Monica,
It seems that you have some sort of "local instruments". That means that
for one group your instruments can "explain" well the instrumented
variable (your R2 of 31%) meanwhile for the other group those explain
badly the endogenous variable (5%). When you collapse them into one big
group you end up with 8% (*). I think that you could keep the analysis
at the subsample level, maybe there are other more relevant instruments
for each group that lost significance in the collapse analysis. Think a
labor-example, you could have a educational policy that helps
less-skilled worked (primary education) but has no effect in
college-education.
I hope this helps you.
Rodrigo.
(*) That could be explained by the dispersion generated by the small R2.
For example, we can think on the collapsed R2 as an harmonic average of
subsample R2's: 1/R2 = w1*(1/R21) + w2*(1/R22), where R21 and R22 are
the R2's of the subsamples, and w1 and w2 are the weigths based on the
realtive size of the subsamples. Your first group (young) weights 0.46
(=4959/10799, there are 87 lost cases!), then 1/R2 = 0.46*(1/0.31) +
0.54*(1/0.05), with that R2 = 0.08.
-----Mensaje original-----
De: [email protected]
[mailto:[email protected]] En nombre de
[email protected]
Enviado el: Martes, 01 de Abril de 2008 01:44 a.m.
Para: [email protected]
Asunto: st: F-stat of excl inst too big in robsut estimation with
dummydependent variable (xtivreg2)
Hi,
I need some help understanding why the F-statistic of the excluded
instruments is so big for a subset of my sample.
I am estimating an IV fe model using xtivreg: y x1 x2 t ct (x1=z), fe
robust fwl(t ct) i(nidhogar) t(year). My full sample is composed by
1334 households and 9 years (10886 obs. in total). The dependent
variable is a dummy taking the value of 1 if the household owns a
business and zero otherwise. The endogenous variable is an income
variable truncated at zero for all households but taking positive
values for some households in all or some years. In addition I am
controlling for time-varying household characteristics, time dummies and
community-time dummies (there are 8 communities- too few to cluster on).
I use the fwl option to partial out the time and community-time dummies.
The instrument is a variable similar to the endogenous variable, that
is, an income type variable but exogenous. I am confident about my
instrument doing a good job in correcting for endogeneity. Also, in
previous versions I had included additional instruments and tested for
overidentification, endogeneity and redundancy and this one performed
quite well.
Looking at the results of the full sample, I believe the estimation is
just fine. My concern is that when I partition the sample in two
(household heads younger and older than 39 years old) the F-statistic of
excluded instruments becomes very large for the sample of younger heads
(5712.73). However, standard errors in both first stage and main
regression and other statistics seem fine. What could be happening?
What am I missing? Are my results still valid? How can I fix this, if
there is something to fix?
I read an earlier post (by Vince Wiggins answering to Mark Schaffer) in
which the problem was the F-stat not being reported in a
heteroskedastic-robust estimation with dummy variables. My intuition is
that in my case the opposite might be happening, but not sure? I checked
and my dependent variable takes the value of 1 for 23% of the
observations in the sample of young heads.
The sample size for the two age groups is: 3477 observations (591 hhs)
and 7727 (1031 hhs).
Below I?m copying the results of the estimations using the full sample
and the restricted samples. Please let me know if further information
is needed.
I very much appreciate your comments and suggestions.
Regards,
Monica
FULL SAMPLE:
. xtivreg2 dabus_sum
> dagro lnpland_manz `agesq' `hhcharact' `ctfe' `te' ( potincm_fo =
> `inst80s_fo' ) , fe i(nidhogar) r first ffirst rf fwl(`ctfe' `te') ;
Warning - singleton groups detected. 38 observation(s) not used.
