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]
RE: st: RE: IVREG2 with two endogenous variables: CD/KP F-stat vs.APF-stat
From
"Schaffer, Mark E" <[email protected]>
To
<[email protected]>
Subject
RE: st: RE: IVREG2 with two endogenous variables: CD/KP F-stat vs.APF-stat
Date
Fri, 1 Feb 2013 22:57:08 -0000
Miquel,
The results are little odd and I can't work out why.
The (unrobust) Cragg-Donald statistic suggests that the equation is identified but the (robust) K-P statistic suggests it is underidentified. I am not sure why that would be.
Plus the individual (robust) A-P test statistics suggest that the individual coeffs on the two endogenous variables are separately identified, but the (similarly-robust) K-P underidentification test says they aren't jointly identified.
I see you're using the -partial- option. What happens if you partial out all the exogenous regressors?
--Mark
> -----Original Message-----
> From: [email protected] [mailto:owner-
> [email protected]] On Behalf Of Miguel Àngel García López
> Sent: 01 February 2013 21:17
> To: [email protected]
> Subject: Re: st: RE: IVREG2 with two endogenous variables: CD/KP F-stat
> vs.APF-stat
>
> Hi Mark,
>
> You're right, the Kleibergen-Paap rk LM statistic indicates that the equation is
> underidentified.
> So, which coefficient am I identifying? x1=disnode20061991 or
> x2=disnode20061991dhistink (this is the interacted endogenous variable)?
>
> Thanks a lot!
>
> Here is the output:
>
>
> First-stage regressions
> -----------------------
>
> First-stage regression of disnode20061991:
>
> OLS estimation
> --------------
>
> Estimates efficient for homoskedasticity only Statistics robust to
> heteroskedasticity
>
> Number of obs = 603
> F( 26, 464) = 2.21
> Prob > F = 0.0006
> Total (centered) SS = 2045.649523 Centered R2 = 0.2016
> Total (uncentered) SS = 2045.649523 Uncentered R2 = 0.2016
> Residual SS = 1633.166249 Root MSE = 1.876
>
> ------------------------------------------------------------------------------
> | Robust
> disnod~61991 | Coef. Std. Err. t P>|t| [95% Conf. Interval]
> -------------+----------------------------------------------------------
> -------------+------
> dhistlink | -.0549742 .3248397 -0.17 0.866 -.6933134 .583365
> lnpob1991 | .0061794 .141505 0.04 0.965 -.2718907 .2842495
> lnkm2 | .1863085 .166866 1.12 0.265 -.1415981 .5142151
> disCBD | -.0709631 .0430135 -1.65 0.100 -.1554885 .0135622
> discoast | -.0602264 .0400691 -1.50 0.134 -.1389658 .0185131
> alti | -.0023203 .0027915 -0.83 0.406 -.0078058 .0031652
> tri | -.0115232 .0067075 -1.72 0.086 -.0247041 .0016577
> lat1 | 5.500978 3.274879 1.68 0.094 -.9344523 11.93641
> long1 | -.9525434 2.43592 -0.39 0.696 -5.739344 3.834257
> taxa_at~1991 | 6.955092 4.800384 1.45 0.148 -2.478093 16.38828
> taxa_oc~1991 | .008856 3.143292 0.00 0.998 -6.167996 6.185708
> porc_ocup1~n | -3.03856 1.642994 -1.85 0.065 -6.267191 .1900711
> porc_pob1~25 | -1.580318 3.761586 -0.42 0.675 -8.972172 5.811537
> porc_pob19~i | 8.381571 3.778023 2.22 0.027 .9574171 15.80572
> porc_pob19~r | -2.879711 4.192779 -0.69 0.493 -11.1189 5.359476
> lnpob1981 | -.0628442 .0730382 -0.86 0.390 -.2063707 .0806824
> lnpob1970 | -.1758226 .1700468 -1.03 0.302 -.5099798 .1583347
> lnpob1960 | .5060852 .5099257 0.99 0.321 -.4959645 1.508135
> lnpob1950 | -.3825519 .4967032 -0.77 0.442 -1.358618 .5935145
> lnpob1940 | -.3062138 .7478307 -0.41 0.682 -1.775768 1.163341
> lnpob1930 | .6099881 1.073278 0.57 0.570 -1.499099 2.719075
> lnpob1920 | -1.898675 1.43143 -1.33 0.185 -4.711564 .9142143
