Statalist


[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: st: AW: double hurdle model with Tobit second stage estimation


From   Kirimi Sindi <sindijul@msu.edu>
To   statalist@hsphsun2.harvard.edu
Subject   Re: st: AW: double hurdle model with Tobit second stage estimation
Date   Wed, 25 Feb 2009 14:46:32 -0500

Hi Erickson,

I do not seem to get the paper in The Stata Journal. Actually I have no access to the journal. However, if I get you correctly do I estimate a probit model of the first stage then before estimating the tobit model I write " drop if <var> == 0 ", then estimate the Tobit directly? Will this solve my problem whereas the Tobit model does not account for the joint correlation of the error terms in the two models?

For your information the dependent variable in the second stage is a continuous variable with 70% pile up of zeros not count.

Saha et. al, 1997 ( Applied Economics, 1997, 29, 1311-1316) in their paper on "Calculating Marginal effects for zero expenditures ....." argue that the problem can be solved by adding inverse of the Mills Ration of the first stage Probit. This is equivalent to the two stage Heckman selection model which I am not sure solves my problem.

I am wondering which would help?

Kirimi

Erickson, Ken wrote:
Hi David and 'Statalisters',

My colleague Mike just suggested this:

...Ken,

Can you send my response to him?

You need to use a drop statement   drop if <var> == 0

Do this before the second stage...

See "From the help desk: hurdle models," The Stata Journal, 2003, 3, Number 2, pp.178-184.

Hope this helps!
Ken Erickson

-----Original Message-----
From: owner-statalist@hsphsun2.harvard.edu [mailto:owner-statalist@hsphsun2.harvard.edu] On Behalf Of David Jacobs
Sent: Wednesday, February 25, 2009 1:47 PM
To: statalist@hsphsun2.harvard.edu
Subject: Re: st: AW: double hurdle model with Tobit second stage estimation

You've probably thought of this as well, but what about the -zip- or -znb- models in Stata. Those options allow large proportions of zeros in your outcome variable.

Is the problem with these models that the values above zero are not integers? If they are integers or could be made into integers without messing things up, then these two equation count estimators might work.

Dave Jacobs

At 01:16 PM 2/25/2009, you wrote:
Heckprob is not appropriate in my case. Heckprob would treat the pile of zeros i have as unobserved whereas the Tobit treats them as observed zeros. This is the case in my data and therefore heckprob is not an appropriate model. I could run a first stage Probit then a second stage Tobit. However, the Tobit in the second stage does not use all the data observed and hence the standard errors will not be correct leading to wrong t statistics and be unable to interpret the results.

Kirimi

Martin Weiss wrote:
<>
Have you looked into


*************
help heckprob
*************

?


HTH
Martin


-----Ursprüngliche Nachricht-----
Von: owner-statalist@hsphsun2.harvard.edu
[mailto:owner-statalist@hsphsun2.harvard.edu] Im Auftrag von Kirimi Sindi
Gesendet: Mittwoch, 25. Februar 2009 17:43
An: statalist@hsphsun2.harvard.edu
Betreff: st: double hurdle model with Tobit second stage estimation

Hey,

I working on a double hurdle model. The first stage is a Probit model and the second stage is Tobit due to pile up of zeros. Is there a do program that can assist me in this procedure to get the correct standard errors accounting for the first stage estimation?

Thank you,

Kirimi


--
*******************************
Imagination is more important than knowledge. For while knowledge defines all we currently know and understand, imagination points to all we might yet discover and create.
*******************************

Kirimi Sindi
PhD Candidate
Department of Agricultural,
Food, and Resource Economics
Room 20 Cook Hall
Michigan State University
East Lansing,  MI  48824
Telephone: +1-517-353-5320
Home Tel : +1-517-355-8151
Fax: +1-517-432-1800

*
*   For searches and help try:
*   http://www.stata.com/help.cgi?search
*   http://www.stata.com/support/statalist/faq
*   http://www.ats.ucla.edu/stat/stata/



*
*   For searches and help try:
*   http://www.stata.com/help.cgi?search
*   http://www.stata.com/support/statalist/faq
*   http://www.ats.ucla.edu/stat/stata/

*
*   For searches and help try:
*   http://www.stata.com/help.cgi?search
*   http://www.stata.com/support/statalist/faq
*   http://www.ats.ucla.edu/stat/stata/



--
*******************************
Imagination is more important than knowledge. For while knowledge defines all we currently know and understand, imagination points to all we might yet discover and create.
*******************************

Kirimi Sindi
PhD Candidate
Department of Agricultural,
Food, and Resource Economics
Room 20 Cook Hall
Michigan State University
East Lansing,  MI  48824
Telephone: +1-517-353-5320
Home Tel : +1-517-355-8151	
Fax: +1-517-432-1800

*
*   For searches and help try:
*   http://www.stata.com/help.cgi?search
*   http://www.stata.com/support/statalist/faq
*   http://www.ats.ucla.edu/stat/stata/



© Copyright 1996–2014 StataCorp LP   |   Terms of use   |   Privacy   |   Contact us   |   What's new   |   Site index