Warning - collinearities detected
Vars dropped: dcomm1_99 dcomm2_98 dcomm3_99 dcomm4_96 dcomm5_94
dcomm7_98
dcomm8_99 dyr9
FIXED EFFECTS ESTIMATION
------------------------
Number of groups = 1334 Obs per group: min =
2
avg =
8.2
max =
9
Reduced-form regression: dabus_sum
----------------------------------
FIXED EFFECTS ESTIMATION
------------------------
Number of groups = 1334 Obs per group: min =
2
avg =
8.2
max =
9
OLS estimation
--------------
Statistics robust to heteroskedasticity
Number of obs =
10886
F( 6, 9546) =
11.03
Prob > F =
0.0000
Total (centered) SS = 280.0052317 Centered R2 =
0.0155
Total (uncentered) SS = 280.0052317 Uncentered R2 =
0.0155
Residual SS = 275.6691323 Root MSE =
.1699
------------------------------------------------------------------------
------
| Robust
dabus_sum | Coef. Std. Err. t P>|t| [95% Conf.
Interval]
-------------+----------------------------------------------------------
-------------+------
dagro | -.1568444 .0304295 -5.15 0.000 -.2164927
-.0971961
lnpland_manz | -.0091142 .0113462 -0.80 0.422 -.0313552
.0131269
agehead | .024494 .0045082 5.43 0.000 .0156569
.0333311
ageheadsqr | -.0001439 .0000387 -3.72 0.000 -.0002198
-.0000681
kids14less | .003819 .003182 1.20 0.230 -.0024184
.0100565
potincm80s~o | -.0143489 .0072045 -1.99 0.046 -.0284713
-.0002265
------------------------------------------------------------------------
------
Included instruments: dagro lnpland_manz agehead ageheadsqr kids14less
potincm80s_fo
------------------------------------------------------------------------
------
First-stage regressions
-----------------------
First-stage regression of potincm_fo:
FIXED EFFECTS ESTIMATION
------------------------
Number of groups = 1334 Obs per group: min =
2
avg =
8.2
max =
9
OLS estimation
--------------
Statistics robust to heteroskedasticity
Number of obs =
10886
F( 6, 9546) =
16.47
Prob > F =
0.0000
Total (centered) SS = 10371.54444 Centered R2 =
0.0902
Total (uncentered) SS = 10371.54444 Uncentered R2 =
0.0902
Residual SS = 9436.014852 Root MSE =
.9942
------------------------------------------------------------------------
------
| Robust
potincm_fo | Coef. Std. Err. t P>|t| [95% Conf.
Interval]
-------------+----------------------------------------------------------
-------------+------
dagro | .031293 .0752245 0.42 0.677 -.1161631
.1787491
lnpland_manz | -.0771407 .0256491 -3.01 0.003 -.1274185
-.026863
agehead | -.0498908 .0247468 -2.02 0.044 -.0983999
-.0013818
ageheadsqr | .0011953 .000256 4.67 0.000 .0006935
.0016972
kids14less | -.0290633 .0182759 -1.59 0.112 -.0648879
.0067613
potincm80s~o | .6552578 .1159389 5.65 0.000 .4279929
.8825227
------------------------------------------------------------------------
------
Included instruments: dagro lnpland_manz agehead ageheadsqr kids14less
potincm80s_fo
------------------------------------------------------------------------
------
Partial R-squared of excluded instruments: 0.0839
Test of excluded instruments:
F( 1, 9546) = 31.94
Prob > F = 0.0000
Summary results for first-stage regressions
-------------------------------------------
Variable | Shea Partial R2 | Partial R2 | F( 1, 9546)
P-value
potincm_fo | 0.0839 | 0.0839 | 31.94
0.0000
NB: first-stage F-stat heteroskedasticity-robust
Underidentification tests
Ho: matrix of reduced form coefficients has rank=K-1 (underidentified)
Ha: matrix has rank=K (identified)
Chi-sq(1) P-value
