Notice: On April 23, 2014, Statalist moved from an email list to a forum, based at statalist.org.
| From | jing liu <pku.liujing@gmail.com> |
| To | statalist@hsphsun2.harvard.edu |
| Subject | st: 'mills' output is missing for heckman regression |
| Date | Thu, 13 May 2010 11:52:02 +0800 |
Hi,
I have two problems of calculating the marginal effects of
coefficients from a heckman regression.
here is my coding:
xi: heckman lnincome male i.edunew2 urban2 inearthquake certi hr07
ages2 dep_ratio i.occupation ///
if ad01==8 & hr07>=16 & inlist(od04,0,1) ///
[pweight=expweigh] ///
, select (male i.edunew2 i.hhecosta inearthquake urban2) ///
mills(mills200411)
margins male i.edunew2 urban2 inearthquake
predict selpr200411, psel
gen Dpr200411=mills200411*(mills200411+selpr200411)
display Dpr200411
problem 1, margins command doesn't work at all. the effor message is
'factor male not found in e(b)'.
does anyone know how to use margins command after heckman regression?
problem 2, then I cut off the margins command, and calculate the
marginal effects by hand. but the 'Dpr200411' doesn't show up, it
comes back with only '.' instead of a figure. Then I 'display
mills200411', the '.' appears again. But there is a output when i
'display selpr200411'.
the new code is:
xi: heckman lnincome male i.edunew2 urban2 inearthquake certi hr07
ages2 dep_ratio i.occupation ///
if ad01==8 & hr07>=16 & inlist(od04,0,1) ///
[pweight=expweigh] ///
, select (male i.edunew2 i.hhecosta inearthquake urban2) ///
mills(mills200411)
predict selpr200411, psel
gen Dpr200411=mills200411*(mills200411+selpr200411)
display Dpr200411
here is the output:
i.edunew2 _Iedunew2_0-4 (naturally coded; _Iedunew2_0 omitted)
i.occupation _Ioccupatio_0-6 (naturally coded; _Ioccupatio_0 omitted)
i.hhecosta5000 _Ihhecosta5_0-3 (naturally coded; _Ihhecosta5_0 omitted)
note: _Ioccupatio_6 omitted because of collinearity
Iteration 0: log pseudolikelihood = -29511151 (not concave)
Iteration 1: log pseudolikelihood = -27814870
Iteration 2: log pseudolikelihood = -27377953 (not concave)
Iteration 3: log pseudolikelihood = -26636133
Iteration 4: log pseudolikelihood = -26458324
Iteration 5: log pseudolikelihood = -26454536
Iteration 6: log pseudolikelihood = -26454530
Iteration 7: log pseudolikelihood = -26454530
Heckman selection model Number of obs = 8179
(regression model with sample selection) Censored obs = 6890
Uncensored obs = 1289
Wald chi2(15) = 96.10
Log pseudolikelihood = -2.65e+07 Prob > chi2 = 0.0000
------------------------------------------------------------------------------
| Robust
lnincome | Coef. Std. Err. z P>|z| [95% Conf. Interval]
-------------+----------------------------------------------------------------
lnincome |
male | -.0770914 .0420301 -1.83 0.067 -.1594688 .0052859
_Iedunew2_1 | -.151851 .0561303 -2.71 0.007 -.2618645 -.0418376
_Iedunew2_2 | -.2193169 .0574577 -3.82 0.000 -.3319319 -.1067018
_Iedunew2_3 | -.2828423 .0740978 -3.82 0.000 -.4280715 -.1376132
_Iedunew2_4 | .3236262 .2218939 1.46 0.145 -.1112778 .7585302
urban2 | .0602396 .0467554 1.29 0.198 -.0313993 .1518785
inearthquake | -.1362456 .0370262 -3.68 0.000 -.2088156 -.0636757
certi | .0316799 .0637047 0.50 0.619 -.093179 .1565388
hr07 | .0346725 .0086475 4.01 0.000 .0177237 .0516213
ages2 | -.0504113 .0112049 -4.50 0.000 -.0723725 -.0284502
dep_ratio | -.2012548 .0757752 -2.66 0.008 -.3497714 -.0527382
_Ioccupati~1 | .1470003 .0750557 1.96 0.050 -.0001063 .2941068
_Ioccupati~2 | .3799047 .1064299 3.57 0.000 .171306 .5885035
_Ioccupati~3 | .0955231 .0798577 1.20 0.232 -.0609951 .2520413
_Ioccupati~4 | .1403077 .0831009 1.69 0.091 -.0225672 .3031825
_Ioccupati~6 | (omitted)
_cons | 6.630357 .2008758 33.01 0.000 6.236648 7.024066
-------------+----------------------------------------------------------------
select |
male | .5482426 .0394785 13.89 0.000 .4708661 .625619
_Iedunew2_1 | .4617968 .0549549 8.40 0.000 .3540871 .5695065
_Iedunew2_2 | .7589722 .0486543 15.60 0.000 .6636116 .8543329
_Iedunew2_3 | 1.236086 .0688923 17.94 0.000 1.101059 1.371112
_Iedunew2_4 | -.4383474 .2718042 -1.61 0.107 -.9710739 .0943792
_Ihhecosta~1 | .588973 .0505102 11.66 0.000 .489975 .6879711
_Ihhecosta~2 | .2846132 .0427787 6.65 0.000 .2007686 .3684578
_Ihhecosta~3 | .2546093 .0601532 4.23 0.000 .1367111 .3725075
inearthquake | .1958563 .0450808 4.34 0.000 .1074996 .2842131
urban2 | .439123 .0471768 9.31 0.000 .3466581 .5315879
_cons | -2.084462 .0427642 -48.74 0.000 -2.168278 -2.000645
-------------+----------------------------------------------------------------
/athrho | -1.118788 .085606 -13.07 0.000 -1.286573 -.9510036
/lnsigma | -.3448958 .0395288 -8.73 0.000 -.4223709 -.2674207
-------------+----------------------------------------------------------------
rho | -.807147 .0298349 -.8582264 -.740237
sigma | .7082942 .027998 .6554909 .765351
lambda | -.5716975 .0419986 -.6540133 -.4893818
------------------------------------------------------------------------------
Wald test of indep. eqns. (rho = 0): chi2(1) = 170.80 Prob > chi2 = 0.0000
------------------------------------------------------------------------------
. predict selpr200412, psel
(32284 missing values generated)
. gen Dpr200412=mills200412*(mills200412+selpr200412)
(159277 missing values generated)
. display Dpr200412
.
. display selpr200412
.30568352
. display mills200412
.
does any one know what's going on??
thanks a lot
--
Best
刘京 Jing LIU
*
* 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/