 |
[Date Prev][Date Next][Thread Prev][Thread Next][Date index][Thread index]
RE: st: RE: Questions on ivprobit (probit model with an endogenous regressor)
|
From |
"Schaffer, Mark E" <[email protected]> |
|
To |
<[email protected]> |
|
Subject |
RE: st: RE: Questions on ivprobit (probit model with an endogenous regressor) |
|
Date |
Mon, 7 Jan 2008 11:03:21 -0000 |
Stephen,
The output below is messy because of all the line breaks, so I'll reply up here.
- You are right, there is a problem with the syntax, though it probably matters only for appearances. The varlist following the = should have only the EXCLUDED instruments, and should not include instruments that are also exogneous regressors. Thus you have
ivprobit vrics mhol acmt iacm bdze mdir dual ihol aud_1 aud_2 aud_3 aud_4 size perf levg ind_1 ind_2 ind_3 ind_4 ind_5 ind_6 ind_7 ind_8 grth efin optn (bdin = dual ihol mhol efin size levg ind_1 ind_2 ind_3 ind_4 ind_5 ind_6 ind_7 ind_8 optn vrics_1), first
But it should be
ivprobit vrics mhol acmt iacm bdze mdir dual ihol aud_1 aud_2 aud_3 aud_4 size perf levg ind_1 ind_2 ind_3 ind_4 ind_5 ind_6 ind_7 ind_8 grth efin optn (bdin = vrics_1), first
I am pretty sure this doesn't matter, though, in terms of results - Stata will (probably?) take care of the collinearities in the varlists in such a way that you will get the same results either way.
- This probably also explains your query about the number of "instrumented variables", though presumably you actually mean "instrumental variables" or just "instruments" ("instrumentED variables" are endogenous regressors).
- In your estimation results, you have a large number of statistically insignificant regressors. This could be a sign of multicollinearity (I'm not sure how you know you don't have this problem). In any case, it may explain why it takes so long for your model to converge.
- The statement at the end
note: 47 failures and 30 successes completely determined
is a common error message in Stata output for limited dependent variable models (logit, probit, etc.). Have a look at the FAQ for the logit case:
http://www.stata.com/support/faqs/stat/logitcd.html
Cheers,
Mark
> -----Original Message-----
> From: [email protected]
> [mailto:[email protected]] On Behalf Of
> [email protected]
> Sent: Monday, January 07, 2008 1:51 AM
> To: [email protected]
> Cc: Schaffer, Mark E
> Subject: Re: st: RE: Questions on ivprobit (probit model with
> an endogenous regressor)
>
> Hi Mark:
>
> Thanks for your prompt response to my questions. I am
> reposing three of the earlier questions, which you needed
> more information. I have now provided enough information
> below to enable you to help me.
>
> 1. Is there anything with my syntax below? If yes, how can it
> be corrected?
>
> ivprobit vrics mhol acmt iacm bdze mdir dual ihol aud_1 aud_2
> aud_3 aud_4 size perf levg ind_1 ind_2 ind_3 ind_4 ind_5
> ind_6 ind_7 ind_8 grth efin optn (bdin = dual ihol mhol fcash
> size levg ind_1 ind_2 ind_3 ind_4 ind_5 ind_6 ind_7 ind_8
> optn vrics_1), first
>
> 2. Stata increases the number of the instrumented variables
> in its results than I originally specified in the syntax?
> Why that? Is it because my model is under-identified?
>
> 3. For the fitting of the full model (i.e., probit model with
> the endogenous regressor), Stata goes through iteration from
> 1 to 1070, reporting that the intervening iterations (i.e., 1
> to 1068) are not concave. Why this long iteration? Does it
> suggest that the model is mis-specified? Or implies that the
> results are not correct? I DON'T HAVE MULTICOLLINEARITY
> PROBLEM IN THE DATA.
