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 vs. Manual IV
From
Erkal Ersoy <[email protected]>
To
[email protected]
Subject
Re: st: RE: ivreg2 vs. Manual IV
Date
Mon, 7 Mar 2011 19:15:53 +0000
Professor Schaffer,
Thank you for your quick response. I am using Stata 11.1 and ivreg2's
version 3.0.06, 30Jan2011.
1. I tried the partial option and only the R^2 terms change--all
coefficients and z-stats stay the same (output below).
2. "ivregress 2sls loghrearn age agesq YR* (educ=QTR*), robust" gave
the same output as ivreg2 (none of the weak ID and Sargan stats, of
course)
"ivreg loghrearn age agesq YR* (educ=QTR*), robust" gives almost the
same output as the one I get doing the 1st and 2nd stage regressions
manually. The coefficients are the same on educ, age and agesq. But
with -ivreg-, educ, age and agesq are all significant at the 5%
level--using the manual way, agesq was not significant.
3. When I do "ivreg2 loghrearn age agesq YR* (educ=yhat), robust" I
get the same output as "ivreg2 loghrearn age agesq YR* (educ=QTR*),
robust"
With "ivregress 2sls loghrearn age agesq YR* (educ=yhat), robust" I
get the same output as "ivreg2 loghrearn age agesq YR* (educ=QTR*),
robust"
Lastly, with "ivreg loghrearn age agesq YR* (educ=yhat), robust" I get
the same output as "ivreg loghrearn age agesq YR* (educ=QTR*), robust"
I am still confused as to which approach I should be using to get as
robust estimates as possible. Which one would you recommend?
Best,
Erkal
Output:
. ivreg2 loghrearn age agesq YR* (educ=QTR*), robust partial(YR*)
Warning - collinearities detected
Vars dropped: YR57 YR58
IV (2SLS) estimation
--------------------
Estimates efficient for homoskedasticity only
Statistics robust to heteroskedasticity
Number of obs = 1930
F( 3, 1899) = 10.81
Prob > F = 0.0000
Total (centered) SS = 400.8160242 Centered R2 = 0.2049
Total (uncentered) SS = 400.8160242 Uncentered R2 = 0.2049
Residual SS = 318.6886262 Root MSE = .4064
------------------------------------------------------------------------------
| Robust
loghrearn | Coef. Std. Err. z P>|z| [95% Conf. Interval]
-------------+----------------------------------------------------------------
educ | .0775439 .0160336 4.84 0.000 .0461186 .1089692
age | .0960862 .1423336 0.68 0.500 -.1828825 .3750549
agesq | -.0010696 .0017911 -0.60 0.550 -.0045801 .0024409
------------------------------------------------------------------------------
Underidentification test (Kleibergen-Paap rk LM statistic): 87.887
Chi-sq(87) P-val = 0.4532
------------------------------------------------------------------------------
Weak identification test (Cragg-Donald Wald F statistic): 1.147
(Kleibergen-Paap rk Wald F statistic): 1.139
Stock-Yogo weak ID test critical values: 5% maximal IV relative bias 21.12
10% maximal IV relative bias 10.91
20% maximal IV relative bias 5.69
30% maximal IV relative bias 3.92
