Printing one estimation's stat on a table for a different estimation result, in -esttab-

[Nate Breg]

Hello,

I am trying to make one table with the first stage of an IV and a second table with the 2SLS results. The first stage is run using -reg-, and the 2SLS is run with -ivreg2-. The test statistic in question is the Montiel-Pflueger (2013) effective F statistic, which is obtained by running the user-contributed -weakivtest- (which can be installed from SSC) after -ivreg2-, and calling the stored value r(F_eff). I am storing results and exporting Latex tables of the regression results with user-contributed commands -eststo- and -esttab- (available in -estout- from SSC).

I would like to obtain the effective F stat from -weakivtest- after running the 2SLS in -ivreg2- and then print that effective F stat on the separate table I have for the first stage. How can I do this?

Here is my attempt using the cars data as a toy example. This code results in printing the Effective F on the table for 2SLS, correctly, but it prints a blank space in the row for the Effective F stat on the table for the first stage, rather than printing the correct Effective F value:

 sysuse auto
ssc install estout
ssc install weakivtest
 
*first stage reg price weight, r
    eststo m1
 
* 2SLS ivreg2 displacement (price=weight), r
  eststo m2
    weakivtest
  quietly estadd scalar weakf=r(F_eff), replace
 
*save table for 2SLS esttab m2 using "cars_weakiv_2sls", ///
      replace booktabs label se star(* 0.10 ** 0.05 *** 0.01) ///
      s(weakf, ///
      label("Effective F")) ///
      substitute(_ \_ )
 
ivreg2 displacement (price=weight), r
weakivtest
  est restore m2 // bring back the first stage entry
  quietly estadd scalar weakf=r(F_eff), replace
    esttab m1 using "cars_weakiv_ols", ///
      replace booktabs label se star(* 0.10 ** 0.05 *** 0.01) ///
      s(weakf, ///
      label("Effective F")) ///
      substitute(_ \_ )

Thanks in advance for your help!

[Andrew Musau]

 sysuse auto
*ssc install estout
*ssc install weakivtest
 

 
ivreg2 displacement (price=weight), r
  eststo m2
    weakivtest
  quietly estadd scalar weakf=r(F_eff), replace
  local weakf=r(F_eff)
 
*save table for 2SLS esttab m2 using "cars_weakiv_2sls", ///
      replace booktabs label se star(* 0.10 ** 0.05 *** 0.01) ///
      s(weakf, ///
      label("Effective F")) ///
      substitute(_ \_ )
 
   
eststo m1: reg price weight, r
  quietly estadd scalar weakf=`weakf'
    esttab m1, ///
      replace  label se star(* 0.10 ** 0.05 *** 0.01) ///
      s(weakf, ///
      label("Effective F")) ///
      substitute(_ \_ )

You do not need to run the first-stage regression separately. See #9 of the linked thread below.

https://www.statalist.org/forums/for...eg2-by-outreg2

[Nate Breg]

That's a clever solution, thanks very much, Andrew! And thanks for pointing me to that feature for saving first-stage results as part of the 2SLS estimation -- that's handy.