Anderson canon. corr. -N*ln(1-CCEV) LR stat. 837.05 0.0000
Cragg-Donald N*CDEV statistic 874.82 0.0000
Robust chi-square statistic 31.96 0.0000
Weak identification tests
Ho: equation is weakly identified
Cragg-Donald (N-L)*CDEV/L1 F-statistic 874.27
Robust F-statistic 31.94
See main output for Cragg-Donald weak id test critical values
NB: Anderson and Cragg-Donald under- and weak identification stats not
robust
Robust identification stats heteroskedasticity-robust
Weak-instrument-robust inference
Tests of joint significance of endogenous regressors B1 in main equation
Ho: B1=0 and overidentifying restrictions are valid
Anderson-Rubin test F(1,9546)=3.97 P-val=0.0464
Anderson-Rubin test Chi-sq(1)=3.97 P-val=0.0463
Stock-Wright S statistic Chi-sq(1)=2.98 P-val=0.0845
NB: Test statistics heteroskedasticity-robust
Number of observations N = 10886
Number of regressors K = 6
Number of instruments L = 6
Number of excluded instruments L1 = 1
IV (2SLS) estimation
--------------------
Statistics robust to heteroskedasticity
Number of obs =
10886
F( 6, 9546) =
11.10
Prob > F =
0.0000
Total (centered) SS = 280.0052317 Centered R2 =
0.0008
Total (uncentered) SS = 280.0052317 Uncentered R2 =
0.0008
Residual SS = 279.7700696 Root MSE =
.1711
------------------------------------------------------------------------
------
| Robust
dabus_sum | Coef. Std. Err. z P>|z| [95% Conf.
Interval]
-------------+----------------------------------------------------------
-------------+------
potincm_fo | -.0218981 .0100669 -2.18 0.030 -.0416289
-.0021673
dagro | -.1561592 .0302779 -5.16 0.000 -.2155027
-.0968156
lnpland_manz | -.0108034 .010944 -0.99 0.324 -.0322532
.0106464
agehead | .0234015 .0046015 5.09 0.000 .0143827
.0324203
ageheadsqr | -.0001177 .0000403 -2.92 0.003 -.0001967
-.0000388
kids14less | .0031826 .0032288 0.99 0.324 -.0031456
.0095109
------------------------------------------------------------------------
------
Anderson canon. corr. LR statistic (underidentification test):
837.049
Chi-sq(1) P-val =
0.0000
Test statistic(s) not robust
------------------------------------------------------------------------
------
Cragg-Donald F statistic (weak identification test):
874.270
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
Test statistic(s) not robust
Source: Stock-Yogo (2005). Reproduced by permission.
------------------------------------------------------------------------
------
Hansen J statistic (overidentification test of all instruments):
0.000
(equation exactly
identified)
------------------------------------------------------------------------
------
Instrumented: potincm_fo
Included instruments: dagro lnpland_manz agehead ageheadsqr kids14less
Excluded instruments: potincm80s_fo Partialled-out (FWL): dcomm1_91
dcomm1_92 dcomm1_93 dcomm1_94 dcomm1_95
dcomm1_96 dcomm1_97 dcomm1_98 dcomm2_91 dcomm2_92
dcomm2_93 dcomm2_94 dcomm2_95 dcomm2_96 dcomm2_97
dcomm2_99 dcomm3_91 dcomm3_92 dcomm3_93 dcomm3_94
dcomm3_95 dcomm3_96 dcomm3_97 dcomm3_98 dcomm4_91
dcomm4_92 dcomm4_93 dcomm4_94 dcomm4_95 dcomm4_97
dcomm4_98 dcomm4_99 dcomm5_91 dcomm5_92 dcomm5_93
dcomm5_95 dcomm5_96 dcomm5_97 dcomm5_98 dcomm5_99
dcomm7_91 dcomm7_92 dcomm7_93 dcomm7_94 dcomm7_95
dcomm7_96 dcomm7_97 dcomm7_99 dcomm8_91 dcomm8_92
dcomm8_93 dcomm8_94 dcomm8_95 dcomm8_96 dcomm8_97
dcomm8_98 dyr2 dyr3 dyr4 dyr5 dyr6 dyr7 dyr8
nb: variable counts and small-sample adjustments
do not include partialled-out variables.
