150571
   Celebrity |  -.0627029   .0528724    -1.19   0.236    -.1663399    .0409341
teaching_exp |  -.0029099   .0063328    -0.46   0.646     -.015323    .0095032
       Lunch |   .0094199   .0078472     1.20   0.230    -.0059617    .0248016
     School2 |   .0209415   .1934575     0.11   0.914    -.3582613    .4001442
     School3 |  -.0271421   .1076267    -0.25   0.801    -.2381049    .1838206
       _cons |   .3742611   .0381852     9.80   0.000      .299413    .4491091
------------------------------------------------------------------------------

...


----------------------------------------------------------------------------------
                 |               Robust
      z_el_maths |      Coef.   Std. Err.      z    P>|z|     [95% Conf. Interval]
-----------------+----------------------------------------------------------------
fully_vaccinated |   .5364026   .3611458     1.49   0.137    -.1714302    1.244235
           Lunch |  -.0123833   .0174608    -0.71   0.478     -.046606    .0218393
         School2 |  -.0106998   .1235372    -0.09   0.931    -.2528283    .2314287
         School3 |  -.0656685   .1268587    -0.52   0.605     -.314307    .1829699
           _cons |  -.1736665   .1205772    -1.44   0.150    -.4099934    .0626603
----------------------------------------------------------------------------------

And Lunch is included in first stage(and i don't want that)

[Nils Enevoldsen]

Even if you think you want to do this, you probably don't.

See here for an explanation, along with a workaround: https://www.stata.com/support/faqs/s...es-regression/

[Ivan Popov]

thanks!