>
> The following are the equations I have been estimating with
> the ivprobit:
>
> BDIN, = F(&dgr;0 + &dgr;1MHOL + &dgr;2BDZE + &dgr;3DUAL + &dgr;4IHOL + &dgr;5SIZE + &dgr;6LEVG
> + &dgr;7GRTH + + &dgr;17EFIN + &dgr;18OPTN + &dgr;19VRICSt_1)
> (1)
>
> Prob (VRICS = 1) = F(&bgr;0 + &bgr;1MHOL + &bgr;2ACMT + &bgr;3IACM + &bgr;4BDZE
> + &bgr;5FT_BDIN + &bgr;6MDIR + &bgr;7DUAL + &bgr;b8IHOL + + &bgr;14SIZE
> + &bgr;15PERF + &bgr;16LEVG + + &bgr;26GRTH + &bgr;27EFIN + &bgr;28OPTN)
> (2)
>
> Equation 2 is the model of interest. FT_BDIN in equation 2
> is the fitted values of BDIN in equation 1. My sample size
> is 198 companies (110 experimental and 88 control sub-sample).
>
> The syntax and results are as follows:
>
> ivprobit vrics mhol acmt iacm bdze mdir dual ihol aud_1 aud_2
> aud_3 aud_4 size perf levg ind_1 ind_2 ind_3 ind_4 ind_5
> ind_6 ind_7 ind_8 grth efin optn (bdin = dual ihol mhol efin
> size levg ind_1 ind_2 ind_3 ind_4 ind_5 ind_6 ind_7 ind_8
> optn vrics_1), first Fitting exogenous probit model
>
> Iteration 0: log likelihood = -136.01839
> .
> Iteration 6: log likelihood = -103.21347
>
> Fitting full model
>
> Iteration 0: log likelihood = -507.53953 (not concave)
> Iteration 1: log likelihood = -507.44146 (not concave)
> .
> .
> Iteration 1061:log likelihood = -467.50251 (backed up)
> Iteration 1062:log likelihood = -465.17425 Iteration 1063:log
> likelihood = -462.61643 .
> .
> Iteration 1070:log likelihood = -462.0811
>
> Probit model with endogenous regressors Number of
> obs = 198
>
> Wald chi2(26) = 392.55
> Log likelihood = -462.0811
> Prob > chi2 = 0.0000
> --------------------------------------------------------------
> ----------------
> | Coef. Std. Err. z P>|z|
> [95% Conf. Interval]
> -------------+------------------------------------------------
> ----------
> -------------+------
> vrics |
> bdin | .535787 .0270656 19.80 0.000
> .4827393 .5888347
> mhol | .0082955 .6629267 0.01 0.990
> -1.291017 1.307608
> acmt | .0173609 .0293927 0.59 0.555
> -.0402477 .0749695
> iacm | -.2845267 .2214701 -1.28 0.199
> -.7186002 .1495468
> bdze | .1125746 .0368413 3.06 0.002
> .040367 .1847823
> mdir | -.1984201 .0348804 -5.69 0.000
> -.2667844 -.1300558
> dual | .3046553 .1763825 1.73 0.084
> -.041048 .6503587
> ihol | .5343345 .3381111 1.58 0.114
> -.1283511 1.19702
> aud_1 | -.118587 .2618686 -0.45 0.651
> -.63184 .3946661
> aud_2 | -.1315854 .1602961 -0.82 0.412
> -.4457599 .1825892
> aud_3 | -.2391904 .3023197 -0.79 0.429
> -.8317261 .3533454
> aud_4 | -.1764671 .2734199 -0.65 0.519
> -.7123601 .359426
> size | -.394326 .085888 -4.59 0.000
> -.5626634 -.2259885
> perf | .1184288 .5215651 0.23 0.820
> -.90382 1.140678
> levg | -.6340103 .4221428 -1.50 0.133
> -1.461395 .1933744
> ind_1 | .638568 .4017463 1.59 0.112
> -.1488403 1.425976
> ind_2 | .5798336 .3313281 1.75 0.080
> -.0695576 1.229225
> ind_3 | .7170456 .3188755 2.25 0.025
> .0920611 1.34203
> ind_4 | .3894297 .3882769 1.00 0.316
> -.371579 1.150438
> ind_5 | .7477128 .3261907 2.29 0.022
> .1083907 1.387035
> ind_6 | .8033681 .4126983 1.95 0.052