> lnpob1910 | 1.829188 1.056597 1.73 0.084 -.2471203 3.905495
> lnpob1900 | -.1280441 .0932029 -1.37 0.170 -.3111962 .0551081
> dis1760 | -.1942985 .045257 -4.29 0.000 -.2832326 -.1053645
> dis1760dhi~k | -.0141311 .0116053 -1.22 0.224 -.0369365 .0086743
> ------------------------------------------------------------------------------
> Included instruments: dhistlink lnpob1991 lnkm2 disCBD discoast alti tri lat1
> long1 taxa_atur1991 taxa_ocup1991 porc_ocup1991man
> porc_pob1991m25 porc_pob1991uni porc_pob1991estr lnpob1981
> lnpob1970 lnpob1960 lnpob1950 lnpob1940 lnpob1930
> lnpob1920 lnpob1910 lnpob1900 dis1760 dis1760dhistlink
> ------------------------------------------------------------------------------
> F test of excluded instruments:
> F( 2, 464) = 11.50
> Prob > F = 0.0000
> Angrist-Pischke multivariate F test of excluded instruments:
> F( 1, 464) = 20.90
> Prob > F = 0.0000
>
> First-stage regression of disnode20061991dhistlink:
>
> OLS estimation
> --------------
>
> Estimates efficient for homoskedasticity only Statistics robust to
> heteroskedasticity
>
> Number of obs = 603
> F( 26, 464) = 2.67
> Prob > F = 0.0000
> Total (centered) SS = 28055.52566 Centered R2 = 0.4866
> Total (uncentered) SS = 28055.52566 Uncentered R2 = 0.4866
> Residual SS = 14404.85354 Root MSE = 5.572
>
> ------------------------------------------------------------------------------
> | Robust
> disnode200~k | Coef. Std. Err. t P>|t| [95% Conf. Interval]
> -------------+----------------------------------------------------------
> -------------+------
> dhistlink | 2.515352 1.115445 2.26 0.025 .3234024 4.707301
> lnpob1991 | .3581472 .3807795 0.94 0.347 -.3901187 1.106413
> lnkm2 | -.8305778 .4571589 -1.82 0.070 -1.728936 .0677805
> disCBD | .035202 .1068078 0.33 0.742 -.1746849 .2450889
> discoast | -.0824591 .0797964 -1.03 0.302 -.2392662 .0743481
> alti | .0139589 .0089892 1.55 0.121 -.0037056 .0316234
> tri | -.0481706 .0227555 -2.12 0.035 -.0928872 -.0034541
> lat1 | -1.303162 8.268787 -0.16 0.875 -17.55207 14.94575
> long1 | 4.330386 6.440689 0.67 0.502 -8.326146 16.98692
> taxa_at~1991 | -13.02405 17.44058 -0.75 0.456 -47.29635 21.24825
> taxa_oc~1991 | -.380474 8.68993 -0.04 0.965 -17.45697 16.69602
> porc_ocup1~n | 3.577323 3.999994 0.89 0.372 -4.283023 11.43767
> porc_pob1~25 | -18.19801 11.58473 -1.57 0.117 -40.96303 4.567024
> porc_pob19~i | -2.35895 9.349228 -0.25 0.801 -20.73102 16.01312
> porc_pob19~r | -2.417754 7.014291 -0.34 0.730 -16.20147 11.36596
> lnpob1981 | -.5197725 .4901694 -1.06 0.290 -1.482999 .4434545
> lnpob1970 | -.4056388 1.066181 -0.38 0.704 -2.50078 1.689502
> lnpob1960 | .1810121 1.474015 0.12 0.902 -2.71556 3.077584
> lnpob1950 | .8724695 .9530778 0.92 0.360 -1.000414 2.745353
> lnpob1940 | .0072886 1.337592 0.01 0.996 -2.621199 2.635776
> lnpob1930 | .1341517 1.96242 0.07 0.946 -3.72218 3.990484
> lnpob1920 | -2.063717 3.313446 -0.62 0.534 -8.574937 4.447502
> lnpob1910 | 2.01 2.898491 0.69 0.488 -3.685795 7.705795
> lnpob1900 | -.1108913 .319602 -0.35 0.729 -.738938 .5171553
> dis1760 | .2462499 .1164858 2.11 0.035 .017345 .4751549
> dis1760dhi~k | -.4080694 .086959 -4.69 0.000 -.5789517 -.2371871
> ------------------------------------------------------------------------------
> Included instruments: dhistlink lnpob1991 lnkm2 disCBD discoast alti tri lat1
> long1 taxa_atur1991 taxa_ocup1991 porc_ocup1991man
> porc_pob1991m25 porc_pob1991uni porc_pob1991estr lnpob1981
> lnpob1970 lnpob1960 lnpob1950 lnpob1940 lnpob1930
