Stata The Stata listserver
[Date Prev][Date Next][Thread Prev][Thread Next][Date index][Thread index]

st: performing a nested logit model

From   "Kevin Steensma" <>
To   <>
Subject   st: performing a nested logit model
Date   Tue, 5 Nov 2002 17:13:49 -0800

Greetings.  My situation is as follows:  I collected data at time 1 on 150
joint ventures.  At time 2, I observed a four-category outcome -

1)  Goes bankrupt, n=26
2)  remains a joint venture, n=61
3) becomes a foreign owned wholly-owned subsidiary, n=32
4) becomes a locally owned wholly-owned subsidiary, n=31

I conceptualize this as a three-level nested logit model.

1. Do they go bankrupt (n=26), or survive (n=124)?
2. Of those that survive, do they remain a joint venture (n=61), or do they
become a wholly owned subsidiary (n=63)
3. Of those that become a wholly-owned subsidiary, are they foreign owned
(n=32), or locally owned (n=31)

When I run this in LIMDEP using three continuous independent variables, I
get the following:

Error:   803: Hessian is not positive definite at start values.
B0 is too far from solution for Newton method.
Switching to BFGS as a better solution method.
NOTE: Convergence in initial iterations is rarely
at a true function optimum. This may not be a
solution (especially if initial iterations stopped).
Exit from iterative procedure.    1 iterations completed.
Check convergence values shown below.
Gradient value: Tolerance= .1000D-05, current value= .0000D+00
Function chg. : Tolerance= .0000D+00, current value= .0000D+00
Parameters chg: Tolerance= .0000D+00, current value= .0000D+00
Smallest abs. param. change from start value = .0000D+00
At least one parameter did not leave start value.
Normal exit from iterations. Exit status=0.

My questions:

1. Am I expecting too much from my data?  Is nested logit not appropriate?
2. Would STATA have better functionality for nested logit?
3. Could I simply use multinomial logit?  What are the disadvantages?

Any insight would be much appreciated.  Thanks!!

H. Kevin Steensma
Associate Professor of Management and Org.
Gerhardt Faculty Fellow
University of Washington
Mackenzie Hall, Box 353200
Seattle, WA  98195
Fax: 206-685-9392

*   For searches and help try:

© Copyright 1996–2015 StataCorp LP   |   Terms of use   |   Privacy   |   Contact us   |   What's new   |   Site index