> -.0055056 1.612242
> ind_7 | .5320546 .4018902 1.32 0.186
> -.2556357 1.319745
> ind_8 | 1.585647 .5801551 2.73 0.006
> .4485636 2.72273
> grth | .0025123 .0013356 1.88 0.060
> -.0001053 .0051299
> efin | .0080272 .0116433 0.69 0.491
> -.0147932 .0308476
> optn | .0069244 .011954 0.58 0.562
> -.0165049 .0303537
> _cons | 7.2861 1.909368 3.82 0.000
> 3.543807 11.02839
> -------------+------------------------------------------------
> ----------
> -------------+------
> bdin |
> mhol | .0378184 1.235286 0.03 0.976
> -2.383298 2.458935
> acmt | -.033273 .0544709 -0.61 0.541
> -.1400339 .073488
> iacm | .596355 .367546 1.62 0.105
> -.1240219 1.316732
> bdze | -.2195357 .0680645 -3.23 0.001
> -.3529396 -.0861318
> mdir | .3700483 .0622494 5.94 0.000
> .2480418 .4920548
> dual | -.573009 .326602 -1.75 0.079
> -1.213137 .0671192
> ihol | -1.012138 .6303383 -1.61 0.108
> -2.247578 .2233028
> aud_1 | .14425 .4679218 0.31 0.758
> -.7728599 1.06136
> aud_2 | .1355325 .236811 0.57 0.567
> -.3286086 .5996736
> aud_3 | .300847 .4963293 0.61 0.544
> -.6719404 1.273634
> aud_4 | .2759295 .4846338 0.57 0.569
> -.6739352 1.225794
> size | .7537389 .1458796 5.17 0.000
> .4678201 1.039658
> perf | -.181126 .9590768 -0.19 0.850
> -2.060882 1.69863
> levg | 1.132908 .7862537 1.44 0.150
> -.4081206 2.673937
> ind_1 | -1.167724 .744777 -1.57 0.117
> -2.627461 .2920118
> ind_2 | -1.046297 .6184434 -1.69 0.091
> -2.258424 .1658298
> ind_3 | -1.318878 .590602 -2.23 0.026
> -2.476436 -.1613189
> ind_4 | -.7132557 .721654 -0.99 0.323
> -2.127672 .7011601
> ind_5 | -1.400158 .6024688 -2.32 0.020
> -2.580976 -.2193413
> ind_6 | -1.427494 .7589549 -1.88 0.060
> -2.915018 .0600307
> ind_7 | -1.033129 .7444047 -1.39 0.165
> -2.492135 .4258778
> ind_8 | -2.996058 1.066981 -2.81 0.005
> -5.087302 -.9048135
> grth | -.0046962 .0024757 -1.90 0.058
> -.0095485 .000156
> efin | -.0175522 .0208806 -0.84 0.401
> -.0584774 .0233729
> optn | -.0136099 .0222878 -0.61 0.541
> -.0572931 .0300734
> vrics_1 | .198787 .3208153 0.62 0.536
> -.4299995 .8275734
> _cons | -13.92576 3.327674 -4.18 0.000
> -20.44788 -7.403639
> -------------+------------------------------------------------
> ----------
> -------------+------
> /athrho | -3.964625 1.619449 -2.45 0.014
> -7.138687 -.7905633
> /lnsigma | .6231125 .050253 12.40 0.000
> .5246184 .7216065
> -------------+------------------------------------------------
> ----------
> -------------+------
> rho | -.9992801 .0023307
> -.9999987 -.658728
> sigma | 1.864723 .0937079
> 1.689814 2.057736
> --------------------------------------------------------------
> ----------------
> Instrumented: bdin
> Instruments: mhol acmt iacm bdze mdir dual ihol aud_1 aud_2
> aud_3 aud_4 size
> perf levg ind_1 ind_2 ind_3 ind_4 ind_5 ind_6
> ind_7 ind_8 grth
> efin optn vrics_1*
> --------------------------------------------------------------
> ----------------
> Wald test of exogeneity (/athrho = 0): chi2(1) = 5.99
> Prob > chi2 = 0.0144
>
> *Some of the instrumented variables were not specified in the
> original syntax.