10% maximal IV size 222.24
15% maximal IV size 113.33
20% maximal IV size 76.67
25% maximal IV size 58.36
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): 82.395
Chi-sq(86) P-val = 0.5901
------------------------------------------------------------------------------
Instrumented: educ
Included instruments: age agesq
Excluded instruments: QTR230 QTR231 QTR232 QTR233 QTR234 QTR235 QTR236 QTR237
QTR238 QTR239 QTR240 QTR241 QTR242 QTR243 QTR244 QTR245
QTR246 QTR247 QTR248 QTR249 QTR250 QTR251 QTR252 QTR253
QTR254 QTR255 QTR256 QTR257 QTR258 QTR330 QTR331 QTR332
QTR333 QTR334 QTR335 QTR336 QTR337 QTR338 QTR339 QTR340
QTR341 QTR342 QTR343 QTR344 QTR345 QTR346 QTR347 QTR348
QTR349 QTR350 QTR351 QTR352 QTR353 QTR354 QTR355 QTR356
QTR357 QTR358 QTR430 QTR431 QTR432 QTR433 QTR434 QTR435
QTR436 QTR437 QTR438 QTR439 QTR440 QTR441 QTR442 QTR443
QTR444 QTR445 QTR446 QTR447 QTR448 QTR449 QTR450 QTR451
QTR452 QTR453 QTR454 QTR455 QTR456 QTR457 QTR458
Partialled-out: YR30 YR31 YR32 YR33 YR34 YR35 YR36 YR37 YR38 YR39 YR40
YR41 YR42 YR43 YR44 YR45 YR46 YR47 YR48 YR49 YR50 YR51
YR52 YR53 YR54 YR55 YR56 _cons
nb: small-sample adjustments account for
partialled-out variables
Dropped collinear: YR57 YR58
------------------------------------------------------------------------------
. ivregress 2sls loghrearn age agesq YR* (educ=QTR*), robust
note: YR57 omitted because of collinearity
note: YR58 omitted because of collinearity
Instrumental variables (2SLS) regression Number of obs = 1930
Wald chi2(30) = 312.53
Prob > chi2 = 0.0000
R-squared = 0.2791
Root MSE = .40635
------------------------------------------------------------------------------
| Robust
loghrearn | Coef. Std. Err. z P>|z| [95% Conf. Interval]
-------------+----------------------------------------------------------------
educ | .0775439 .0160336 4.84 0.000 .0461186 .1089692
age | .0960863 .1423337 0.68 0.500 -.1828826 .3750552
agesq | -.0010696 .0017911 -0.60 0.550 -.0045801 .0024409
YR30 | .0729702 .1068258 0.68 0.495 -.1364044 .2823448
YR31 | .0770476 .1309373 0.59 0.556 -.1795848 .3336799
YR32 | .0973404 .1634284 0.60 0.551 -.2229734 .4176542
YR33 | .141803 .1929298 0.73 0.462 -.2363325 .5199385
YR34 | -.1038156 .2259734 -0.46 0.646 -.5467153 .3390841
YR35 | -.0904715 .2649789 -0.34 0.733 -.6098205 .4288776
YR36 | .0211239 .2888762 0.07 0.942 -.5450631 .5873109
YR37 | -.0095477 .309436 -0.03 0.975 -.6160311 .5969358
YR38 | -.1243089 .3284458 -0.38 0.705 -.7680508 .519433
YR39 | .031897 .3397094 0.09 0.925 -.6339212 .6977152
YR40 | -.0124043 .3501857 -0.04 0.972 -.6987556 .673947
YR41 | -.0104207 .3627671 -0.03 0.977 -.7214312 .7005898
YR42 | -.0702436 .3685106 -0.19 0.849 -.7925111 .652024
YR43 | .0575309 .374727 0.15 0.878 -.6769206 .7919823
YR44 | .0096373 .3702528 0.03 0.979 -.7160449 .7353195
YR45 | .0027419 .3662022 0.01 0.994 -.7150012 .7204851
