Dear Profs and colleagues,
I run 2sls, and the F test of excluded instruments is almost 3 million! It worries me. Although the tests for weak IV do not show that IV is weak, this large value... what can be the problem?
Any ideas are appreciated.
Cheers,
Paris
I run 2sls, and the F test of excluded instruments is almost 3 million! It worries me. Although the tests for weak IV do not show that IV is weak, this large value... what can be the problem?
Code:
. ivreg2 ln_labor_productivity_w (blau_indexethnic=region_avg_excl) logsize foreign_aff i.year i.sector i.region ltenur lfirmage
> multi lageworker share_9 share_12 share_uni ethnic1 ethnic2 ethnic3 ethnic4 ethnic5 ethnic6 ethnic7 ethnic8, first robust
First-stage regressions
-----------------------
First-stage regression of blau_indexethnic:
Statistics robust to heteroskedasticity
Number of obs = 1461978
---------------------------------------------------------------------------------
| Robust
blau_indexeth~c | Coef. Std. Err. t P>|t| [95% Conf. Interval]
----------------+----------------------------------------------------------------
region_avg_excl | -.9480597 .0005548 -1708.87 0.000 -.9491471 -.9469723
logsize | -.0004566 .0000213 -21.42 0.000 -.0004983 -.0004148
foreign_aff | .0009557 .000132 7.24 0.000 .000697 .0012144
|
year |
2011 | .0030839 .0000779 39.58 0.000 .0029312 .0032366
2012 | .0018291 .0000755 24.23 0.000 .0016811 .001977
2013 | .0026799 .0000762 35.17 0.000 .0025306 .0028293
2014 | .0006011 .0000741 8.11 0.000 .0004558 .0007464
2015 | .0019014 .0000746 25.50 0.000 .0017552 .0020475
2016 | .0058746 .0000743 79.07 0.000 .005729 .0060202
2017 | .0141067 .0000774 182.29 0.000 .013955 .0142584
2018 | .0241563 .0000799 302.51 0.000 .0239998 .0243129
2019 | .0299199 .000084 356.32 0.000 .0297553 .0300844
|
sector |
6 | .000509 .0000546 9.33 0.000 .000402 .0006159
7 | .0002749 .000043 6.40 0.000 .0001907 .0003591
9 | .0017633 .0000772 22.83 0.000 .0016119 .0019147
10 | .0002688 .0001237 2.17 0.030 .0000263 .0005113
11 | .0005773 .0001086 5.31 0.000 .0003644 .0007902
12 | .0004209 .0000612 6.88 0.000 .000301 .0005408
13 | .0010389 .0001027 10.12 0.000 .0008377 .0012401
|
region |
2 | -.0142636 .00004 -356.72 0.000 -.0143419 -.0141852
3 | .206585 .0001428 1446.32 0.000 .2063051 .206865
4 | .0914737 .0001634 559.89 0.000 .0911535 .0917939
5 | -.004737 .0000845 -56.06 0.000 -.0049027 -.0045714
6 | -.0469843 .0001225 -383.51 0.000 -.0472244 -.0467442
7 | .0144741 .000158 91.60 0.000 .0141644 .0147838
|
ltenur | .0007754 .0000652 11.90 0.000 .0006477 .0009031
lfirmage | -.000405 .000019 -21.30 0.000 -.0004422 -.0003677
multi | -.0000759 .0000755 -1.00 0.315 -.0002239 .0000721
lageworker | -.0012993 .0000943 -13.77 0.000 -.0014842 -.0011144
share_9 | .0005679 .0000898 6.32 0.000 .0003919 .0007438
share_12 | .0010241 .0000976 10.49 0.000 .0008328 .0012154
share_uni | .0014498 .0001079 13.43 0.000 .0012383 .0016613
ethnic1 | .0355977 .0014165 25.13 0.000 .0328214 .038374
ethnic2 | -.0737139 .0012486 -59.04 0.000 -.076161 -.0712667
ethnic3 | .045857 .0011811 38.83 0.000 .0435421 .0481719
ethnic4 | .0176868 .0001698 104.14 0.000 .0173539 .0180196
ethnic5 | .0264138 .0003238 81.57 0.000 .0257792 .0270485
ethnic6 | .0330661 .0003604 91.75 0.000 .0323597 .0337724
ethnic7 | .0389608 .0017553 22.20 0.000 .0355206 .042401
ethnic8 | .0185332 .0001801 102.91 0.000 .0181803 .0188862
_cons | .1077592 .000304 354.51 0.000 .1071634 .1083549
---------------------------------------------------------------------------------
F test of excluded instruments:
F( 1,1461937) = 2.9e+06
Prob > F = 0.0000
Sanderson-Windmeijer multivariate F test of excluded instruments:
F( 1,1461937) = 2.9e+06
Prob > F = 0.0000
.
.
.