>
> Below is the syntax and results when I used two-step
> estimator. What does the note at the bottom of the results mean?
>
> ivprobit vrics mhol acmt iacm bdze mdir dual ihol aud_1 aud_2
> aud_3 aud_4 size perf levg ind_1 ind_2 ind_3 ind_4 ind_5
> ind_6 ind_7 ind_8 grth efin optn (bdin = dual ihol mhol efin
> size levg ind_1 ind_2 ind_3 ind_4 ind_5 ind_6 ind_7 ind _8
> optn vrics_1), twostep first
>
> Checking reduced-form model
> First stage regression
>
> Source | SS df MS
> Number of obs = 198
> -------------+------------------------------
> F( 26, 171) = 5.18
> Model | 542.099097 26 20.8499653
> Prob > F = 0.0000
> Residual | 688.483679 171 4.02622035
> R-squared = 0.4405
> -------------+------------------------------
> Adj R-squared = 0.3555
> Total | 1230.58278 197 6.24661308
> Root MSE = 2.0065
> --------------------------------------------------------------
> ----------------
> bdin | Coef. Std. Err. t P>|t|
> [95% Conf. Interval]
> -------------+------------------------------------------------
> ----------
> -------------+------
> vrics_1 | .1988048 .3453415 0.58 0.566
> -.4828765 .8804861
> mhol | .0378225 1.329227 0.03 0.977
> -2.585984 2.661629
> acmt | -.0332726 .0586135 -0.57 0.571
> -.1489718 .0824266
> iacm | .5963493 .3955089 1.51 0.133
> -.1843592 1.377058
> bdze | -.2195362 .0732413 -3.00 0.003
> -.3641098 -.0749626
> mdir | .3700488 .0669836 5.52 0.000
> .2378275 .50227
> dual | -.5730081 .3514396 -1.63 0.105
> -1.266727 .1207105
> ihol | -1.012142 .6782774 -1.49 0.137
> -2.351017 .3267326
> aud_1 | .14425 .5035061 0.29 0.775
> -.8496378 1.138138
> aud_2 | .1355322 .2548215 0.53 0.596
> -.3674686 .638533
> aud_3 | .3008486 .5340752 0.56 0.574
> -.7533807 1.355078
> aud_4 | .2759339 .5214952 0.53 0.597
> -.7534632 1.305331
> size | .7537372 .1569757 4.80 0.000
> .4438776 1.063597
> perf | -.1811319 1.032013 -0.18 0.861
> -2.218258 1.855994
> levg | 1.132905 .8460482 1.34 0.182
> -.5371384 2.802948
> ind_1 | -1.167724 .801416 -1.46 0.147
> -2.749666 .4142183
> ind_2 | -1.046293 .6654784 -1.57 0.118
> -2.359903 .2673175
> ind_3 | -1.318876 .6355171 -2.08 0.039
> -2.573345 -.0644069
> ind_4 | -.7132544 .7765309 -0.92 0.360
> -2.246075 .8195664
> ind_5 | -1.400157 .6482868 -2.16 0.032
> -2.679832 -.1204813
> ind_6 | -1.427492 .8166718 -1.75 0.082
> -3.039548 .1845637
> ind_7 | -1.033129 .8010162 -1.29 0.199
> -2.614282 .5480241
> ind_8 | -2.996057 1.148121 -2.61 0.010
> -5.262373 -.7297423
> grth | -.0046962 .002664 -1.76 0.080
> -.0099548 .0005623
> efin | -.0175522 .0224685 -0.78 0.436
> -.0619036 .0267992
> optn | -.0136099 .0239828 -0.57 0.571
> -.0609505 .0337306
> _cons | -13.92572 3.580775 -3.89 0.000
> -20.99394 -6.857512
> --------------------------------------------------------------
> ----------------
>
> Two-step probit with endogenous regressors Number of
> obs = 198
>
> Wald chi2(26) = 0.99
>
> Prob > chi2 = 1.0000
> --------------------------------------------------------------