YR46 | -.0634878 .3580731 -0.18 0.859 -.7652981 .6383225
YR47 | -.0433442 .3435527 -0.13 0.900 -.716695 .6300067
YR48 | .0054649 .3285693 0.02 0.987 -.6385191 .6494489
YR49 | -.0997214 .3100031 -0.32 0.748 -.7073162 .5078735
YR50 | -.0773026 .2898275 -0.27 0.790 -.645354 .4907489
YR51 | .0145725 .2641776 0.06 0.956 -.5032062 .5323511
YR52 | -.0292526 .2321381 -0.13 0.900 -.484235 .4257298
YR53 | .0184267 .1987075 0.09 0.926 -.3710328 .4078863
YR54 | -.0401415 .1639573 -0.24 0.807 -.3614919 .2812088
YR55 | -.0174243 .1241563 -0.14 0.888 -.2607661 .2259175
YR56 | -.0569665 .0846968 -0.67 0.501 -.2229692 .1090362
YR57 | (omitted)
YR58 | (omitted)
_cons | -.7113387 2.494937 -0.29 0.776 -5.601325 4.178648
------------------------------------------------------------------------------
Instrumented: educ
Instruments: age agesq YR30 YR31 YR32 YR33 YR34 YR35 YR36 YR37 YR38 YR39
YR40 YR41 YR42 YR43 YR44 YR45 YR46 YR47 YR48 YR49 YR50 YR51
YR52 YR53 YR54 YR55 YR56 QTR230 QTR231 QTR232 QTR233 QTR234
QTR235 QTR236 QTR237 QTR238 QTR239 QTR240 QTR241 QTR242
QTR243 QTR244 QTR245 QTR246 QTR247 QTR248 QTR249 QTR250
QTR251 QTR252 QTR253 QTR254 QTR255 QTR256 QTR257 QTR258
QTR330 QTR331 QTR332 QTR333 QTR334 QTR335 QTR336 QTR337
QTR338 QTR339 QTR340 QTR341 QTR342 QTR343 QTR344 QTR345
QTR346 QTR347 QTR348 QTR349 QTR350 QTR351 QTR352 QTR353
QTR354 QTR355 QTR356 QTR357 QTR358 QTR430 QTR431 QTR432
QTR433 QTR434 QTR435 QTR436 QTR437 QTR438 QTR439 QTR440
QTR441 QTR442 QTR443 QTR444 QTR445 QTR446 QTR447 QTR448
QTR449 QTR450 QTR451 QTR452 QTR453 QTR454 QTR455 QTR456
QTR457 QTR458
. ivreg loghrearn age agesq YR* (educ=QTR*), robust
Instrumental variables (2SLS) regression Number of obs = 1930
F( 30, 1899) = 10.25
Prob > F = 0.0000
R-squared = 0.2791
Root MSE = .40966
------------------------------------------------------------------------------
| Robust
loghrearn | Coef. Std. Err. t P>|t| [95% Conf. Interval]
-------------+----------------------------------------------------------------
educ | .0775439 .0161639 4.80 0.000 .045843 .1092449
age | .0694072 .0275518 2.52 0.012 .0153722 .1234423
agesq | -.0007319 .0003619 -2.02 0.043 -.0014417 -.0000221
YR30 | .0824261 .0942624 0.87 0.382 -.1024427 .2672948
YR31 | .0952839 .087805 1.09 0.278 -.0769205 .2674882
YR32 | .1236817 .0844014 1.47 0.143 -.0418475 .2892109
YR33 | .175574 .0761246 2.31 0.021 .0262772 .3248707
YR34 | -.0632905 .082945 -0.76 0.446 -.2259633 .0993823
YR35 | -.0438676 .1183119 -0.37 0.711 -.2759026 .1881675
YR36 | .0731311 .1048125 0.70 0.485 -.1324285 .2786908
YR37 | .0471875 .0985247 0.48 0.632 -.1460405 .2404155
YR38 | -.0635212 .0990568 -0.64 0.521 -.2577928 .1307504
YR39 | .0960618 .0859788 1.12 0.264 -.072561 .2646846
YR40 | .0544622 .0782422 0.70 0.486 -.0989875 .2079118
YR41 | .058472 .085802 0.68 0.496 -.1098041 .2267481
YR42 | (omitted)
YR43 | .1284498 .1083614 1.19 0.236 -.08407 .3409697