Dropped collinear: dcomm1_99 dcomm2_98 dcomm3_99 dcomm4_96 dcomm5_94
dcomm7_98 dcomm8_99 dyr9
------------------------------------------------------------------------
------
YOUNGER HEADS SAMPLE:
. xtivreg2 dabus_sum
> dagro lnpland_manz `agesq' `hhcharact' `ctfe' `te' ( potincm_fo =
> `inst80s_fo' ) if dagehead2539 ==1 , fe i(nidhogar) r first ffirst
> rf fwl(`ctfe' `te') ;
Warning - singleton groups detected. 69 observation(s) not used.
Warning - collinearities detected
Vars dropped: dcomm1_99 dcomm2_99 dcomm3_99 dcomm4_99 dcomm5_97
dcomm7_98
dcomm8_99 dyr9
FIXED EFFECTS ESTIMATION
------------------------
Number of groups = 749 Obs per group: min =
2
avg =
6.6
max =
9
Reduced-form regression: dabus_sum
----------------------------------
FIXED EFFECTS ESTIMATION
------------------------
Number of groups = 749 Obs per group: min =
2
avg =
6.6
max =
9
OLS estimation
--------------
Statistics robust to heteroskedasticity
Number of obs =
4959
F( 6, 4204) =
7.23
Prob > F =
0.0000
Total (centered) SS = 119.6871152 Centered R2 =
0.0322
Total (uncentered) SS = 119.6871152 Uncentered R2 =
0.0322
Residual SS = 115.8295136 Root MSE =
.166
------------------------------------------------------------------------
------
| Robust
dabus_sum | Coef. Std. Err. t P>|t| [95% Conf.
Interval]
-------------+----------------------------------------------------------
-------------+------
dagro | -.1429524 .0403911 -3.54 0.000 -.2221404
-.0637645
lnpland_manz | -.0025269 .0153996 -0.16 0.870 -.0327183
.0276645
agehead | .0358974 .0136655 2.63 0.009 .0091058
.0626891
ageheadsqr | -.0003511 .0002082 -1.69 0.092 -.0007592
.000057
kids14less | .0056506 .0051194 1.10 0.270 -.0043861
.0156873
potincm80s~o | -.0641757 .0172277 -3.73 0.000 -.0979511
-.0304002
------------------------------------------------------------------------
------
Included instruments: dagro lnpland_manz agehead ageheadsqr kids14less
potincm80s_fo
------------------------------------------------------------------------
------
First-stage regressions
-----------------------
First-stage regression of potincm_fo:
FIXED EFFECTS ESTIMATION
------------------------
Number of groups = 749 Obs per group: min =
2
avg =
6.6
max =
9
OLS estimation
--------------
Statistics robust to heteroskedasticity
Number of obs =
4959
F( 6, 4204) =
1000.42
Prob > F =
0.0000
Total (centered) SS = 1882.108159 Centered R2 =
0.3080
Total (uncentered) SS = 1882.108159 Uncentered R2 =
0.3080
Residual SS = 1302.35443 Root MSE =
.5566
------------------------------------------------------------------------
------
| Robust
potincm_fo | Coef. Std. Err. t P>|t| [95% Conf.
