Hi,
I would like to know the reasons why the endogeneity test and
overidentification test results differ after 2SLS estimation with
-ivreg2- and -ivregress-. When GMM estimators are used, both commands
give the same endogeneity and overidentification results.
Here are the codes that I used:
webuse nlswork
sort idcode year
tsset idcode year
ivreg2 wks_ue ( tenure = hours c_city), cl(idcode) endog(tenure)
ivregress 2sls wks_ue ( tenure = hours c_city), cl(idcode)
estat endog
estat overid
Results from -ivreg2-
------------------------------------------------------------------------
------
Underidentification test (Kleibergen-Paap rk LM statistic):
201.142
Chi-sq(2) P-val =
0.0000
------------------------------------------------------------------------
------
Weak identification test (Kleibergen-Paap rk Wald F statistic):
101.210
Stock-Yogo weak ID test critical values: 10% maximal IV size
19.93
15% maximal IV size
11.59
20% maximal IV size
8.75
25% maximal IV size
7.25
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):
4.057
Chi-sq(1) P-val =
0.0440
-endog- option:
Endogeneity test of endogenous regressors:
2.825
Chi-sq(1) P-val =
0.0928
Regressors tested: tenure
------------------------------------------------------------------------
------
Results from -ivregress-
. estat endog
Tests of endogeneity
Ho: variables are exogenous
Robust regression F(1,4629) = 3.59988 (p = 0.0578)
(Adjusted for 4630 clusters in idcode)
.
. estat overid
Test of overidentifying restrictions:
Score chi2(1) = 5.30928 (p = 0.0212)
When GMM estimators are used, both commands give the same results.
ivreg2 wks_ue ( tenure = hours c_city), cl(idcode) endog(tenure) gmm
ivregress gmm wks_ue ( tenure = hours c_city), cl(idcode)
estat endog
estat overid
Thanks. Please advise me which command to use when clustered standard
errors are used.
Regards,
Kelvin
*
* 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/
