Dear Statalist,
I am trying to figure out how to deal with this model. I want to examine two decisions that are somewhat interdependent: return migration and occupational selection for self-employment. However, I want to control for an earlier selection bias: selection on domestic migration and international migration. Thus, my model is:
1. Migration probability. A multinomial dependent variable (non-migration, internal migration, and international migration) with a sample of non-migrants, current migrants, and returnees.
2. Internal return migration: binomial dependent variable, with a sample of current internal migrants and internal migration returnees. (RETURN2)
3. International return migration: binomial dependent variable, with a sample of current international migrants and international returnees. (RETURN3)
3. Self-employment: binomial dependent variable, with a sample of non-migrants and returnees.
I think that the CMP can be useful here. However, I have doubts about how to model return selection in migration, since this is based on two of the three equations computed in multinomial probit.
I came up with the option of combining the alternative-specific syntax, althoug none of my independent variables do not vary across options.
This gives me a cmp_lf1(): 3200 conformability error.
I am trying to figure out how to deal with this model. I want to examine two decisions that are somewhat interdependent: return migration and occupational selection for self-employment. However, I want to control for an earlier selection bias: selection on domestic migration and international migration. Thus, my model is:
1. Migration probability. A multinomial dependent variable (non-migration, internal migration, and international migration) with a sample of non-migrants, current migrants, and returnees.
2. Internal return migration: binomial dependent variable, with a sample of current internal migrants and internal migration returnees. (RETURN2)
3. International return migration: binomial dependent variable, with a sample of current international migrants and international returnees. (RETURN3)
3. Self-employment: binomial dependent variable, with a sample of non-migrants and returnees.
I think that the CMP can be useful here. However, I have doubts about how to model return selection in migration, since this is based on two of the three equations computed in multinomial probit.
I came up with the option of combining the alternative-specific syntax, althoug none of my independent variables do not vary across options.
Code:
cmp (selfemployment=x d.RETURN2 i.RETURN3 ) (RETURN2 = internalmigration# x i) (RETURN3 = internationalmigration# x i) (stayer: MIGRATION1 = z) (internalmigration: MIGRATION2 = z) (internationalmigration: MIGRATION3 = z) , ind( $cmp_int $cmp_probit $cmp_probit (6 6 6) ) tech(dfp)
mopt__calluser_lf2(): - function returned error
opt__eval_dfp_lf1(): - function returned error
opt__eval(): - function returned error
opt__looputil_iter0_common(): - function returned error
opt__looputil_iter0_dfp(): - function returned error
opt__loop_dfp(): - function returned error
opt__loop(): - function returned error
_moptimize(): - function returned error
Mopt_maxmin(): - function returned error
<istmt>: - function returned error
Mata run-time error
Mata run-time error
opt__eval_dfp_lf1(): - function returned error
opt__eval(): - function returned error
opt__looputil_iter0_common(): - function returned error
opt__looputil_iter0_dfp(): - function returned error
opt__loop_dfp(): - function returned error
opt__loop(): - function returned error
_moptimize(): - function returned error
Mopt_maxmin(): - function returned error
<istmt>: - function returned error
Mata run-time error
Mata run-time error