> lnpob1920 lnpob1910 lnpob1900 dis1760 dis1760dhistlink
> ------------------------------------------------------------------------------
> F test of excluded instruments:
> F( 2, 464) = 11.53
> Prob > F = 0.0000
> Angrist-Pischke multivariate F test of excluded instruments:
> F( 1, 464) = 22.72
> Prob > F = 0.0000
>
>
>
> Summary results for first-stage regressions
> -------------------------------------------
>
> (Underid) (Weak id)
> Variable | F( 2, 464) P-val | AP Chi-sq( 1) P-val | AP F( 1, 464)
> disnode20061 | 11.50 0.0000 | 27.16 0.0000 | 20.90
> disnode20061 | 11.53 0.0000 | 29.53 0.0000 | 22.72
>
> NB: first-stage test statistics heteroskedasticity-robust
>
> Stock-Yogo weak ID 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 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(1)=0.00 P-val=0.9997
>
> Weak identification test
> Ho: equation is weakly identified
> Cragg-Donald Wald F statistic 27.11
> Kleibergen-Paap Wald rk F statistic 0.00
>
> Stock-Yogo weak ID test critical values for K1=2 and L1=2:
> 10% maximal IV size 7.03
> 15% maximal IV size 4.58
> 20% maximal IV size 3.95
> 25% maximal IV size 3.63
> 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(2,464)= 4.17 P-val=0.0161
> Anderson-Rubin Wald test Chi-sq(2)= 10.83 P-val=0.0044
> Stock-Wright LM S statistic Chi-sq(2)= 10.18 P-val=0.0062
>
> NB: Underidentification, weak identification and weak-identification-robust
> test statistics heteroskedasticity-robust
>
> Number of observations N = 603
> Number of regressors K = 26
> Number of endogenous regressors K1 = 2
> Number of instruments L = 26
> Number of excluded instruments L1 = 2
> Number of partialled-out regressors/IVs = 113
> NB: K & L do not included partialled-out variables
>
> IV (2SLS) estimation
> --------------------
>
> Estimates efficient for homoskedasticity only Statistics robust to
> heteroskedasticity
>
> Number of obs = 603
> F( 26, 464) = 17.35
> Prob > F = 0.0000
> Total (centered) SS = 32.85467898 Centered R2 = 0.4789
> Total (uncentered) SS = 32.85467898 Uncentered R2 = 0.4789
> Residual SS = 17.12178597 Root MSE = .1685
>
> ------------------------------------------------------------------------------
> | Robust
> lnp~20111991 | Coef. Std. Err. z P>|z| [95% Conf. Interval]
> -------------+----------------------------------------------------------
> -------------+------
> disnod~61991 | -.0422962 .0157094 -2.69 0.007 -.0730861 -.0115064
> disnode200~k | .0033454 .002269 1.47 0.140 -.0011017 .0077925
> dhistlink | -.0193715 .0301388 -0.64 0.520 -.0784425 .0396996
> lnpob1991 | -.1716033 .0175352 -9.79 0.000 -.2059716 -.1372351
> lnkm2 | .0772166 .0201062 3.84 0.000 .0378092 .1166241
> disCBD | -.0014151 .0038103 -0.37 0.710 -.0088832 .006053
> discoast | -.0020445 .0030389 -0.67 0.501 -.0080007 .0039116
> alti | -.0003639 .0002803 -1.30 0.194 -.0009132 .0001854
> tri | .0002327 .0007257 0.32 0.748 -.0011896 .001655
> lat1 | .2161914 .27737 0.78 0.436 -.3274437 .7598266
> long1 | -.3242189 .1943732 -1.67 0.095 -.7051834 .0567457
> taxa_at~1991 | -.0633292 .4341575 -0.15 0.884 -.9142623 .7876039
> taxa_oc~1991 | .5256404 .2733951 1.92 0.055 -.0102043 1.061485
> porc_ocup1~n | -.2804171 .1410424 -1.99 0.047 -.5568552 -.0039791
> porc_pob1~25 | -2.038739 .4678653 -4.36 0.000 -2.955738 -1.121739
> porc_pob19~i | 1.98907 .4010011 4.96 0.000 1.203123 2.775018
> porc_pob19~r | 1.207574 .3368401 3.59 0.000 .5473792 1.867768
> lnpob1981 | -.0206563 .0138923 -1.49 0.137 -.0478848 .0065722
> lnpob1970 | -.0664541 .031215 -2.13 0.033 -.1276344 -.0052739
