[Andrew Musau]

Another option is David Roodman's cmp command from SSC that can estimate multiple equation models. If you send him an email, he may be able to suggest if and how to exclude the intercepts in each equation. If you are successful in that, you have solved your problem.

webuse mroz87, clear
set seed 03302022
gen outcome2= runiformint(1, 4920)
tobit whrs75 c.nwinc c.wedyrs c.wexper c.wexper#c.wexper c.wifeage c.kl6 c.k618, ll(0) nolog
tobit outcome2 c.nwinc c.wedyrs c.wexper c.wexper#c.wexper c.wifeage c.kl6 c.k618, ll(0) nolog
cmp (whrs75= c.nwinc c.wedyrs c.wexper c.wexper#c.wexper c.wifeage c.kl6 c.k618) ///
(outcome2= c.nwinc c.wedyrs c.wexper c.wexper#c.wexper c.wifeage c.kl6 c.k618), ///
ind("cond(whrs75>0, $cmp_cont, $cmp_left)" "cond(outcome2>0, $cmp_cont, $cmp_left)") qui

Res.:

. tobit whrs75 c.nwinc c.wedyrs c.wexper c.wexper#c.wexper c.wifeage c.kl6 c.k618, ll(0) nolog

Tobit regression                                Number of obs     =        753
                                                   Uncensored     =        428
Limits: lower = 0                                  Left-censored  =        325
        upper = +inf                               Right-censored =          0

                                                LR chi2(7)        =     271.59
                                                Prob > chi2       =     0.0000
Log likelihood = -3819.0946                     Pseudo R2         =     0.0343

-----------------------------------------------------------------------------------
           whrs75 |      Coef.   Std. Err.      t    P>|t|     [95% Conf. Interval]
------------------+----------------------------------------------------------------
            nwinc |  -8.814227   4.459089    -1.98   0.048    -17.56808   -.0603708
           wedyrs |   80.64541   21.58318     3.74   0.000     38.27441    123.0164
           wexper |    131.564   17.27935     7.61   0.000     97.64211     165.486
                  |
c.wexper#c.wexper |  -1.864153   .5376606    -3.47   0.001    -2.919661   -.8086455
                  |
          wifeage |  -54.40491   7.418483    -7.33   0.000     -68.9685   -39.84133
              kl6 |  -894.0202   111.8777    -7.99   0.000    -1113.653   -674.3875
             k618 |  -16.21805    38.6413    -0.42   0.675    -92.07668    59.64057
            _cons |   965.3068   446.4351     2.16   0.031     88.88827    1841.725
------------------+----------------------------------------------------------------
     var(e.whrs75)|    1258927   93304.48                       1088458     1456093
-----------------------------------------------------------------------------------

. tobit outcome2 c.nwinc c.wedyrs c.wexper c.wexper#c.wexper c.wifeage c.kl6 c.k618, ll(0) nolog

Tobit regression                                Number of obs     =        753
                                                   Uncensored     =        753
Limits: lower = 0                                  Left-censored  =          0
        upper = +inf                               Right-censored =          0

                                                LR chi2(7)        =       5.70
                                                Prob > chi2       =     0.5753
Log likelihood = -6550.7516                     Pseudo R2         =     0.0004

-----------------------------------------------------------------------------------
         outcome2 |      Coef.   Std. Err.      t    P>|t|     [95% Conf. Interval]
------------------+----------------------------------------------------------------
            nwinc |    -2.8083   4.923527    -0.57   0.569    -12.47392    6.857317
           wedyrs |   23.37747   25.07164     0.93   0.351    -25.84191    72.59684
           wexper |   25.51334   19.28185     1.32   0.186    -12.33979    63.36648
                  |
c.wexper#c.wexper |  -.8997776   .6281175    -1.43   0.152    -2.132866    .3333107
                  |
          wifeage |     7.6236   8.445607     0.90   0.367    -8.956385    24.20359
              kl6 |   159.3098   113.8887     1.40   0.162    -64.27071    382.8903
             k618 |   47.07904   44.85406     1.05   0.294    -40.97618    135.1343
            _cons |   1768.561   524.0627     3.37   0.001     739.7478    2797.374
------------------+----------------------------------------------------------------
   var(e.outcome2)|    2107889   108633.9                       1905058     2332315
-----------------------------------------------------------------------------------