> ----------------
> | Coef. Std. Err. z P>|z|
> [95% Conf. Interval]
> -------------+------------------------------------------------
> ----------
> -------------+------
> bdin | 14.12199 24.61414 0.57 0.566
> -34.12084 62.36482
> mhol | .2186578 18.79424 0.01 0.991
> -36.61738 37.05469
> acmt | .4575839 1.235632 0.37 0.711
> -1.96421 2.879378
> iacm | -7.49924 17.15531 -0.44 0.662
> -41.12302 26.12454
> bdze | 2.967177 5.363456 0.55 0.580
> -7.545004 13.47936
> mdir | -5.22986 9.026449 -0.58 0.562
> -22.92138 12.46166
> dual | 8.029926 15.17887 0.53 0.597
> -21.72012 37.77997
> ihol | 14.08375 25.45977 0.55 0.580
> -35.81648 63.98398
> aud_1 | -3.125751 7.994754 -0.39 0.696
> -18.79518 12.54368
> aud_2 | -3.468392 5.087113 -0.68 0.495
> -13.43895 6.502165
> aud_3 | -6.304659 10.32934 -0.61 0.542
> -26.54979 13.94047
> aud_4 | -4.651355 9.189619 -0.51 0.613
> -22.66268 13.35997
> size | -10.39339 19.15959 -0.54 0.587
> -47.9455 27.15873
> perf | 3.121618 14.83004 0.21 0.833
> -25.94473 32.18797
> levg | -16.71092 31.29799 -0.53 0.593
> -78.05385 44.63201
> ind_1 | 16.83106 30.99911 0.54 0.587
> -43.92608 77.58821
> ind_2 | 15.28294 28.46859 0.54 0.591
> -40.51448 71.08035
> ind_3 | 18.89951 34.12934 0.55 0.580
> -47.99277 85.79179
> ind_4 | 10.26438 21.0224 0.49 0.625
> -30.93877 51.46753
> ind_5 | 19.70779 36.16063 0.55 0.586
> -51.16575 90.58132
> ind_6 | 21.17483 37.29913 0.57 0.570
> -51.93013 94.27978
> ind_7 | 14.02357 27.7139 0.51 0.613
> -40.29468 68.34181
> ind_8 | 41.79358 75.57157 0.55 0.580
> -106.324 189.9111
> grth | .0662178 .1233329 0.54 0.591
> -.1755102 .3079459
> efin | .2115723 .5480824 0.39 0.699
> -.8626495 1.285794
> optn | .1825097 .4637365 0.39 0.694
> -.7263971 1.091416
> _cons | 192.0422 356.4177 0.54 0.590
> -506.5236 890.6081
> --------------------------------------------------------------
> ----------------
> Instrumented: bdin
> Instruments: mhol acmt iacm bdze mdir dual ihol aud_1 aud_2
> aud_3 aud_4
> size perf levg ind_1 ind_2 ind_3 ind_4 ind_5
> ind_6 ind_7
> ind_8 grth efin optn vrics_1
> --------------------------------------------------------------
> ----------------
> Wald test of exogeneity: chi2(1) = 51.79
> Prob > chi2 = 0.0000
>
> note: 47 failures and 30 successes completely determined.
>
> Regards,
>
> Stephen
>
> ---- "Schaffer wrote:
> > Stephen,
> >
> > > -----Original Message-----
> > > From: [email protected]
> > > [mailto:[email protected]] On Behalf Of
> > > [email protected]
> > > Sent: 06 January 2008 21:45
> > > To: [email protected]
> > > Subject: st: Questions on ivprobit (probit model with an
> endogenous
> > > regressor)
> > >
> > > Dear colleagues:
> > >
> > > I need your help with respect to the following questions about
> > > ivprobit (command for probit model with an endogenous
> > > regressor):
> > >
> > > 1. Is there anything with my syntax below? If yes, how can it be
> > > corrected?
