dear all,
I'm investigating financing decisions of small firms and I'm using a nested
logit to model them.
When I use STATA 8, I always get dots in the (IV params) section, as follows:
nlogit Dchoice (posschoi= lsdsd_o lsdsd_b lrdinten_o lrdinten_b lage_o
lage_b ) (ext_int= llncapinv_a lincen_a ltgasset_a lcashflo_a llever_a
difflev_a), group(grp)
tree structure specified for the nested logit model
top --> bottom
ext_int posschoi
--------------------------
debt 0
1
autof 2
initial: = -890.61013
rescale: = -890.61013
rescale eq: = -881.41885
Iteration 0: log likelihood = -881.41885
Iteration 1: log likelihood = -881.11793 (backed up)
Iteration 2: log likelihood = -881.11422 (backed up)
Iteration 3: log likelihood = -880.78024 (backed up)
Iteration 4: log likelihood = -880.54555 (backed up)
Iteration 5: log likelihood = -880.43086 (backed up)
Iteration 6: log likelihood = -880.41612 (backed up)
Iteration 7: log likelihood = -880.36322 (backed up)
Iteration 8: log likelihood = -880.32841 (backed up)
Iteration 9: log likelihood = -880.29477 (backed up)
Iteration 10: log likelihood = -879.81536 (backed up)
Iteration 11: log likelihood = -879.79723 (backed up)
Iteration 12: log likelihood = -879.37898
Iteration 13: log likelihood = -879.03583
Iteration 14: log likelihood = -878.85494
BFGS stepping has contracted, resetting BFGS Hessian (0)
Iteration 15: log likelihood = -878.77095
Iteration 16: log likelihood = -878.75297 (backed up)
Iteration 17: log likelihood = -878.75255 (backed up)
Nested logit
Levels = 2 Number of obs = 3972
Dependent variable = Dchoice LR chi2(12) = 1151.62
Log likelihood = -878.75255 Prob > chi2 = 0.0000
------------------------------------------------------------------------------
| Coef. Std. Err. z P>|z| [95% Conf. Interval]
-------------+----------------------------------------------------------------
posschoi |
lsdsd_o | .004309 .0186682 0.23 0.817 -.03228 .040898
lsdsd_b | .0039733 .0186788 0.21 0.832 -.0326365 .0405831
lrdinten_o | -.257973 .4507444 -0.57 0.567 -1.141416 .6254699
lrdinten_b | -.0662926 .446717 -0.15 0.882 -.9418418 .8092566
lage_o | -2.014462 .0959822 -20.99 0.000 -2.202584 -1.826341
lage_b | -.8130758 . . . . .
-------------+----------------------------------------------------------------
ext_int |
llncapinv_a | .1565505 .0318874 4.91 0.000 .0940524 .2190486
lincen_a | -.2021057 .1224135 -1.65 0.099 -.4420317 .0378203
ltgasset_a | .0052008 .0041965 1.24 0.215 -.0030242 .0134257
lcashflo_a | -.0000758 .0000223 -3.41 0.001 -.0001195 -.0000322
llever_a | -.0287271 .0023722 -12.11 0.000 -.0333766 -.0240776
difflev_a | .0520366 .0059348 8.77 0.000 .0404046 .0636687
-------------+----------------------------------------------------------------
(IV params) |
|
ext_int |
/debt | -.0194676 .0893842 -0.22 0.828 -.1946575 .1557223
/autof | 1 . . . . .
------------------------------------------------------------------------------
LR test of homoskedasticity (iv = 1): chi2(0)= 3.46 Prob > chi2 = .
------------------------------------------------------------------------------
.
end of do-file
.
Notice that all regressors are interacted with alternative specific constants
("o" for alternative 0, "b" for alternative 1, "a" for alternative 2 in the
tree above), as all explanatory variables are firm specific attributes. I do
not have choice specific attributes.
I have tried several specifications, some of them include alternative specific
constants, some do not (as the one reported above), but the same result pops
up for the INCLUSIVE VALUE /autof: its coefficient is either 1 or 0.5, but its
standard error is always missing...
What could it be the cause of it?
Many thanks in advance
Mariarosaria
Mariarosaria Agostino, PhD student
Dipartimento di Economia e Statistica
Universit� della Calabria
Arcavacata di Rende (CS)
Italia
-------------------------------------------------
This mail sent through IMP: http://horde.org/imp/
*
* 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/
