[Limdep Nlogit List] Ordered Probit with Endogenous Binary Variable(treatment effect)

sshreay at mail.wsu.edu sshreay at mail.wsu.edu
Sun Oct 28 21:15:58 EST 2007


Hi,
To my knowledge this feature is available in NLOGIT 4 (LIMDEP 9). I hope
this will help.
Regards,
Sanatan,
PhD candidate,
Washington State Univ




> Hello,
>
> I'm using Nlogit 3 (Limdep 8) and I've been trying to run an ordered
> probit model with an endogenous binary variable. In the spirit of a
> treatment effect model
>
> I've followed the Help/Manual guidelines, and have written the code
> bellow:
>
> My questions are the follows:
>
> 1) Why the variable that I'm trying to endogenize does not show up in
> the output windows?
>
> 2) the "; All" option does not seem to make any difference on the
> number of observation used when applied after the ORDERED PROBIT
> command. In both cases all the ~2500 obs from the sample are being
> used. The same is not true, after the SELECTION comand (OLS).
>
> 3) Should I include the variable that I'm trying to endogenize on the
> list of Xs that I have? Whenever I tried to do that I get an error
> message, stating that there is no variation. I understand that this
> should be the case without the "; All$" option, however I keep getting
> the same error message even when using the "; All$" option. Curiously
> the same is not true if instead of an ordered probit, I estimate an
> OLS on the second stage (SELECTION).
>
> Many thanks in advance for all the help,
>
> cheers,
>
> JP
>
>
> ?======================================================?
> ? Instruments
> ?
> NAMES ; W = ONE,MUNIC_MO$
> ?
> ? Treatment variables.
> ?
> CREATE ; Z = FUTURA$
> ?
> ? RHS variables in the primary regression
> ?
> NAMES ; X = ONE,SEX1,AGE2,AGE3,EDUCA1,EDUCA2,EDUCA3,EDUCA4,EDUCA5,COR1$
> ?
> ? Dependent variable in the regression in the ordered probit model.
> ?
> CREATE ; Y = DESEJO$
> ?
> ?======================================================?
> ? Selection Equation
> PROBIT ; Lhs = Z ; Rhs = W ; Wts = PESO; Hold results$
> ?
> ? Estimate the ordered probit model and collect results =
> ORDERED PROBIT ; Lhs = Y ; Rhs = X; Wts = PESO ; Select ; All $
> ORDERED PROBIT ; Lhs = Y ; Rhs = X; Wts = PESO ; Select $
> ?
> ?======================================================?
>
> SELECTION;
> 	Lhs=DESEJO;
> 	Rhs=ONE,SEX1,AGE2,AGE3,EDUCA1
> 		,EDUCA2,EDUCA3,EDUCA4,EDUCA5,COR1,FUTURA;
> 	Wts=PESO;
> 	All$
>
> SELECTION;
> 	Lhs=DESEJO;
> 	Rhs=ONE,SEX1,AGE2,AGE3,EDUCA1
> 		,EDUCA2,EDUCA3,EDUCA4,EDUCA5,COR1;
> 	Wts=PESO$
> _______________________________________________
> Limdep site list
> Limdep at limdep.itls.usyd.edu.au
> http://limdep.itls.usyd.edu.au
>




More information about the Limdep mailing list