. cmp (whrs75= c.nwinc c.wedyrs c.wexper c.wexper#c.wexper c.wifeage c.kl6 c.k618) ///
> (outcome2= c.nwinc c.wedyrs c.wexper c.wexper#c.wexper c.wifeage c.kl6 c.k618), ///
> ind("cond(whrs75>0, $cmp_cont, $cmp_left)" "cond(outcome2>0, $cmp_cont, $cmp_left)") qui

Fitting individual models as starting point for full model fit.

Fitting constant-only model for LR test of overall model fit.

Fitting full model.

Mixed-process regression                        Number of obs     =        753
                                                LR chi2(14)       =     277.15
Log likelihood = -10369.821                     Prob > chi2       =     0.0000

-----------------------------------------------------------------------------------
                  |      Coef.   Std. Err.      z    P>|z|     [95% Conf. Interval]
------------------+----------------------------------------------------------------
whrs75            |
            nwinc |  -8.810777   4.458305    -1.98   0.048     -17.5489   -.0726592
           wedyrs |   80.56007   21.58372     3.73   0.000     38.25676    122.8634
           wexper |   131.6076    17.2807     7.62   0.000     97.73804    165.4772
                  |
c.wexper#c.wexper |  -1.865532   .5377009    -3.47   0.001    -2.919406   -.8116571
                  |
          wifeage |   -54.3756   7.418726    -7.33   0.000    -68.91604   -39.83517
              kl6 |  -893.2394   111.8979    -7.98   0.000    -1112.555   -673.9236
             k618 |  -16.17063    38.6418    -0.42   0.676    -91.90716    59.56589
            _cons |    964.672   446.4197     2.16   0.031     89.70553    1839.639
------------------+----------------------------------------------------------------
outcome2          |
            nwinc |  -2.808302   4.923525    -0.57   0.568    -12.45823    6.841629
           wedyrs |   23.37749   25.07163     0.93   0.351      -25.762    72.51699
           wexper |   25.51333   19.28184     1.32   0.186    -12.27837    63.30504
                  |
c.wexper#c.wexper |  -.8997773   .6281172    -1.43   0.152    -2.130864    .3313099
                  |
          wifeage |   7.623598   8.445603     0.90   0.367     -8.92948    24.17668
              kl6 |   159.3097   113.8886     1.40   0.162    -63.90796    382.5273
             k618 |   47.07903   44.85404     1.05   0.294    -40.83328    134.9913
            _cons |   1768.561   524.0625     3.37   0.001     741.4172    2795.704
------------------+----------------------------------------------------------------
         /lnsig_1 |   7.022846   .0370561   189.52   0.000     6.950217    7.095475
         /lnsig_2 |   7.280598   .0257684   282.54   0.000     7.230093    7.331103
     /atanhrho_12 |  -.0090192   .0401423    -0.22   0.822    -.0876966    .0696582
------------------+----------------------------------------------------------------
            sig_1 |   1121.975   41.57604                      1043.376    1206.495
            sig_2 |   1451.857     37.412                      1380.351    1527.066
           rho_12 |  -.0090189    .040139                     -.0874724    .0695458
-----------------------------------------------------------------------------------

.