YR44 | .0805562 .0827555 0.97 0.330 -.081745 .2428575
YR45 | .0729855 .0897922 0.81 0.416 -.1031163 .2490873
YR46 | .005405 .0832444 0.06 0.948 -.1578551 .168665
YR47 | .0235223 .0713527 0.33 0.742 -.1164156 .1634602
YR48 | .0696297 .0677743 1.03 0.304 -.0632903 .2025496
YR49 | -.0389337 .06661 -0.58 0.559 -.1695702 .0917028
YR50 | -.0205674 .0725417 -0.28 0.777 -.1628372 .1217024
YR51 | .0665797 .0719262 0.93 0.355 -.0744829 .2076424
YR52 | .0173513 .0595863 0.29 0.771 -.0995101 .1342126
YR53 | .0589519 .0551962 1.07 0.286 -.0492997 .1672034
YR54 | -.0063706 .0538644 -0.12 0.906 -.1120102 .099269
YR55 | .008917 .0516867 0.17 0.863 -.0924517 .1102857
YR56 | -.0387302 .0481101 -0.81 0.421 -.1330844 .055624
YR57 | .0094559 .0500105 0.19 0.850 -.0886254 .1075371
YR58 | (omitted)
_cons | -.255431 .4852748 -0.53 0.599 -1.207159 .6962967
------------------------------------------------------------------------------
Instrumented: educ
Instruments: age agesq YR30 YR31 YR32 YR33 YR34 YR35 YR36 YR37 YR38 YR39
YR40 YR41 YR42 YR43 YR44 YR45 YR46 YR47 YR48 YR49 YR50 YR51
YR52 YR53 YR54 YR55 YR56 YR57 YR58 QTR230 QTR231 QTR232
QTR233 QTR234 QTR235 QTR236 QTR237 QTR238 QTR239 QTR240
QTR241 QTR242 QTR243 QTR244 QTR245 QTR246 QTR247 QTR248
QTR249 QTR250 QTR251 QTR252 QTR253 QTR254 QTR255 QTR256
QTR257 QTR258 QTR330 QTR331 QTR332 QTR333 QTR334 QTR335
QTR336 QTR337 QTR338 QTR339 QTR340 QTR341 QTR342 QTR343
QTR344 QTR345 QTR346 QTR347 QTR348 QTR349 QTR350 QTR351
QTR352 QTR353 QTR354 QTR355 QTR356 QTR357 QTR358 QTR430
QTR431 QTR432 QTR433 QTR434 QTR435 QTR436 QTR437 QTR438
QTR439 QTR440 QTR441 QTR442 QTR443 QTR444 QTR445 QTR446
QTR447 QTR448 QTR449 QTR450 QTR451 QTR452 QTR453 QTR454
QTR455 QTR456 QTR457 QTR458
------------------------------------------------------------------------------
. ivreg2 loghrearn age agesq YR* (educ=yhat), robust
Warning - collinearities detected
Vars dropped: YR57 YR58
IV (2SLS) estimation
--------------------
Estimates efficient for homoskedasticity only
Statistics robust to heteroskedasticity
Number of obs = 1930
F( 30, 1899) = 10.25
Prob > F = 0.0000
Total (centered) SS = 442.0938306 Centered R2 = 0.2791
Total (uncentered) SS = 10400.84512 Uncentered R2 = 0.9694
Residual SS = 318.6886296 Root MSE = .4064
------------------------------------------------------------------------------
| Robust
loghrearn | Coef. Std. Err. z P>|z| [95% Conf. Interval]
-------------+----------------------------------------------------------------
educ | .0775439 .0160336 4.84 0.000 .0461186 .1089692
age | .0960844 .1423327 0.68 0.500 -.1828826 .3750515
agesq | -.0010696 .0017911 -0.60 0.550 -.0045801 .0024409
YR30 | .0729708 .1068256 0.68 0.495 -.1364036 .2823452
YR31 | .0770488 .1309368 0.59 0.556 -.1795827 .3336803
YR32 | .0973422 .1634277 0.60 0.551 -.2229702 .4176545
YR33 | .1418053 .1929288 0.74 0.462 -.2363282 .5199388