> lnpob1960 | .035141 .0807666 0.44 0.663 -.1231586 .1934405
> lnpob1950 | .0388572 .0753196 0.52 0.606 -.1087665 .1864808
> lnpob1940 | -.0780994 .0883634 -0.88 0.377 -.2512886 .0950897
> lnpob1930 | .0474178 .1058593 0.45 0.654 -.1600626 .2548983
> lnpob1920 | .2845001 .1418916 2.01 0.045 .0063977 .5626024
> lnpob1910 | -.2330463 .1296206 -1.80 0.072 -.4870979 .0210054
> lnpob1900 | -.0073257 .0135746 -0.54 0.589 -.0339315 .01928
> ------------------------------------------------------------------------------
> Underidentification test (Kleibergen-Paap rk LM statistic): 0.000
> Chi-sq(1) P-val = 0.9997
> ------------------------------------------------------------------------------
> Weak identification test (Cragg-Donald Wald F statistic): 27.106
> (Kleibergen-Paap rk Wald F statistic): 0.000
> Stock-Yogo weak ID test critical values: 10% maximal IV size 7.03
> 15% maximal IV size 4.58
> 20% maximal IV size 3.95
> 25% maximal IV size 3.63
> 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)
> ------------------------------------------------------------------------------
> Instrumented: disnode20061991 disnode20061991dhistlink
> Included instruments: dhistlink lnpob1991 lnkm2 disCBD discoast alti tri lat1
> long1 taxa_atur1991 taxa_ocup1991 porc_ocup1991man
> porc_pob1991m25 porc_pob1991uni porc_pob1991estr lnpob1981
> lnpob1970 lnpob1960 lnpob1950 lnpob1940 lnpob1930
> lnpob1920 lnpob1910 lnpob1900 Excluded instruments: dis1760
> dis1760dhistlink
> Partialled-out: _Iam129_2 _Iam129_3 _Iam129_4 _Iam129_5 _Iam129_6
>
>
> Miquel
>
> ----- Missatge original -----
> De: "Schaffer, Mark E" <[email protected]>
> Data: Divendres, Febrer 1, 2013 9:03 pm
> Assumpte: st: RE: IVREG2 with two endogenous variables: CD/KP F-stat vs.
> APF-stat
>
> > Miguel,
> >
> > > -----Original Message-----
> > > From: [email protected] [owner-
> > > [email protected]] On Behalf Of Miguel Àngel García
> > López> Sent: 01 February 2013 09:22
> > > To: [email protected]
> > > Subject: st: IVREG2 with two endogenous variables: CD/KP F-stat
> > vs. AP F-
> > > stat
> > >
> > > Hi all,
> > >
> > > I am using ivreg2 for running my model with two endogenous
> > variables:>
> > > ivreg2 y1 (x1 x2 = z1 z2) x3, robust
> > >
> > > where x2 and z2 are interactions of x1 and z1, respectively. In
> > other words,
> > > x2=dummy*x1 is an interacted endogenous variable and z2=dummy*z1
> > is an
> > > interacted instrument.
> > >
> > > When I look at the first-stage statistics I find that the KP F-
> > stat = 0.00 while
> > > the individual AP F-stats are around 20.
> > > And I am puzzle about which F-stat I should consider.
> > >
> > > Following previous discussions such as
> > > http://www.stata.com/statalist/archive/2011-04/msg00853.html, I
> > think I
> > > should only use AP F-stats.
> > > Am I right?
> > >
> > > Another question is why my CD F-stat is 0. Maybe because my
> > instruments> are highly correlated?
> >
> > Can you show us the output?
> >
> > It sounds like your equation is underidentified: you can identify
> > either the coeff on x1, or the coeff on x2, but not both coeffs
> > simultaneously (hence the zero under- and weak-identification
> > statistics).
> > But it's hard to tell without seeing more.
> >
> > --Mark
> >
> > >
> > > Thanks,
> > >
> > > Miquel
> > >
> > > *
> > > * 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
> > howto 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/
> >
>
>
> *
> * 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/