Bookmark and Share

Notice: On April 23, 2014, Statalist moved from an email list to a forum, based at statalist.org.


[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

st: RE: ivreg2 with two endogenous regressors: KP stats vs AP stats


From   "Schaffer, Mark E" <M.E.Schaffer@hw.ac.uk>
To   "statalist@hsphsun2.harvard.edu" <statalist@hsphsun2.harvard.edu>
Subject   st: RE: ivreg2 with two endogenous regressors: KP stats vs AP stats
Date   Tue, 2 Jul 2013 14:45:40 +0000

Esha,

Which version of ivreg2 are you using?  There was a bug in the K-P statistic for case of >1 endogenous regressors that affected the K-P stat for overall identification but not the stats for individual regressors, but it's now fixed.  The fix is in the Mata library livreg.mlib (not in ivreg2.ado) and we made it available in early May, if I recall correctly.

--Mark

> -----Original Message-----
> From: owner-statalist@hsphsun2.harvard.edu [mailto:owner-
> statalist@hsphsun2.harvard.edu] On Behalf Of Sraboni, Esha (IFPRI-Dhaka)
> Sent: 02 July 2013 12:26
> To: statalist@hsphsun2.harvard.edu
> Subject: st: ivreg2 with two endogenous regressors: KP stats vs AP stats
> 
> Dear Statalisters,
> 
> 
> I am using instrumental variables regression with two endogenous regressors.
> The 2SLS IV diagnostics suggest that my model is unidentified and the
> instruments are weak. The Kleibergen-Paap rk LM p-value is 1, and Kleibergen-
> Paap Wald rk F value (not Cragg-Donald, since I am using weights) is 0.
> However, the AP statistics (first stage under identification and weak
> identification tests) look good for the individual endogenous variables. I thought
> it strange that the first stage results look favorable, but the joint tests do not.
> 
> Another thing which I could not understand is that, I tried out a large number of
> different combinations of instruments for each of the endogenous regressors,
> but the KP rk LM p value and KP Wald rk F value remain unchanged at 1 and 0
> respectively.
> 
> I have pasted below the first stage and 2SLS output (I have deleted the results
> for the other controls, in order to avoid cluttering). I was hoping you could help
> me understand what might be happening here, or whether I am
> misinterpreting the statistics.
> 
> 
> Thank you for your time!
> 
> 
> Esha Sraboni
> 
> 
> 
> 
> local emp1 "emp_score2" //endogenous regressor local prodvar
> "num_allcrops"//endogenous regressor local hhchars "age_hhhead
> agesq_hhhead eduy_hhhead hh_head_farmer hh_head_trader hhsize
> pctmale04-pctmale1959 pctfemale04-pctfemale60 lnowncland_dec elec_conn
> rice_price d1-d6 num_milkcow  owns_handtbwell"
> 
> *** Instruments
> local z1 "agediff_m_f infcreditexclNGO_sum commparticip_any" //instruments
> for emp1 local w "soil_type4 soil_type5 pct_irrig" //instruments for prodvar
> 
> ***Regression
> 
> ivreg2 hhdietscore12 `hhchars'  (`emp1' `prodvar' = `z1' `w' ) [pw=hhweight],
> endog(`emp1' `prodvar') first ffirst savefirst savefp(hhm1`i') (sum of wgt is
> 2.0031e+07)
> 
> 
> First-stage regressions
> -----------------------
> 
> First-stage regression of emp_score2:
> 
> OLS estimation
> --------------
> 
> Estimates efficient for homoskedasticity only Statistics robust to
> heteroskedasticity
> 
>                                                       Number of obs =     4195
>                                                       F( 32,  4162) =    34.72
>                                                       Prob > F      =   0.0000
> Total (centered) SS     =  233.9254195                Centered R2   =   0.2098
> Total (uncentered) SS   =  2000.668046                Uncentered R2 =   0.9076
> Residual SS             =  184.8518389                Root MSE      =    .2107
> 
> ------------------------------------------------------------------------------
>                             |               Robust
>   emp_score2 |      Coef.   Std. Err.      t    P>|t|     [95% Conf. Interval]
> -------------+----------------------------------------------------------
> -------------+------
>  agediff_m_f |  -.0025851   .0008446    -3.06   0.002     -.004241   -.0009292
> infcredite~m |   .0286863   .0025982    11.04   0.000     .0235926    .0337801
> commpartic~y |   .0937299   .0072531    12.92   0.000     .0795098    .1079499
>   soil_type4 |   .0039734   .0093112     0.43   0.670    -.0142815    .0222283
>   soil_type5 |  -.0233485    .011089    -2.11   0.035    -.0450888   -.0016082
>    pct_irrig |   .0007367   .0000999     7.38   0.000     .0005409    .0009326
>        _cons |   .3315584   .0693319     4.78   0.000     .1956308     .467486
> ------------------------------------------------------------------------------
> Included instruments: age_hhhead agesq_hhhead eduy_hhhead
> hh_head_farmer
>                       hh_head_trader hhsize pctmale04 pctmale510 pctmale1118
>                       pctmale1959 pctfemale04 pctfemale510 pctfemale1118
>                       pctfemale1959 pctfemale60 lnowncland_dec elec_conn
>                       rice_price d1 d2 d3 d4 d5 d6 num_milkcow owns_handtbwell