YR34 | -.1038129 .2259721 -0.46 0.646 -.5467101 .3390843
YR35 | -.0904684 .2649774 -0.34 0.733 -.6098146 .4288779
YR36 | .0211273 .2888746 0.07 0.942 -.5450564 .5873111
YR37 | -.0095438 .3094342 -0.03 0.975 -.6160237 .596936
YR38 | -.1243048 .3284438 -0.38 0.705 -.7680429 .5194333
YR39 | .0319013 .3397073 0.09 0.925 -.6339128 .6977154
YR40 | -.0123998 .3501835 -0.04 0.972 -.6987468 .6739472
YR41 | -.0104161 .3627649 -0.03 0.977 -.7214222 .70059
YR42 | -.0702389 .3685083 -0.19 0.849 -.7925019 .6520242
YR43 | .0575356 .3747247 0.15 0.878 -.6769114 .7919825
YR44 | .009642 .3702505 0.03 0.979 -.7160356 .7353196
YR45 | .0027466 .3661999 0.01 0.994 -.7149921 .7204852
YR46 | -.0634832 .3580708 -0.18 0.859 -.7652891 .6383227
YR47 | -.0433398 .3435505 -0.13 0.900 -.7166864 .6300068
YR48 | .0054691 .3285672 0.02 0.987 -.6385108 .649449
YR49 | -.0997174 .3100011 -0.32 0.748 -.7073084 .5078736
YR50 | -.0772989 .2898257 -0.27 0.790 -.6453468 .490749
YR51 | .0145758 .264176 0.06 0.956 -.5031997 .5323513
YR52 | -.0292497 .2321367 -0.13 0.900 -.4842293 .4257298
YR53 | .0184292 .1987063 0.09 0.926 -.3710279 .4078863
YR54 | -.0401395 .1639563 -0.24 0.807 -.3614879 .2812089
YR55 | -.0174228 .1241556 -0.14 0.888 -.2607633 .2259176
YR56 | -.0569656 .0846964 -0.67 0.501 -.2229675 .1090363
_cons | -.7113064 2.494921 -0.29 0.776 -5.601261 4.178649
------------------------------------------------------------------------------
Underidentification test (Kleibergen-Paap rk LM statistic): 73.415
Chi-sq(1) P-val = 0.0000
------------------------------------------------------------------------------
Weak identification test (Cragg-Donald Wald F statistic): 104.504
(Kleibergen-Paap rk Wald F statistic): 86.424
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: educ
Included instruments: age agesq YR30 YR31 YR32 YR33 YR34 YR35 YR36 YR37 YR38
YR39 YR40 YR41 YR42 YR43 YR44 YR45 YR46 YR47 YR48 YR49
YR50 YR51 YR52 YR53 YR54 YR55 YR56
Excluded instruments: yhat
Dropped collinear: YR57 YR58
------------------------------------------------------------------------------
. ivregress 2sls loghrearn age agesq YR* (educ=yhat), robust
note: YR57 omitted because of collinearity
note: YR58 omitted because of collinearity
Instrumental variables (2SLS) regression Number of obs = 1930
Wald chi2(30) = 312.53
Prob > chi2 = 0.0000
R-squared = 0.2791
Root MSE = .40635
------------------------------------------------------------------------------
| Robust
loghrearn | Coef. Std. Err. z P>|z| [95% Conf. Interval]
-------------+----------------------------------------------------------------
educ | .0775439 .0160336 4.84 0.000 .0461186 .1089692
age | .0960863 .1423337 0.68 0.500 -.1828826 .3750552
agesq | -.0010696 .0017911 -0.60 0.550 -.0045801 .0024409
YR30 | .0729702 .1068258 0.68 0.495 -.1364044 .2823448
YR31 | .0770476 .1309373 0.59 0.556 -.1795848 .3336799
