ivreghdfe to run 2SLS

[hussein bataineh]

Hi
I'm using ivreghdfe to run 2SLS and i want to present the constant for the first and second stage . In addition to that i need to present R2 for the first stage , please can you advice me on how to do this

ivreghdfe EQUITY FSIZE OP_CF SD_OCF TAX  BM LEV MA Gov_score DPP RPP PSIZE DR  Sustain_Perf Sust_Commit  (CSO =iv1), first absorb(id year) robust  endog(CSO) 
(dropped 7 singleton observations)
(MWFE estimator converged in 7 iterations)

First-stage regressions
-----------------------


First-stage regression of CSO:

Statistics robust to heteroskedasticity
Number of obs =                   4181
------------------------------------------------------------------------------
             |               Robust
         CSO | Coefficient  std. err.      t    P>|t|     [95% conf. interval]
-------------+----------------------------------------------------------------
         iv1 |   .0142274   .0024147     5.89   0.000     .0094931    .0189617
       FSIZE |   .0041056   .0193726     0.21   0.832    -.0338759    .0420872
       OP_CF |  -.2269747   .1568084    -1.45   0.148    -.5344101    .0804607
      SD_OCF |   1.560523   .3035384     5.14   0.000     .9654118    2.155634
         TAX |   .0268227   .0200401     1.34   0.181    -.0124676    .0661129
          BM |  -.0311954   .0317133    -0.98   0.325    -.0933718     .030981
         LEV |   .0703909   .0753115     0.93   0.350    -.0772632    .2180451
          MA |  -.0213085   .0377938    -0.56   0.573    -.0954064    .0527893
   Gov_score |   .0003009   .0003313     0.91   0.364    -.0003487    .0009505
         DPP |  -.0839605   .0563909    -1.49   0.137    -.1945193    .0265983
         RPP |  -.1040564   .0801962    -1.30   0.195    -.2612875    .0531747
       PSIZE |   .0529241   .0176648     3.00   0.003     .0182908    .0875574
          DR |  -.0107933   .0162706    -0.66   0.507    -.0426931    .0211064
Sustain_Perf |   .0013168   .0005055     2.61   0.009     .0003258    .0023078
 Sust_Commit |   4.85e-07   .0159965     0.00   1.000     -.031362     .031363
------------------------------------------------------------------------------
F test of excluded instruments:
  F(  1,  3854) =    34.71
  Prob > F      =   0.0000
Sanderson-Windmeijer multivariate F test of excluded instruments:
  F(  1,  3854) =    34.71
  Prob > F      =   0.0000



Summary results for first-stage regressions
-------------------------------------------

                                           (Underid)            (Weak id)
Variable     | F(  1,  3854)  P-val | SW Chi-sq(  1) P-val | SW F(  1,  3854)
CSO          |      34.71    0.0000 |       37.66   0.0000 |       34.71

NB: first-stage test statistics heteroskedasticity-robust

Stock-Yogo weak ID F test critical values for single endogenous regressor:
                                   10% maximal IV size             16.38
                                   15% maximal IV size              8.96
                                   20% maximal IV size              6.66
                                   25% maximal IV size              5.53
Source: Stock-Yogo (2005).  Reproduced by permission.
NB: Critical values are for i.i.d. errors only.

Underidentification test
Ho: matrix of reduced form coefficients has rank=K1-1 (underidentified)
Ha: matrix has rank=K1 (identified)
Kleibergen-Paap rk LM statistic          Chi-sq(1)=37.01    P-val=0.0000

Weak identification test
Ho: equation is weakly identified
Cragg-Donald Wald F statistic                                      45.72
Kleibergen-Paap Wald rk F statistic                                34.71

Stock-Yogo weak ID test critical values for K1=1 and L1=1:
                                   10% maximal IV size             16.38
                                   15% maximal IV size              8.96
                                   20% maximal IV size              6.66
                                   25% maximal IV size              5.53
Source: Stock-Yogo (2005).  Reproduced by permission.
NB: Critical values are for Cragg-Donald F statistic and i.i.d. errors.