Interval]
-------------+----------------------------------------------------------
-------------+------
dagro | .0813442 .070368 1.16 0.248 -.0566143
.2193027
lnpland_manz | -.0382054 .0165208 -2.31 0.021 -.070595
-.0058159
agehead | .1004971 .040134 2.50 0.012 .0218132
.1791809
ageheadsqr | -.0009395 .0006719 -1.40 0.162 -.0022568
.0003778
kids14less | -.0212255 .0166119 -1.28 0.201 -.0537936
.0113425
potincm80s~o | .9976588 .0131996 75.58 0.000 .9717807
1.023537
------------------------------------------------------------------------
------
Included instruments: dagro lnpland_manz agehead ageheadsqr kids14less
potincm80s_fo
------------------------------------------------------------------------
------
Partial R-squared of excluded instruments: 0.3043
Test of excluded instruments:
F( 1, 4204) = 5712.73
Prob > F = 0.0000
Summary results for first-stage regressions
-------------------------------------------
Variable | Shea Partial R2 | Partial R2 | F( 1, 4204)
P-value
potincm_fo | 0.3043 | 0.3043 | 5712.73
0.0000
NB: first-stage F-stat heteroskedasticity-robust
Underidentification tests
Ho: matrix of reduced form coefficients has rank=K-1 (underidentified)
Ha: matrix has rank=K (identified)
Chi-sq(1) P-value
Anderson canon. corr. -N*ln(1-CCEV) LR stat. 1527.72 0.0000
Cragg-Donald N*CDEV statistic 1841.71 0.0000
Robust chi-square statistic 5720.89 0.0000
Weak identification tests
Ho: equation is weakly identified
Cragg-Donald (N-L)*CDEV/L1 F-statistic 1839.09
Robust F-statistic 5712.73
See main output for Cragg-Donald weak id test critical values
NB: Anderson and Cragg-Donald under- and weak identification stats not
robust
Robust identification stats heteroskedasticity-robust
Weak-instrument-robust inference
Tests of joint significance of endogenous regressors B1 in main equation
Ho: B1=0 and overidentifying restrictions are valid
Anderson-Rubin test F(1,4204)=13.88 P-val=0.0002
Anderson-Rubin test Chi-sq(1)=13.90 P-val=0.0002
Stock-Wright S statistic Chi-sq(1)=5.61 P-val=0.0178
NB: Test statistics heteroskedasticity-robust
Number of observations N = 4959
Number of regressors K = 6
Number of instruments L = 6
Number of excluded instruments L1 = 1
IV (2SLS) estimation
--------------------
Statistics robust to heteroskedasticity
Number of obs =
4959
F( 6, 4204) =
6.80
Prob > F =
0.0000
Total (centered) SS = 119.6871152 Centered R2 =
0.0067
Total (uncentered) SS = 119.6871152 Uncentered R2 =
0.0067
Residual SS = 118.8838886 Root MSE =
.168
------------------------------------------------------------------------
------
| Robust
dabus_sum | Coef. Std. Err. z P>|z| [95% Conf.
Interval]
-------------+----------------------------------------------------------
-------------+------
potincm_fo | -.0643263 .0173661 -3.70 0.000 -.0983631
-.0302894
dagro | -.1377199 .0407396 -3.38 0.001 -.217568
-.0578717
lnpland_manz | -.0049845 .0154831 -0.32 0.748 -.0353308
.0253617
agehead | .0423621 .0139741 3.03 0.002 .0149734
.0697507
ageheadsqr | -.0004116 .0002127 -1.94 0.053 -.0008284
5.30e-06
kids14less | .0042853 .0051692 0.83 0.407 -.0058461
.0144167
------------------------------------------------------------------------
------
Anderson canon. corr. LR statistic (underidentification test):
1527.719
Chi-sq(1) P-val =
0.0000
Test statistic(s) not robust
------------------------------------------------------------------------
------
Cragg-Donald F statistic (weak identification test):
1839.088
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
Test statistic(s) not robust
Source: Stock-Yogo (2005). Reproduced by permission.