>                       agediff_m_f infcreditexclNGO_sum commparticip_any
>                       soil_type4 soil_type5 pct_irrig
> ------------------------------------------------------------------------------
> F test of excluded instruments:
>   F(  6,  4162) =    75.15
>   Prob > F      =   0.0000
> Angrist-Pischke multivariate F test of excluded instruments:
>   F(  5,  4162) =    68.97
>   Prob > F      =   0.0000
> 
> First-stage regression of num_allcrops:
> 
> OLS estimation
> --------------
> 
> Estimates efficient for homoskedasticity only Statistics robust to
> heteroskedasticity
> 
>                                                       Number of obs =     4195
>                                                       F( 32,  4162) =    75.34
>                                                       Prob > F      =   0.0000
> Total (centered) SS     =  7786.104652                Centered R2   =   0.3880
> Total (uncentered) SS   =    12656.797                Uncentered R2 =   0.6235
> Residual SS             =  4765.006576                Root MSE      =     1.07
> 
> ------------------------------------------------------------------------------
>              |               Robust
> num_allcrops |      Coef.   Std. Err.      t    P>|t|     [95% Conf. Interval]
> -------------+----------------------------------------------------------
> -------------+------
>  agediff_m_f |   .0032771   .0041113     0.80   0.425    -.0047833    .0113375
> infcredite~m |   .0322633   .0115057     2.80   0.005      .009706    .0548206
> commpartic~y |    .050581    .036366     1.39   0.164    -.0207159    .1218779
>   soil_type4 |   .4495017    .056289     7.99   0.000     .3391451    .5598582
>   soil_type5 |   .5629801   .0660527     8.52   0.000     .4334815    .6924786
>    pct_irrig |   .0103765   .0005456    19.02   0.000     .0093069    .0114461
>        _cons |  -.5467235   .3383645    -1.62   0.106    -1.210099    .1166517
> ------------------------------------------------------------------------------
> Included instruments: age_hhhead agesq_hhhead eduy_hhhead
> hh_head_farmer
>                       hh_head_trader hhsize pctmale04 pctmale510 pctmale1118
>                       pctmale1959 pctfemale04 pctfemale510 pctfemale1118
>                       pctfemale1959 pctfemale60 lnowncland_dec elec_conn
>                       rice_price d1 d2 d3 d4 d5 d6 num_milkcow owns_handtbwell
>                       agediff_m_f infcreditexclNGO_sum commparticip_any
>                       soil_type4 soil_type5 pct_irrig
> ------------------------------------------------------------------------------
> F test of excluded instruments:
>   F(  6,  4162) =   171.29
>   Prob > F      =   0.0000
> Angrist-Pischke multivariate F test of excluded instruments:
>   F(  5,  4162) =   140.12
>   Prob > F      =   0.0000
> 
> 
> 
> Summary results for first-stage regressions
> -------------------------------------------
> 
>                                            (Underid)            (Weak id)
> Variable     | F(  6,  4162)  P-val | AP Chi-sq(  5) P-val | AP F(  5,  4162)
> emp_score2   |      75.15    0.0000 |      347.60   0.0000 |       68.97
> num_allcrops |     171.29    0.0000 |      706.16   0.0000 |      140.12
> 
> NB: first-stage test statistics heteroskedasticity-robust
> 
> Stock-Yogo weak ID test critical values for single endogenous regressor:
>                                     5% maximal IV relative bias    19.28
>                                    10% maximal IV relative bias    10.83
>                                    20% maximal IV relative bias     6.77
>                                    30% maximal IV relative bias     5.25
>                                    10% maximal IV size             26.87
>                                    15% maximal IV size             15.09
>                                    20% maximal IV size             10.98
>                                    25% maximal IV size              8.84
> Source: Stock-Yogo (2005).  Reproduced by permission.
> NB: Critical values are for Cragg-Donald F statistic and i.i.d. errors.
> 
> 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(5)=0.00     P-val=1.0000
> 
> Weak identification test
> Ho: equation is weakly identified
> Cragg-Donald Wald F statistic                                      56.94
> Kleibergen-Paap Wald rk F statistic                                 0.00
> 
> Stock-Yogo weak ID test critical values for K1=2 and L1=6:
>                                     5% maximal IV relative bias    15.72
>                                    10% maximal IV relative bias     9.48
>                                    20% maximal IV relative bias     6.08
>                                    30% maximal IV relative bias     4.78
>                                    10% maximal IV size             21.68
>                                    15% maximal IV size             12.33
>                                    20% maximal IV size              9.10
>                                    25% maximal IV size              7.42