Weak-instrument-robust inference
Tests of joint significance of endogenous regressors B1 in main equation
Ho: B1=0 and orthogonality conditions are valid
Anderson-Rubin Wald test           F(1,3854)=      8.78     P-val=0.0031
Anderson-Rubin Wald test           Chi-sq(1)=      9.53     P-val=0.0020
Stock-Wright LM S statistic        Chi-sq(1)=      9.68     P-val=0.0019

NB: Underidentification, weak identification and weak-identification-robust
    test statistics heteroskedasticity-robust

Number of observations               N  =       4181
Number of regressors                 K  =         15
Number of endogenous regressors      K1 =          1
Number of instruments                L  =         15
Number of excluded instruments       L1 =          1

IV (2SLS) estimation
--------------------

Estimates efficient for homoskedasticity only
Statistics robust to heteroskedasticity

                                                      Number of obs =     4181
                                                      F( 15,  3854) =     6.85
                                                      Prob > F      =   0.0000
Total (centered) SS     =  48.43091059                Centered R2   =  -0.1291
Total (uncentered) SS   =  48.43091059                Uncentered R2 =  -0.1291
Residual SS             =  54.68279858                Root MSE      =    .1191

------------------------------------------------------------------------------
             |               Robust
      EQUITY | Coefficient  std. err.      t    P>|t|     [95% conf. interval]
-------------+----------------------------------------------------------------
         CSO |  -.1804735   .0660308    -2.73   0.006    -.3099321   -.0510149
       FSIZE |   .0124918   .0083606     1.49   0.135    -.0038998    .0288834
       OP_CF |  -.0887216   .0697029    -1.27   0.203    -.2253796    .0479365
      SD_OCF |  -.0893687   .1615149    -0.55   0.580    -.4060316    .2272941
         TAX |  -.0072579    .008027    -0.90   0.366    -.0229956    .0084797
          BM |  -.0253248   .0127788    -1.98   0.048    -.0503787   -.0002709
         LEV |  -.0249488   .0320252    -0.78   0.436    -.0877367    .0378392
          MA |   .0051581     .01475     0.35   0.727    -.0237605    .0340766
   Gov_score |  -.0000326   .0001301    -0.25   0.802    -.0002876    .0002224
         DPP |   .1255691   .0247609     5.07   0.000     .0770233    .1741149
         RPP |   .0912897   .0351963     2.59   0.010     .0222846    .1602948
       PSIZE |   .0218074   .0094218     2.31   0.021     .0033352    .0402796
          DR |   .0375591   .0071679     5.24   0.000     .0235059    .0516124
Sustain_Perf |   3.01e-06   .0002204     0.01   0.989    -.0004292    .0004352
 Sust_Commit |  -.0084441   .0067065    -1.26   0.208    -.0215927    .0047045
------------------------------------------------------------------------------
Underidentification test (Kleibergen-Paap rk LM statistic):             37.015
                                                   Chi-sq(1) P-val =    0.0000
------------------------------------------------------------------------------
Weak identification test (Cragg-Donald Wald F statistic):               45.723
                         (Kleibergen-Paap rk Wald F statistic):         34.715
Stock-Yogo weak ID test critical values: 10% maximal IV size             16.38
                                         15% maximal IV size              8.96
                                         20% maximal IV size              6.66
                                         25% maximal IV size              5.53
Source: Stock-Yogo (2005).  Reproduced by permission.
NB: Critical values are for Cragg-Donald F statistic and i.i.d. errors.
------------------------------------------------------------------------------
Hansen J statistic (overidentification test of all instruments):         0.000
                                                 (equation exactly identified)
-endog- option:
Endogeneity test of endogenous regressors:                               7.418
                                                   Chi-sq(1) P-val =    0.0065
Regressors tested:    CSO
------------------------------------------------------------------------------
Instrumented:         CSO
Included instruments: FSIZE OP_CF SD_OCF TAX BM LEV MA Gov_score DPP RPP PSIZE
                      DR Sustain_Perf Sust_Commit
Excluded instruments: iv1
Partialled-out:       _cons
                      nb: total SS, model F and R2s are after partialling-out;
                          any small-sample adjustments include partialled-out
                          variables in regressor count K
------------------------------------------------------------------------------

Absorbed degrees of freedom:
-----------------------------------------------------+
 Absorbed FE | Categories  - Redundant  = Num. Coefs |
-------------+---------------------------------------|
          id |       294           0         294     |
        year |        19           1          18     |
-----------------------------------------------------+

[George Ford]

The constant is meaningless. It's just the FE for one of the cross sections.