[Andrew Musau]

Actually, answering the question in #16, it's just the standard -nocons-. Whenever this pays off (if you are a student, when you graduate, get a job and receive your first paycheck), send your contribution to David!

webuse mroz87, clear
set seed 03302022
gen outcome2= runiformint(1, 4920)
tobit whrs75 c.nwinc c.wedyrs c.wexper c.wexper#c.wexper c.wifeage c.kl6 c.k618, ll(0) nocons nolog 
tobit outcome2 c.nwinc c.wedyrs c.wexper c.wexper#c.wexper c.wifeage c.kl6 c.k618, ll(0) nocons nolog
cmp (whrs75= c.nwinc c.wedyrs c.wexper c.wexper#c.wexper c.wifeage c.kl6 c.k618, nocons) ///
(outcome2= c.nwinc c.wedyrs c.wexper c.wexper#c.wexper c.wifeage c.kl6 c.k618, nocons), ///
ind("cond(whrs75>0, $cmp_cont, $cmp_left)" "cond(outcome2>0, $cmp_cont, $cmp_left)") qui

Res.:

. tobit whrs75 c.nwinc c.wedyrs c.wexper c.wexper#c.wexper c.wifeage c.kl6 c.k618, ll(0) nocons nolog 

Tobit regression                                Number of obs     =        753
                                                   Uncensored     =        428
Limits: lower = 0                                  Left-censored  =        325
        upper = +inf                               Right-censored =          0

                                                F(   7,    746)   =      52.61
                                                Prob > F          =     0.0000
Log likelihood = -3821.3981                     Pseudo R2         =     0.0368

 ( 1)  [whrs75]_cons = 0
-----------------------------------------------------------------------------------
           whrs75 |      Coef.   Std. Err.      t    P>|t|     [95% Conf. Interval]
------------------+----------------------------------------------------------------
            nwinc |  -9.847428    4.48635    -2.19   0.028     -18.6548   -1.040053
           wedyrs |   109.4473   17.29735     6.33   0.000        75.49    143.4046
           wexper |   140.9599   17.00301     8.29   0.000     107.5805    174.3394
                  |
c.wexper#c.wexper |  -2.188859   .5240843    -4.18   0.000    -3.217714   -1.160003
                  |
          wifeage |   -42.4201   4.861033    -8.73   0.000    -51.96304   -32.87717
              kl6 |  -831.0646   108.6034    -7.65   0.000    -1044.269   -617.8599
             k618 |    20.3103   35.07418     0.58   0.563    -48.54554    89.16615
            _cons |          0  (omitted)
------------------+----------------------------------------------------------------
     var(e.whrs75)|    1284148   94835.66                       1110838     1484497
-----------------------------------------------------------------------------------

. tobit outcome2 c.nwinc c.wedyrs c.wexper c.wexper#c.wexper c.wifeage c.kl6 c.k618, ll(0) nocons nolog

Tobit regression                                Number of obs     =        753
                                                   Uncensored     =        753
Limits: lower = 0                                  Left-censored  =          0
        upper = +inf                               Right-censored =          0

                                                F(   7,    746)   =     322.56
                                                Prob > F          =     0.0000
Log likelihood = -6556.4033                     Pseudo R2         =     0.0737

 ( 1)  [outcome2]_cons = 0
-----------------------------------------------------------------------------------
         outcome2 |      Coef.   Std. Err.      t    P>|t|     [95% Conf. Interval]
------------------+----------------------------------------------------------------
            nwinc |   -4.47991   4.935452    -0.91   0.364    -14.16894    5.209118
           wedyrs |   75.32679   19.93851     3.78   0.000     36.18452    114.4691
           wexper |   39.26225   18.98851     2.07   0.039     1.984987    76.53952
                  |
c.wexper#c.wexper |  -1.418335   .6136206    -2.31   0.021    -2.622964   -.2137068
                  |
          wifeage |     30.512   5.070643     6.02   0.000     20.55757    40.46642
              kl6 |   286.6867   108.2618     2.65   0.008     74.15271    499.2207
             k618 |   112.2164   40.79374     2.75   0.006     32.13222    192.3006
            _cons |          0  (omitted)
------------------+----------------------------------------------------------------
   var(e.outcome2)|    2139770   110276.9                       1933871     2367590
-----------------------------------------------------------------------------------