Number of obs = 1461978
F( 40,1461937) = 6499.70
Prob > F = 0.0000
Total (centered) SS = 1173915.105 Centered R2 = 0.1620
Total (uncentered) SS = 136022757.4 Uncentered R2 = 0.9928
Residual SS = 983783.6863 Root MSE = .8203
----------------------------------------------------------------------------------
| Robust
ln_labor_produ~w | Coef. Std. Err. z P>|z| [95% Conf. Interval]
-----------------+----------------------------------------------------------------
blau_indexethnic | -.1152863 .0076515 -15.07 0.000 -.1302828 -.1002897
logsize | .1689567 .0008155 207.19 0.000 .1673585 .170555
foreign_aff | .5448005 .0055822 97.60 0.000 .5338596 .5557414
|
year |
2011 | -.0738303 .0031599 -23.36 0.000 -.0800237 -.067637
2012 | -.0732055 .0032468 -22.55 0.000 -.0795692 -.0668419
2013 | -.023119 .0032587 -7.09 0.000 -.0295059 -.0167321
2014 | -.0262547 .0032535 -8.07 0.000 -.0326315 -.0198779
2015 | -.0187217 .0031769 -5.89 0.000 -.0249483 -.012495
2016 | -.0109866 .0031328 -3.51 0.000 -.0171268 -.0048464
2017 | -.021945 .00309 -7.10 0.000 -.0280013 -.0158886
2018 | -.0470292 .0030751 -15.29 0.000 -.0530563 -.0410021
2019 | -.0694518 .0030783 -22.56 0.000 -.0754853 -.0634184
|
sector |
6 | .389331 .002249 173.11 0.000 .384923 .3937389
7 | .1154389 .0019711 58.57 0.000 .1115756 .1193022
9 | -.4035993 .0025582 -157.77 0.000 -.4086132 -.3985854
10 | -.0566752 .0056741 -9.99 0.000 -.0677963 -.0455541
11 | .2250767 .0058422 38.53 0.000 .2136262 .2365272
12 | .0391298 .0031416 12.46 0.000 .0329724 .0452873
13 | .0236135 .0040498 5.83 0.000 .0156761 .0315509
|
region |
2 | .0757906 .0017336 43.72 0.000 .0723927 .0791884
3 | .1312091 .0019546 67.13 0.000 .1273781 .1350401
4 | .0582004 .0034991 16.63 0.000 .0513421 .0650586
5 | .058269 .0031391 18.56 0.000 .0521164 .0644216
6 | .0936566 .0056656 16.53 0.000 .0825522 .1047609
7 | -.0038452 .0053333 -0.72 0.471 -.0142982 .0066078
|
ltenur | .0843379 .0028634 29.45 0.000 .0787257 .0899501
lfirmage | .0566891 .0007829 72.41 0.000 .0551545 .0582236
multi | -.0555838 .0024369 -22.81 0.000 -.06036 -.0508077
lageworker | -.0217752 .0031043 -7.01 0.000 -.0278595 -.0156909
share_9 | .1880041 .0035665 52.71 0.000 .1810139 .1949943
share_12 | .4319362 .0039801 108.52 0.000 .4241354 .439737
share_uni | .7743468 .0048596 159.34 0.000 .7648221 .7838714
ethnic1 | -.1274692 .0167966 -7.59 0.000 -.1603898 -.0945485
ethnic2 | -.2308133 .0123156 -18.74 0.000 -.2549513 -.2066752
ethnic3 | -.0063384 .0131721 -0.48 0.630 -.0321553 .0194784
ethnic4 | .0426324 .0035377 12.05 0.000 .0356987 .0495661
ethnic5 | .1053024 .0056499 18.64 0.000 .0942288 .116376
ethnic6 | -.0123884 .0047612 -2.60 0.009 -.0217201 -.0030567
ethnic7 | .1000489 .022069 4.53 0.000 .0567946 .1433033
ethnic8 | -.030844 .0034389 -8.97 0.000 -.0375842 -.0241038
_cons | 8.620184 .0113673 758.33 0.000 8.597905 8.642464
----------------------------------------------------------------------------------
Underidentification test (Kleibergen-Paap rk LM statistic): 7.3e+04
Chi-sq(1) P-val = 0.0000
------------------------------------------------------------------------------
Weak identification test (Cragg-Donald Wald F statistic): 2.4e+07
(Kleibergen-Paap rk Wald F statistic): 2.9e+06
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)
------------------------------------------------------------------------------
Instrumented: blau_indexethnic
Included instruments: logsize foreign_aff 2011.year 2012.year 2013.year
2014.year 2015.year 2016.year 2017.year 2018.year
2019.year 6.sector 7.sector 9.sector 10.sector 11.sector
12.sector 13.sector 2.region 3.region 4.region 5.region
6.region 7.region ltenur lfirmage multi lageworker share_9
share_12 share_uni ethnic1 ethnic2 ethnic3 ethnic4 ethnic5
ethnic6 ethnic7 ethnic8
Excluded instruments: region_avg_excl
------------------------------------------------------------------------------
Any ideas are appreciated.
Cheers,
Paris