------------------------------------------------------------------------
------
Hansen J statistic (overidentification test of all instruments):
0.000
(equation exactly
identified)
------------------------------------------------------------------------
------
Instrumented: potincm_fo
Included instruments: dagro lnpland_manz agehead ageheadsqr kids14less
Excluded instruments: potincm80s_fo Partialled-out (FWL): dcomm1_91
dcomm1_92 dcomm1_93 dcomm1_94 dcomm1_95
dcomm1_96 dcomm1_97 dcomm1_98 dcomm2_91 dcomm2_92
dcomm2_93 dcomm2_94 dcomm2_95 dcomm2_96 dcomm2_97
dcomm2_98 dcomm3_91 dcomm3_92 dcomm3_93 dcomm3_94
dcomm3_95 dcomm3_96 dcomm3_97 dcomm3_98 dcomm4_91
dcomm4_92 dcomm4_93 dcomm4_94 dcomm4_95 dcomm4_96
dcomm4_97 dcomm4_98 dcomm5_91 dcomm5_92 dcomm5_93
dcomm5_94 dcomm5_95 dcomm5_96 dcomm5_98 dcomm5_99
dcomm7_91 dcomm7_92 dcomm7_93 dcomm7_94 dcomm7_95
dcomm7_96 dcomm7_97 dcomm7_99 dcomm8_91 dcomm8_92
dcomm8_93 dcomm8_94 dcomm8_95 dcomm8_96 dcomm8_97
dcomm8_98 dyr2 dyr3 dyr4 dyr5 dyr6 dyr7 dyr8
nb: variable counts and small-sample adjustments
do not include partialled-out variables.
Dropped collinear: dcomm1_99 dcomm2_99 dcomm3_99 dcomm4_99 dcomm5_97
dcomm7_98 dcomm8_99 dyr9
------------------------------------------------------------------------
------
OLDER HEADS SAMPLE :
. xtivreg2 dabus_sum
> dagro lnpland_manz `agesq' `hhcharact' `ctfe' `te' ( potincm_fo =
> `inst80s_fo' ) if dagehead4065 ==1 , fe i(nidhogar) r first ffirst rf
> fwl(`ctfe' `te') ;
Warning - singleton groups detected. 56 observation(s) not used.
Warning - collinearities detected
Vars dropped: dcomm1_99 dcomm2_95 dcomm3_96 dcomm4_97 dcomm5_98
dcomm7_99
dcomm8_99 dyr9
FIXED EFFECTS ESTIMATION
------------------------
Number of groups = 830 Obs per group: min =
2
avg =
7.0
max =
9
Reduced-form regression: dabus_sum
----------------------------------
FIXED EFFECTS ESTIMATION
------------------------
Number of groups = 830 Obs per group: min =
2
avg =
7.0
max =
9
OLS estimation
--------------
Statistics robust to heteroskedasticity
Number of obs =
5840
F( 6, 5004) =
9.95
Prob > F =
0.0000
Total (centered) SS = 130.6892353 Centered R2 =
0.0200
Total (uncentered) SS = 130.6892353 Uncentered R2 =
0.0200
Residual SS = 128.0706507 Root MSE =
.16
------------------------------------------------------------------------
------
| Robust
dabus_sum | Coef. Std. Err. t P>|t| [95% Conf.
Interval]
-------------+----------------------------------------------------------
-------------+------
dagro | -.1704666 .0498948 -3.42 0.001 -.2682824
-.0726509
lnpland_manz | -.0023589 .0059152 -0.40 0.690 -.0139553
.0092375
agehead | .0638039 .0102197 6.24 0.000 .0437689
.0838389
ageheadsqr | -.0005301 .0000889 -5.97 0.000 -.0007043
-.0003559
kids14less | .0099594 .0049148 2.03 0.043 .0003242
.0195947
potincm80s~o | .0093771 .0038328 2.45 0.014 .0018631
.0168911
------------------------------------------------------------------------
------
Included instruments: dagro lnpland_manz agehead ageheadsqr kids14less
potincm80s_fo
------------------------------------------------------------------------
------
First-stage regressions
-----------------------
First-stage regression of potincm_fo:
FIXED EFFECTS ESTIMATION
------------------------
Number of groups = 830 Obs per group: min =
2
avg =
7.0
max =
9
OLS estimation
--------------
Statistics robust to heteroskedasticity
Number of obs =
5840
F( 6, 5004) =
7.64
Prob > F =
0.0000
Total (centered) SS = 8142.451572 Centered R2 =
0.0463
Total (uncentered) SS = 8142.451572 Uncentered R2 =
0.0463
Residual SS = 7765.373403 Root MSE =
1.246
------------------------------------------------------------------------
------
| Robust
potincm_fo | Coef. Std. Err. t P>|t| [95% Conf.