> 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(6,4162)=     11.07     P-val=0.0000
> Anderson-Rubin Wald test           Chi-sq(6)=     66.92     P-val=0.0000
> Stock-Wright LM S statistic        Chi-sq(6)=     65.36     P-val=0.0000
> 
> NB: Underidentification, weak identification and weak-identification-robust
>     test statistics heteroskedasticity-robust
> 
> Number of observations               N  =       4195
> Number of regressors                 K  =         29
> Number of endogenous regressors      K1 =          2
> Number of instruments                L  =         33
> Number of excluded instruments       L1 =          6
> 
> IV (2SLS) estimation
> --------------------
> 
> Estimates efficient for homoskedasticity only Statistics robust to
> heteroskedasticity
> 
>                                                       Number of obs =     4195
>                                                       F( 28,  4166) =    32.84
>                                                       Prob > F      =   0.0000
> Total (centered) SS     =   10642.6554                Centered R2   =   0.1356
> Total (uncentered) SS   =  392905.5563                Uncentered R2 =   0.9766
> Residual SS             =  9199.049659                Root MSE      =    1.481
> 
> ------------------------------------------------------------------------------
>              |               Robust
> hhdietsco~12 |      Coef.   Std. Err.      z    P>|z|     [95% Conf. Interval]
> -------------+----------------------------------------------------------
> -------------+------
>   emp_score2 |   2.073557   .3761406     5.51   0.000     1.336335    2.810779
> num_allcrops |   .0998546   .0510304     1.96   0.050    -.0001631    .1998723
> 
>        _cons |   6.311798    .506199    12.47   0.000     5.319666     7.30393
> ------------------------------------------------------------------------------
> Underidentification test (Kleibergen-Paap rk LM statistic):              0.000
>                                                    Chi-sq(5) P-val =    1.0000
> ------------------------------------------------------------------------------
> Weak identification test (Cragg-Donald Wald F statistic):               56.943
>                          (Kleibergen-Paap rk Wald F statistic):          0.000
> Stock-Yogo weak ID test critical values:  5% maximal IV relative bias    15.72
>                                          10% maximal IV relative bias     9.48
>                                          20% maximal IV relative bias     6.08
>                                          30% maximal IV relative bias     4.78
>                                          10% maximal IV size             21.68
>                                          15% maximal IV size             12.33
>                                          20% maximal IV size              9.10
>                                          25% maximal IV size              7.42
> 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):         6.247
>                                                    Chi-sq(4) P-val =    0.1814
> -endog- option:
> Endogeneity test of endogenous regressors:                              33.049
>                                                    Chi-sq(2) P-val =    0.0000
> Regressors tested:    emp_score2 num_allcrops
> ------------------------------------------------------------------------------
> Instrumented:         emp_score2 num_allcrops
> Included instruments: age_hhhead agesq_hhhead eduy_hhhead
> hh_head_farmer
>                       hh_head_trader hhsize pctmale04 pctmale510 pctmale1118
>                       pctmale1959 pctfemale04 pctfemale510 pctfemale1118
>                       pctfemale1959 pctfemale60 lnowncland_dec elec_conn
>                       rice_price d1 d2 d3 d4 d5 d6 num_milkcow owns_handtbwell
> Excluded instruments: agediff_m_f infcreditexclNGO_sum commparticip_any
>                       soil_type4 soil_type5 pct_irrig
> ------------------------------------------------------------------------------
> 
> *
> *   For searches and help try:
> *   http://www.stata.com/help.cgi?search
> *   http://www.stata.com/support/faqs/resources/statalist-faq/
> *   http://www.ats.ucla.edu/stat/stata/


----- 
Sunday Times Scottish University of the Year 2011-2013
Top in the UK for student experience
Fourth university in the UK and top in Scotland (National Student Survey 2012)

We invite research leaders and ambitious early career researchers to 
join us in leading and driving research in key inter-disciplinary themes. 
Please see www.hw.ac.uk/researchleaders for further information and how
to apply.

Heriot-Watt University is a Scottish charity
registered under charity number SC000278.


*
*   For searches and help try:
*   http://www.stata.com/help.cgi?search
*   http://www.stata.com/support/faqs/resources/statalist-faq/
*   http://www.ats.ucla.edu/stat/stata/


© Copyright 1996–2018 StataCorp LLC   |   Terms of use   |   Privacy   |   Contact us   |   Site index