. cmp (whrs75= c.nwinc c.wedyrs c.wexper c.wexper#c.wexper c.wifeage c.kl6 c.k618, nocons) ///
> (outcome2= c.nwinc c.wedyrs c.wexper c.wexper#c.wexper c.wifeage c.kl6 c.k618, nocons), ///
> ind("cond(whrs75>0, $cmp_cont, $cmp_left)" "cond(outcome2>0, $cmp_cont, $cmp_left)") qui

Fitting individual models as starting point for full model fit.

Fitting full model.

Mixed-process regression                        Number of obs     =        753
                                                Wald chi2(14)     =    2625.27
Log likelihood = -10377.801                     Prob > chi2       =     0.0000

-----------------------------------------------------------------------------------
                  |      Coef.   Std. Err.      z    P>|z|     [95% Conf. Interval]
------------------+----------------------------------------------------------------
whrs75            |
            nwinc |  -9.847732   4.486411    -2.20   0.028    -18.64094   -1.054528
           wedyrs |   109.4588   17.30324     6.33   0.000     75.54513    143.3726
           wexper |   140.9529   17.00509     8.29   0.000     107.6236    174.2823
                  |
c.wexper#c.wexper |  -2.188663   .5241348    -4.18   0.000    -3.215948   -1.161378
                  |
          wifeage |  -42.42194   4.861559    -8.73   0.000    -51.95042   -32.89345
              kl6 |  -831.1396   108.6446    -7.65   0.000    -1044.079   -618.2002
             k618 |   20.30592    35.0744     0.58   0.563    -48.43865    89.05049
------------------+----------------------------------------------------------------
outcome2          |
            nwinc |  -4.479911   4.935449    -0.91   0.364    -14.15321    5.193392
           wedyrs |   75.32682    19.9385     3.78   0.000     36.24807    114.4056
           wexper |   39.26223    18.9885     2.07   0.039     2.045461      76.479
                  |
c.wexper#c.wexper |  -1.418335   .6136203    -2.31   0.021    -2.621008   -.2156611
                  |
          wifeage |   30.51199   5.070641     6.02   0.000     20.57372    40.45026
              kl6 |   286.6866   108.2617     2.65   0.008     74.49749    498.8756
             k618 |   112.2164   40.79372     2.75   0.006     32.26217    192.1706
------------------+----------------------------------------------------------------
         /lnsig_1 |     7.0328   .0369258   190.46   0.000     6.960427    7.105174
         /lnsig_2 |   7.288104   .0257684   282.83   0.000     7.237599    7.338609
     /atanhrho_12 |   .0010275   .0402228     0.03   0.980    -.0778077    .0798627
------------------+----------------------------------------------------------------
            sig_1 |   1133.199   41.84431                      1054.084    1218.254
            sig_2 |   1462.794   37.69385                       1390.75    1538.571
           rho_12 |   .0010275   .0402227                      -.077651    .0796933
-----------------------------------------------------------------------------------

.

[Katy Tabero]

Hi Andrew,

I think I might owe both yourself and David - hopefully, it pays well!

This solution works as you have shown, both results appear in one table and are very close to those calculated individually. When I use margins to get the APE's (margins, predict(ystar(0,.)) dydx(*)), I only get results for the first regression - is there some syntax change to get it for both sets of results?

[Andrew Musau]

This solution works as you have shown, both results appear in one table and are very close to those calculated individually.

The results are identical. Just note that tobit outputs t-statistics and cmp outputs z-statistics.