Interval]
-------------+----------------------------------------------------------
-------------+------
dagro | .0590806 .1640592 0.36 0.719 -.2625472
.3807085
lnpland_manz | -.1176091 .0390987 -3.01 0.003 -.1942598
-.0409584
agehead | .0332028 .0682443 0.49 0.627 -.1005861
.1669916
ageheadsqr | .0003195 .0006371 0.50 0.616 -.0009294
.0015684
kids14less | -.0352024 .031388 -1.12 0.262 -.0967366
.0263317
potincm80s~o | .5135334 .1413639 3.63 0.000 .2363982
.7906685
------------------------------------------------------------------------
------
Included instruments: dagro lnpland_manz agehead ageheadsqr kids14less
potincm80s_fo
------------------------------------------------------------------------
------
Partial R-squared of excluded instruments: 0.0446
Test of excluded instruments:
F( 1, 5004) = 13.20
Prob > F = 0.0003
Summary results for first-stage regressions
-------------------------------------------
Variable | Shea Partial R2 | Partial R2 | F( 1, 5004)
P-value
potincm_fo | 0.0446 | 0.0446 | 13.20
0.0003
NB: first-stage F-stat heteroskedasticity-robust
Underidentification tests
Ho: matrix of reduced form coefficients has rank=K-1 (underidentified)
Ha: matrix has rank=K (identified)
Chi-sq(1) P-value
Anderson canon. corr. -N*ln(1-CCEV) LR stat. 228.47 0.0000
Cragg-Donald N*CDEV statistic 233.76 0.0000
Robust chi-square statistic 13.21 0.0003
Weak identification tests
Ho: equation is weakly identified
Cragg-Donald (N-L)*CDEV/L1 F-statistic 233.48
Robust F-statistic 13.20
See main output for Cragg-Donald weak id test critical values
NB: Anderson and Cragg-Donald under- and weak identification stats not
robust
Robust identification stats heteroskedasticity-robust
Weak-instrument-robust inference
Tests of joint significance of endogenous regressors B1 in main equation
Ho: B1=0 and overidentifying restrictions are valid
Anderson-Rubin test F(1,5004)=5.99 P-val=0.0145
Anderson-Rubin test Chi-sq(1)=5.99 P-val=0.0144
Stock-Wright S statistic Chi-sq(1)=6.48 P-val=0.0109
NB: Test statistics heteroskedasticity-robust
Number of observations N = 5840
Number of regressors K = 6
Number of instruments L = 6
Number of excluded instruments L1 = 1
IV (2SLS) estimation
--------------------
Statistics robust to heteroskedasticity
Number of obs =
5840
F( 6, 5004) =
9.91
Prob > F =
0.0000
Total (centered) SS = 130.6892353 Centered R2 =
0.0038
Total (uncentered) SS = 130.6892353 Uncentered R2 =
0.0038
Residual SS = 130.1981904 Root MSE =
.1612
------------------------------------------------------------------------
------
| Robust
dabus_sum | Coef. Std. Err. z P>|z| [95% Conf.
Interval]
-------------+----------------------------------------------------------
-------------+------
potincm_fo | .0182599 .0079422 2.30 0.021 .0026936
.0338263
dagro | -.1715454 .0502306 -3.42 0.001 -.2699957
-.0730952
lnpland_manz | -.0002114 .0060272 -0.04 0.972 -.0120245
.0116017
agehead | .0631976 .0102148 6.19 0.000 .043177
.0832183
ageheadsqr | -.000536 .0000894 -6.00 0.000 -.0007111
-.0003608
kids14less | .0106022 .0049372 2.15 0.032 .0009254
.020279
------------------------------------------------------------------------
------
Anderson canon. corr. LR statistic (underidentification test):
228.468
Chi-sq(1) P-val =
0.0000
Test statistic(s) not robust
------------------------------------------------------------------------
------
Cragg-Donald F statistic (weak identification test):
233.478
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
Test statistic(s) not robust
Source: Stock-Yogo (2005). Reproduced by permission.