YR32 | .0973404 .1634284 0.60 0.551 -.2229734 .4176542
YR33 | .141803 .1929298 0.73 0.462 -.2363325 .5199385
YR34 | -.1038156 .2259734 -0.46 0.646 -.5467153 .3390841
YR35 | -.0904715 .2649789 -0.34 0.733 -.6098205 .4288776
YR36 | .0211239 .2888762 0.07 0.942 -.5450631 .5873109
YR37 | -.0095477 .309436 -0.03 0.975 -.6160311 .5969358
YR38 | -.1243089 .3284458 -0.38 0.705 -.7680508 .519433
YR39 | .031897 .3397094 0.09 0.925 -.6339212 .6977152
YR40 | -.0124043 .3501857 -0.04 0.972 -.6987556 .673947
YR41 | -.0104207 .3627671 -0.03 0.977 -.7214312 .7005898
YR42 | -.0702436 .3685106 -0.19 0.849 -.7925111 .652024
YR43 | .0575309 .374727 0.15 0.878 -.6769206 .7919823
YR44 | .0096373 .3702528 0.03 0.979 -.7160449 .7353194
YR45 | .0027419 .3662022 0.01 0.994 -.7150012 .7204851
YR46 | -.0634878 .3580731 -0.18 0.859 -.7652981 .6383225
YR47 | -.0433442 .3435527 -0.13 0.900 -.716695 .6300067
YR48 | .0054649 .3285693 0.02 0.987 -.6385191 .6494489
YR49 | -.0997214 .3100031 -0.32 0.748 -.7073162 .5078735
YR50 | -.0773026 .2898275 -0.27 0.790 -.6453541 .4907489
YR51 | .0145725 .2641776 0.06 0.956 -.5032062 .5323511
YR52 | -.0292526 .2321381 -0.13 0.900 -.484235 .4257298
YR53 | .0184267 .1987075 0.09 0.926 -.3710328 .4078863
YR54 | -.0401415 .1639573 -0.24 0.807 -.3614919 .2812088
YR55 | -.0174243 .1241563 -0.14 0.888 -.2607661 .2259175
YR56 | -.0569665 .0846968 -0.67 0.501 -.2229692 .1090362
YR57 | (omitted)
YR58 | (omitted)
_cons | -.7113387 2.494937 -0.29 0.776 -5.601325 4.178648
------------------------------------------------------------------------------
Instrumented: educ
Instruments: age agesq YR30 YR31 YR32 YR33 YR34 YR35 YR36 YR37 YR38 YR39
YR40 YR41 YR42 YR43 YR44 YR45 YR46 YR47 YR48 YR49 YR50 YR51
YR52 YR53 YR54 YR55 YR56 yhat
. ivreg loghrearn age agesq YR* (educ=yhat), robust
Instrumental variables (2SLS) regression Number of obs = 1930
F( 30, 1899) = 10.25
Prob > F = 0.0000
R-squared = 0.2791
Root MSE = .40966
------------------------------------------------------------------------------
| Robust
loghrearn | Coef. Std. Err. t P>|t| [95% Conf. Interval]
-------------+----------------------------------------------------------------
educ | .0775439 .0161639 4.80 0.000 .045843 .1092449
age | .0694072 .0275518 2.52 0.012 .0153722 .1234423
agesq | -.0007319 .0003619 -2.02 0.043 -.0014417 -.0000221
YR30 | .0824261 .0942624 0.87 0.382 -.1024427 .2672948
YR31 | .0952839 .087805 1.09 0.278 -.0769205 .2674882
YR32 | .1236817 .0844014 1.47 0.143 -.0418475 .2892109
YR33 | .175574 .0761246 2.31 0.021 .0262772 .3248707
YR34 | -.0632905 .082945 -0.76 0.446 -.2259633 .0993823
YR35 | -.0438676 .1183119 -0.37 0.711 -.2759026 .1881675
YR36 | .0731311 .1048125 0.70 0.485 -.1324285 .2786908
YR37 | .0471875 .0985247 0.48 0.632 -.1460405 .2404155
YR38 | -.0635212 .0990568 -0.64 0.521 -.2577928 .1307504
YR39 | .0960618 .0859788 1.12 0.264 -.072561 .2646846
