Statalist


[Date Prev][Date Next][Thread Prev][Thread Next][Date index][Thread index]

Re: st: Maximization never converges


From   "Pavlos C. Symeou" <p.symeou@jbs.cam.ac.uk>
To   statalist@hsphsun2.harvard.edu
Subject   Re: st: Maximization never converges
Date   Mon, 15 Oct 2007 12:45:21 +0100

Dear Maarten,

thank you for the reply.

The model does not converge if I leave out the firm and time dummies, unless I also reduce the number of variables in the model. However, reduction of the number of variables puts my model in the risk of misspecification. In addition, since I want the model to exploit the panel data information, I cannot keep firm-specific and time-specific effects out, either. I have followed your suggestions below and have found no problems with my data.

Sometimes when I break the estimation (since after a large number of iterations my model does not converge and coefficients do not confront to concavity) STATA gives me the following "Hessian has become unstable or asymmetric (D)".

Is there anything else that I could do?

Maarten buis wrote:

--- "Pavlos C. Symeou" <p.symeou@jbs.cam.ac.uk> wrote:

I am running a maximisation procedure based on Wang (2002, 2003) programme for the estimation of stochastic frontier models. The programme can estimate cross-sectional models and panel data models
if, according to the author, one also includes firm-specific and time-specific dummies. Using my dataset, consisting of 144 countries
and 15 years, the log-likelihood after a number of iterations (say
50)

takes a value which does not change over subsequent iterations (after
more than 1500 iterations with no change I had to break the
procedure). In addition, STATA 10 SE never finds a concave estimate
Every single iteration is not concave. Do you have a way to overcome
this problem?
Does the model converge if you leave out the firm and time dummies?
Also have a look at a cross tabulation of firm v. time and see if there
are sparse or empty cells. Similarly have a look at a cross tabulation
or conditinal means of these dummies with the other explanatory
variables.

Hope this helps,
Maarten

-----------------------------------------
Maarten L. Buis
Department of Social Research Methodology
Vrije Universiteit Amsterdam
Boelelaan 1081
1081 HV Amsterdam
The Netherlands

visiting address:
Buitenveldertselaan 3 (Metropolitan), room Z434

+31 20 5986715

http://home.fsw.vu.nl/m.buis/
-----------------------------------------


___________________________________________________________ Want ideas for reducing your carbon footprint? Visit Yahoo! For Good http://uk.promotions.yahoo.com/forgood/environment.html
*
* For searches and help try:
* http://www.stata.com/support/faqs/res/findit.html
* http://www.stata.com/support/statalist/faq
* http://www.ats.ucla.edu/stat/stata/

*
*   For searches and help try:
*   http://www.stata.com/support/faqs/res/findit.html
*   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