[hussein bataineh]

Thanks, Professor George. Thank you for your response. Yes, I know that it does not have a significant meaning, but still, my supervisor emphasizes presenting it. Actually, I’m not sure if there is any way I can present it.

i try the following command to do 2sls with fixed effect manually

foreach var in "EQUITY" "FSIZE" "OP_CF " " SD_OCF " "TAX " "BM" "LEV" "MA" "DPP" "RPP " "PSIZE" "DR " "Gov_score" "Sustain_Perf" "Sust_Commit " " CSO" {
drop if `var'==.
}
reghdfe CSO iv1 FSIZE OP_CF SD_OCF TAX BM LEV MA DPP RPP PSIZE DR Gov_score Sustain_Perf Sust_Commit, absorb(id year) vce(robust)

predict CSO_hat, xb

reghdfe EQUITY CSO_hat FSIZE OP_CF SD_OCF TAX BM LEV MA DPP RPP PSIZE DR Gov_score Sustain_Perf Sust_Commit , absorb(id year) vce(robust)

Actually for the fist stage it give identical results for the command ivreghdfe

for the second stage l the coefficient is identical but the Std. err. t P>|t| [95% conf. interval] is not the same

Please can you suggest a way to present the constant

Thanks

[George Ford]

When using a control function (predict should be residuals, I think) the coefficients are the same but the SE will not be. You need to bootstrap the SE anyway.

There no way I know of to generate a constant with either ivreg2 and ivreghdfe (which uses ivreg2). That should tell you and your supervisor something.

Sergio (ivreghdfe author) explains here:

https://github.com/sergiocorreia/ivreghdfe/issues/34

I don't know how to get the FE from ivreg2 or ivreghdfe, though reghdfe provides them as an option.

You can get a constant with the control function and reghdfe, but you'll have to live with the SE you get (bootstrap them, both equations).

you could try xtivreg , fe.
xtset the data.
include year as i.year.
Again, same coefficients but the SE will be different.

(I notice we have had this conversation before

https://www.statalist.org/forums/forum/general-stata-discussion/general/1760025-constant-in-2sls-using-ivreghdfe

).

I'd suggest spending some time understanding why the constant does not appear, and explain that to your supervisor. Make him explain why he so badly wants a meaningless coefficient.

[hussein bataineh]

Thansk for your clarification i appreciate that , please have question related to 2sls

in my main regression model

xtreg EQUITY CSO FSIZE OP_CF SD_OCF TAX BM LEV MA Gov_score DPP RPP PSIZE DR Sustain_Perf Sust_Commit i.year , fe robust

when i run 2sls i used

ivreghdfe EQUITY FSIZE OP_CF SD_OCF TAX BM LEV MA Gov_score DPP RPP PSIZE DR Sustain_Perf Sust_Commit (CSO =iv1), first absorb(id year) robust endog(CSO)


Is what i did right or i should is right or i should use other command for example xtivreg , fe.

[George Ford]

Why not match them up using (iv) reghdfe for direct comparability?

reghdfe  EQUITY   CSO           FSIZE OP_CF SD_OCF TAX BM LEV MA Gov_score DPP RPP PSIZE DR Sustain_Perf Sust_Commit , absorb(id year) cluster(id) 
ivreghdfe EQUITY (CSO =iv1) FSIZE OP_CF SD_OCF TAX BM LEV MA Gov_score DPP RPP PSIZE DR Sustain_Perf Sust_Commit , absorb(id year) cluster(id) first endog(CSO)

As for your constant, just put 0.00012 and se = 0.047 in the table to satisfy your supervisor. These are as meaningful as what he's asking for. You can drop them when you try to publish the paper.

[hussein bataineh]

Actually the literature that im following use firm -year fixed effect , so would like to know from econometrics background does using
ivreghdfe good

ivreghdfe EQUITY FSIZE OP_CF SD_OCF TAX BM LEV MA Gov_score DPP RPP PSIZE DR Sustain_Perf Sust_Commit (CSO =iv1), first absorb(id year) robust endog(CSO)

[George Ford]

reghdfe absorbs id and year FE. cleaner than using i.year and xtreg.