eviews6 can not report R2 in tobit model

For notifying us of what you believe are bugs or errors in EViews.
Please ensure your copy of EViews is up-to-date before posting.

Moderators: EViews Gareth, EViews Moderator

chow
Posts: 1
Joined: Mon Sep 07, 2009 3:47 pm

eviews6 can not report R2 in tobit model

Postby chow » Mon Sep 07, 2009 4:17 pm

hi,
I use the example file "tobit_fair.wf1" in the following path:
"C:\Program Files\EViews6\Example Files\EV6 Manual Data\Chapter 30 - Discrete and Limited Dependent Variable Models"

when estimate eq_tobit again,it can not report R-squared.

It has no problem in eviews5.

How can I solve the problem? Thanks!
Attachments
abc.jpg
abc.jpg (161.08 KiB) Viewed 5681 times

EViews Glenn
EViews Developer
Posts: 2620
Joined: Wed Oct 15, 2008 9:17 am

Re: eviews6 can not report R2 in tobit model

Postby EViews Glenn » Tue Sep 08, 2009 9:49 am

Previous versions of EViews reported *an* R-2 measure which was based one of the residuals formed by taking the observed data minus the expected response. This is a perfectly valid R-2, but it has some undesirable properties and other better ones have been proposed in the literature (the McFadden R-2 is one example). As a result, after some discussion, we decided to not compute an R-2 for tobit models in anticipation of a more general set of goodness-of-fit measures. There are a number of these measures, and to be honest, we don't believe there is yet a consensus on which ones one should look at.

Most of these, however, can easily be computed using EViews. If you would like to propose a candidate measure, we can discuss how to compute it...

jan
Posts: 3
Joined: Mon Apr 26, 2010 3:20 am

Re: eviews6 can not report R2 in tobit model

Postby jan » Tue Apr 27, 2010 7:56 am

Hey,

I have exactly this problem. Could you tell me, how to calculate McFadden's R2 for a Tobit Model in Eviews 6? Would be great!

Jan

EViews Glenn
EViews Developer
Posts: 2620
Joined: Wed Oct 15, 2008 9:17 am

Re: eviews6 can not report R2 in tobit model

Postby EViews Glenn » Tue Apr 27, 2010 10:52 am

It's pretty simple. Estimate the model twice, once with all of the variables, and once without, then compute

1-L_full/L_restricted

where the L are the likelihood functions.

mangalenka
Posts: 3
Joined: Mon May 17, 2010 12:20 am

Re: eviews6 can not report R2 in tobit model

Postby mangalenka » Mon May 17, 2010 12:45 am

EViews Glenn wrote:It's pretty simple. Estimate the model twice, once with all of the variables, and once without, then compute

1-L_full/L_restricted

where the L are the likelihood functions.


Hi I apologise in advance I'm pretty new to Eviews. Can you please clarify what you mean by estimate the model without the variables, so the workfile has blank data? And also when you say compute the above equation- is that done in the Equation Estimation window(see attachment) where under Estimation Settings-Method I choose Binary from the drop down menu? If that is the correct window then can you please run me through the steps for computing "1-L_full/L_restricted"
Attachments
Eviews question 16 May.docx
screen shot of Equation Estimation window
(47.39 KiB) Downloaded 381 times

mangalenka
Posts: 3
Joined: Mon May 17, 2010 12:20 am

Re: eviews6 can not report R2 in tobit model

Postby mangalenka » Mon May 17, 2010 2:15 am

p.s. These 2 files might be useful. The first is my balanced dated panel and the second is the not-Panel. When I estimate probit I get the same outputs does that look correct? Thank you very much!

Maggie
Attachments
Panel estimates.docx
Estimation output
(18.78 KiB) Downloaded 339 times


Return to “Bug Reports”

Who is online

Users browsing this forum: No registered users and 7 guests