When I use margins to get the APE's (margins, predict(ystar(0,.)) dydx(*)), I only get results for the first regression - is there some syntax change to get it for both sets of results?

webuse mroz87, clear
set seed 03302022
gen outcome2= runiformint(1, 4920)
tobit whrs75 c.nwinc c.wedyrs c.wexper c.wexper#c.wexper c.wifeage c.kl6 c.k618, ll(0) nocons nolog 
tobit outcome2 c.nwinc c.wedyrs c.wexper c.wexper#c.wexper c.wifeage c.kl6 c.k618, ll(0) nocons nolog
cmp (whrs75= c.nwinc c.wedyrs c.wexper c.wexper#c.wexper c.wifeage c.kl6 c.k618, nocons) ///
(outcome2= c.nwinc c.wedyrs c.wexper c.wexper#c.wexper c.wifeage c.kl6 c.k618, nocons), ///
ind("cond(whrs75>0, $cmp_cont, $cmp_left)" "cond(outcome2>0, $cmp_cont, $cmp_left)") qui
margins, predict(ys(0 .) eq(#1)) predict(ys(0 .) eq(#2)) dydx(*)

Res.:

. margins, predict(ys(0 .) eq(#1)) predict(ys(0 .) eq(#2)) dydx(*)

Average marginal effects                        Number of obs     =        753
Model VCE    : OIM

dy/dx w.r.t. : nwinc wedyrs wexper wifeage kl6 k618
1._predict   : E(whrs75*|0<whrs75), predict(ys(0 .) eq(#1))
2._predict   : E(outcome2*|0<outcome2), predict(ys(0 .) eq(#2))

------------------------------------------------------------------------------
             |            Delta-method
             |      dy/dx   Std. Err.      z    P>|z|     [95% Conf. Interval]
-------------+----------------------------------------------------------------
nwinc        |
    _predict |
          1  |  -5.750593   2.618037    -2.20   0.028    -10.88185    -.619335
          2  |  -4.276987   4.711722    -0.91   0.364    -13.51179    4.957819
-------------+----------------------------------------------------------------
wedyrs       |
    _predict |
          1  |   63.91861   10.11868     6.32   0.000     44.08635    83.75086
          2  |   71.91478    19.0562     3.77   0.000     34.56532    109.2642
-------------+----------------------------------------------------------------
wexper       |
    _predict |
          1  |    48.7461   3.558567    13.70   0.000     41.77143    55.72076
          2  |   8.679914   8.177933     1.06   0.289    -7.348539    24.70837
-------------+----------------------------------------------------------------
wifeage      |
    _predict |
          1  |  -24.77233   2.671484    -9.27   0.000    -30.00834   -19.53632
          2  |   29.12991   4.858785     6.00   0.000     19.60686    38.65295
-------------+----------------------------------------------------------------
kl6          |
    _predict |
          1  |  -485.3448   62.04923    -7.82   0.000    -606.9591   -363.7306
          2  |   273.7007   103.3471     2.65   0.008     71.14416    476.2572
-------------+----------------------------------------------------------------
k618         |
    _predict |
          1  |   11.85766   20.49125     0.58   0.563    -28.30445    52.01978
          2  |   107.1334   38.95688     2.75   0.006      30.7793    183.4875
------------------------------------------------------------------------------

[Katy Tabero]

The results are identical. Just note that tobit outputs t-statistics and cmp outputs z-statistics.

[/CODE]

Thank you for highlighting this and for the correct 'margins' syntax - it all seems to be working now!

The coefficient results are very slightly different, in your example above it is only after 2-3 decimal places they start to differ, so it is minimal, whereas in mine it is a bit more pronounced (in one case by more than 3.5 in the coefficient of one variable). It is odd as when running the Tobit regressions using cmp command individually e.g. with cmp (team_correct= c.endoimposed c.endonotimposed c.exoimposed c.exonotimposed, nocons), ind("cond(team_correct>0, $cmp_cont, $cmp_left)" ) qui), the results are much closer to those provided by the tobit command. So it seems to be something to do with running them simultaneously that throws things off a little.

[Andrew Musau]

Just ignore those small differences, the results are equivalent.