YR40 | .0544622 .0782422 0.70 0.486 -.0989875 .2079118
YR41 | .058472 .085802 0.68 0.496 -.1098041 .2267481
YR42 | (omitted)
YR43 | .1284498 .1083614 1.19 0.236 -.08407 .3409697
YR44 | .0805562 .0827555 0.97 0.330 -.081745 .2428575
YR45 | .0729855 .0897922 0.81 0.416 -.1031163 .2490873
YR46 | .005405 .0832444 0.06 0.948 -.1578551 .168665
YR47 | .0235223 .0713527 0.33 0.742 -.1164156 .1634602
YR48 | .0696297 .0677743 1.03 0.304 -.0632903 .2025496
YR49 | -.0389337 .06661 -0.58 0.559 -.1695702 .0917028
YR50 | -.0205674 .0725417 -0.28 0.777 -.1628372 .1217024
YR51 | .0665797 .0719262 0.93 0.355 -.0744829 .2076424
YR52 | .0173513 .0595863 0.29 0.771 -.0995101 .1342126
YR53 | .0589519 .0551962 1.07 0.286 -.0492997 .1672034
YR54 | -.0063706 .0538644 -0.12 0.906 -.1120102 .099269
YR55 | .008917 .0516867 0.17 0.863 -.0924517 .1102857
YR56 | -.0387302 .0481101 -0.81 0.421 -.1330844 .055624
YR57 | .0094559 .0500105 0.19 0.850 -.0886254 .1075371
YR58 | (omitted)
_cons | -.255431 .4852748 -0.53 0.599 -1.207159 .6962967
------------------------------------------------------------------------------
Instrumented: educ
Instruments: age agesq YR30 YR31 YR32 YR33 YR34 YR35 YR36 YR37 YR38 YR39
YR40 YR41 YR42 YR43 YR44 YR45 YR46 YR47 YR48 YR49 YR50 YR51
YR52 YR53 YR54 YR55 YR56 YR57 YR58 yhat
------------------------------------------------------------------------------
>
>
> On Mon, Mar 7, 2011 at 10:32 AM, Schaffer, Mark E <[email protected]> wrote:
>> Erkal,
>>
>> You've got a lot of dummy regressors and instruments, most of which are
>> not statistically significant, so my first guess would be something to
>> do with numerical accuracy. You should tell us, though, which versions
>> of Stata and -ivreg2- you are using.
>>
>> Here are a few things you can experiment with:
>>
>> 1. Do your results slightly change again if you partial out the year
>> dummies with the -partial- option?
>>
>> ivreg2 loghrearn age agesq YR* (educ=QTR*), robust partial(YR*)
>>
>> 2. There are two official IV routines in Stata, -ivregress- and
>> -ivreg-. The former is documented in Stata 11, the latter is not, but
>> its syntax is the same as that of -ivreg2-:
>>
>> ivregress 2sls loghrearn age agesq YR* (educ=QTR*), robust
>>
>> ivreg loghrearn age agesq YR* (educ=QTR*), robust
>>
>> The reason to try out -ivreg- is that it is implemented using -regress-.
>> For that reason, it's likely to be very accurate in the face of
>> numerical challenges.
>>
>> 3. What happens if, instead of using your QTR* instruments, you use
>> your predicted value (yhat) as your sole excluded instrument in your IV
>> estimation with -ivreg2-, -ivregress- and -ivreg-? E.g.,
>>
>> ivreg2 loghrearn age agesq YR* (educ=yhat), robust
>>
>> Cheers,
>> Mark
*
* For searches and help try:
* http://www.stata.com/help.cgi?search
* http://www.stata.com/support/statalist/faq
* http://www.ats.ucla.edu/stat/stata/