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]
st: Treatment effect with endogenous treatment for multinominal outcomes
From
Sawa Omori <[email protected]>
To
[email protected]
Subject
st: Treatment effect with endogenous treatment for multinominal outcomes
Date
Thu, 20 Mar 2014 15:43:13 +0900
Dear Stata list members,
I am using Stata13.1 SE and I would like to estimate treatment effect
with endogenous treatment for 3 multinominal outcomes (policy
reversal, no reform, policy reform). I am looking at the treatment
effect of international organizations on country's policy changes.
I hope cmp (Roodman, David. 2011. Stata Journal 11(2): 159-206) works
for these estimates.
However, when I run the command as follows, results do not converge as
shown the following.
Can I use cmp command to estimate treatment effect with endogenous
treatment for multinominal outcomes since Roodman(2011) do not give
examples on this. If could, is my command wrong and just the problems
of entered variables? Any suggestions are very much appreciated.
Thank you so much in advance.
Sawa
Sawa Omori, Ph.D.
Associate Professor
Department of Politics and International Studies
International Christian University, Tokyo, Japan
cmp ( lag_IO = lag_bankingcrisisdummy lag_ln_gdppc) (c3_credit =
lag_directedcredit lag_IO lag_gdpgrow
> th ), indicators ($cmp_probit $cmp_mprobit) cluster(countryNo)
Fitting individual models as starting point for full model fit.
Note: For programming reasons, these initial estimates may deviate
from your specification.
For exact fits of each equation alone, run cmp separately on each.
Iteration 0: log likelihood = -1762.3733
Iteration 1: log likelihood = -1492.4516
Iteration 2: log likelihood = -1489.2398
Iteration 3: log likelihood = -1489.2371
Iteration 4: log likelihood = -1489.2371
Probit regression Number of obs = 2921
LR chi2(2) = 546.27
Prob > chi2 = 0.0000
Log likelihood = -1489.2371 Pseudo R2 = 0.1550
----------------------------------------------------------------------------------------
lag_IO | Coef. Std. Err. z P>|z|
[95% Conf. Interval]
-----------------------+----------------------------------------------------------------
lag_bankingcrisisdummy | .7632921 .0878081 8.69 0.000
.5911913 .935393
lag_ln_gdppc | -.3850068 .0192198 -20.03 0.000
-.4226768 -.3473367
_cons | 2.266576 .1431045 15.84 0.000
1.986096 2.547055
----------------------------------------------------------------------------------------
Iteration 0: log likelihood = -215.68587
Iteration 1: log likelihood = -215.68587
Probit regression Number of obs = 2407
LR chi2(0) = -0.00
Prob > chi2 = .
Log likelihood = -215.68587 Pseudo R2 = -0.0000
------------------------------------------------------------------------------
_mp_cmp_y2 | Coef. Std. Err. z P>|z| [95% Conf. Interval]
-------------+----------------------------------------------------------------
_cons | -2.099997 .0613833 -34.21 0.000 -2.220306 -1.979688
------------------------------------------------------------------------------
Iteration 0: log likelihood = -691.41286
Iteration 1: log likelihood = -672.32085
Iteration 2: log likelihood = -672.10768
Iteration 3: log likelihood = -672.10763
Probit regression Number of obs = 2407
LR chi2(3) = 38.61
Prob > chi2 = 0.0000
Log likelihood = -672.10763 Pseudo R2 = 0.0279
------------------------------------------------------------------------------------
_mp_cmp_y3 | Coef. Std. Err. z P>|z| [95%
Conf. Interval]
-------------------+----------------------------------------------------------------
lag_directedcredit | .1833153 .0349896 5.24 0.000
.114737 .2518936
lag_IO | -.1169937 .0787977 -1.48 0.138
-.2714343 .0374469
lag_gdpgrowth | .0170111 .0070026 2.43 0.015
.0032863 .0307359
_cons | 1.121325 .0655769 17.10 0.000
.9927968 1.249854
------------------------------------------------------------------------------------
Iteration 0: log likelihood = -583.01649
Iteration 1: log likelihood = -551.18079
Iteration 2: log likelihood = -550.17504
Iteration 3: log likelihood = -550.17195
Iteration 4: log likelihood = -550.17195
Probit regression Number of obs = 2407
LR chi2(3) = 65.69
Prob > chi2 = 0.0000
Log likelihood = -550.17195 Pseudo R2 = 0.0563
------------------------------------------------------------------------------------
_mp_cmp_y4 | Coef. Std. Err. z P>|z| [95%
Conf. Interval]
-------------------+----------------------------------------------------------------
lag_directedcredit | -.2926993 .0408909 -7.16 0.000
-.3728439 -.2125546
lag_IO | .1021243 .0863737 1.18 0.237
-.067165 .2714135
lag_gdpgrowth | -.0161491 .0074706 -2.16 0.031
-.0307913 -.0015069
_cons | -1.132346 .0686062 -16.51 0.000
-1.266811 -.9978799
------------------------------------------------------------------------------------
Fitting full model.
Likelihoods for 2407 observations involve cumulative normal
distributions above dimension 2.
Using ghk2() to simulate them. Settings:
Sequence type = halton
Number of draws per observation = 99
Include antithetic draws = no
Scramble = no
Prime bases = 2 3 5
Each observation gets different draws, so changing the order of
observations in the data set would change
> the results.
Iteration 0: log pseudolikelihood = -2389.7011 (not concave)
Iteration 1: log pseudolikelihood = -2268.9075 (not concave)
Iteration 2: log pseudolikelihood = -2247.2773 (not concave)
Iteration 3: log pseudolikelihood = -2244.6163 (not concave)
Iteration 4: log pseudolikelihood = -2243.2808 (not concave)
Iteration 5: log pseudolikelihood = -2242.5749 (not concave)
cannot compute an improvement -- discontinuous region encountered
convergence not achieved
r(430);
*
* For searches and help try:
* http://www.stata.com/help.cgi?search
* http://www.stata.com/support/faqs/resources/statalist-faq/
* http://www.ats.ucla.edu/stat/stata/