> > >
> > > ivprobit vrics mhol acmt iacm bdze mdir dual ihol aud_1 aud_2
> > > aud_3 aud_4 size perf levg ind_1 ind_2 ind_3 ind_4 ind_5
> > > ind_6 ind_7 ind_8 rev_gwth fcash optns (bdin= dual ihol
> mhol fcash
> > > size levg ind_1 ind_2 ind_3 ind_4 ind_5 ind_6
> > > ind_7 ind_8 optns vrics_1), first
> >
> > It's impossible to tell without seeing the actual call to
> -ivprobit-
> > and what Stata makes of it. You should post this.
> >
> > > 2. The dependent variable, vrics, is a dummy coded 1/0.
> > > Hence, my use of ivprobit. However, the endogenous
> regressor, bdin,
> > > is not. Isn't ivprobit reading the data on bdin as dummy?
> >
> > No. It's because you're using the default ML estimator.
> If you use
> > the two-step estimator, you'll see that the first-step
> estimates for
> > bdin are exactly the same estimates you get if you use -regress-.
> >
> > > My question is based on the fact that OLS estimates of the bdin
> > > equation is different from those returned by ivprobit for the
> > > first-stage regression. Also, Stata 10 reports iteration for
> > > "Fitting exogenous probit model".
> > >
> > > 3. Stata increases the number of the instrumented
> variables in its
> > > results than I originally specified in the syntax? Why
> that? Is it
> > > because my model is under-identified?
> >
> > Again, it's impossible to tell unless you show us the call to
> > -ivprobit- and the results.
> >
> > > 4. For the fitting of the full model (i.e., probit model with the
> > > endogenous regressor), Stata goes through iteration from
> > > 1 to 1070, reporting that the intervening iterations (i.e., 1 to
> > > 1068) are not concave. Why this long iteration? Does it
> suggest that
> > > the model is mis-specified? Or implies that the results are not
> > > correct?
> >
> > You're probably asking a lot of the data, maybe too much.
> Perhaps you
> > have some multicollinearity problems. Are many of the coefficients
> > insignificant?
> >
> > > 5. Stata 10 does not report model summary statistics for the
> > > first-stage regression with bdin as the dependent variable.
> > > Is there any way of getting these statistics?
> >
> > With the ML estimator, the "first-stage regression" isn't really a
> > first stage, since it's estimated simultaneously with the main
> > equation. I think this means that you just have to get the
> stats you
> > want from the main -ivprobit- results with the ML estimator. You
> > could switch to the two-step estimator so that the
> first-stage results
> > are reproducible with a simple call to -regress-, but this doesn't
> > seem like a good reason to do this.
> >
> > Hope this helps.
> >
> > Cheers,
> > Mark
> >
> >
> > Prof. Mark Schaffer
> > Director, CERT
> > Department of Economics
> > School of Management & Languages
> > Heriot-Watt University, Edinburgh EH14 4AS tel
> +44-131-451-3494 / fax
> > +44-131-451-3296
> > email: [email protected]
> > web: http://www.sml.hw.ac.uk/ecomes
> >
> > > I look forward to hearing from you. Thanks for your cooperation.
> > >
> > > Regards,
> > >
> > > Stephen
> > > --
> > > Stephen Owusu-Ansah, PhD, CIA, CBM
> > >
> > > *
> > > * For searches and help try:
> > > * http://www.stata.com/support/faqs/res/findit.html
> > > * http://www.stata.com/support/statalist/faq
> > > * http://www.ats.ucla.edu/stat/stata/
> > >
> >
> > *
> > * For searches and help try:
> > * http://www.stata.com/support/faqs/res/findit.html
> > * http://www.stata.com/support/statalist/faq
> > * http://www.ats.ucla.edu/stat/stata/
>
>
> *
> * For searches and help try:
> * http://www.stata.com/support/faqs/res/findit.html
> * http://www.stata.com/support/statalist/faq
> * http://www.ats.ucla.edu/stat/stata/
>
*
* For searches and help try:
* http://www.stata.com/support/faqs/res/findit.html
* http://www.stata.com/support/statalist/faq
* http://www.ats.ucla.edu/stat/stata/