------------------------------------------------------------------------
------
Hansen J statistic (overidentification test of all instruments):
0.000
(equation exactly
identified)
------------------------------------------------------------------------
------
Instrumented: potincm_fo
Included instruments: dagro lnpland_manz agehead ageheadsqr kids14less
Excluded instruments: potincm80s_fo Partialled-out (FWL): dcomm1_91
dcomm1_92 dcomm1_93 dcomm1_94 dcomm1_95
dcomm1_96 dcomm1_97 dcomm1_98 dcomm2_91 dcomm2_92
dcomm2_93 dcomm2_94 dcomm2_96 dcomm2_97 dcomm2_98
dcomm2_99 dcomm3_91 dcomm3_92 dcomm3_93 dcomm3_94
dcomm3_95 dcomm3_97 dcomm3_98 dcomm3_99 dcomm4_91
dcomm4_92 dcomm4_93 dcomm4_94 dcomm4_95 dcomm4_96
dcomm4_98 dcomm4_99 dcomm5_91 dcomm5_92 dcomm5_93
dcomm5_94 dcomm5_95 dcomm5_96 dcomm5_97 dcomm5_99
dcomm7_91 dcomm7_92 dcomm7_93 dcomm7_94 dcomm7_95
dcomm7_96 dcomm7_97 dcomm7_98 dcomm8_91 dcomm8_92
dcomm8_93 dcomm8_94 dcomm8_95 dcomm8_96 dcomm8_97
dcomm8_98 dyr2 dyr3 dyr4 dyr5 dyr6 dyr7 dyr8
nb: variable counts and small-sample adjustments
do not include partialled-out variables.
Dropped collinear: dcomm1_99 dcomm2_95 dcomm3_96 dcomm4_97 dcomm5_98
dcomm7_99 dcomm8_99 dyr9
------------------------------------------------------------------------
------
*
* For searches and help try:
* http://www.stata.com/support/faqs/res/findit.html
* http://www.stata.com/support/statalist/faq
* http://www.ats.ucla.edu/stat/stata/
********************************************************************************
ADVERTENCIA: La informaci�n contenida en esta transmisi�n, y en cualquier archivo adjunto, est� sujeta a reserva legal conforme a la normativa aplicable al Banco Central de Chile, y no puede ser usada o difundida por personas distintas de su o sus destinatarios. Si usted ha recibido esta transmisi�n por error, por favor notifique inmediatamente al remitente respondiendo por este mismo medio y elim�nela de su sistema.
El Banco Central de Chile no se har� responsable de la exactitud y veracidad de la informaci�n contenida en este mensaje, as� como de su modificaci�n, copia, divulgaci�n o reenv�o, total o parcial. Su uso no autorizado puede ser sancionado de conformidad con las leyes chilenas.
El Banco Central de Chile transmite sus decisiones a trav�s de comunicados oficiales, los que pone a disposici�n del p�blico en su p�gina de Internet: www.bcentral.cl
DISCLAIMER: The information contained in this email or any attached file, is subject to legal privilege pursuant to the laws and regulations applicable to the Central Bank of Chile , and may not be used or disseminated by any person other than its intended recipients. If you have received this transmission in error, please notify the sender immediately by reply to this email address and delete it from your system.
The Central Bank of Chile shall not be liable for the accuracy or authenticity of the contents of this message, whether amended, copied, forwarded or disclosed in any form, in whole or in part. Please note that unauthorized use may be penalized in conformity with the Chilean law.
The Central Bank of Chile communicates its decisions by official releases, and
makes them available to the public in its WebPages: www.bcentral.cl
*
* For searches and help try:
* http://www.stata.com/support/faqs/res/findit.html
* http://www.stata.com/support/statalist/faq
* http://www.ats.ucla.edu/stat/stata/
