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

st: Things to consider when regressions don't converge


From   "de la Garza, Adrian" <ADelagarza@imf.org>
To   <statalist@hsphsun2.harvard.edu>
Subject   st: Things to consider when regressions don't converge
Date   Tue, 18 May 2004 10:29:10 -0400

Dear Stata list,

I would like to ask you if anyone knows what things I should look at
when my regressions don't converge (log likelihood function not
concave). They had been yielding very sensible results, and everything
converged after 2 or 3 iterations max... and all of a sudden now they
are not converging (or converge after 30+ or 50+ iterations). 

I am running a Heckman selection model (shown below) and after this
non-converging story I started playing around with my equation and
noticed that my -lR- variable might be the one that's giving me trouble.
-lR- is ln(R), and -R- is generated as follows:

sort borrower indic signdate mtydate amount
by borrower: g R = _n

so -R- is the number of bonds issued by a particular borrower at each
different point in time. Now, when I remove -lR- from the equation,
things do converge quickly. I thought that perhaps it was a problem of
-lR- or -R- being correlated with something but that doesn't seem to be
the problem either... and besides, why did I not encounter this issue
before?

Any thoughts would be greatly appreciated.

Thank you,
Adrian


. heckman lspread lamount maturity  g_usip1 lswap embi_sd crtg edtgdp
ggdp90 sd_xgrth bissht_t resimp bnkcrdst pub pub_finance pub_service
pub_utinfr priv priv_finance priv_service priv_utinfr lac yen mark euro
othcur bshare lR /*
> */if crtg<=50, /*
> */select (indic=g_usip1 lswap embi_sd crtg edtgdp tdsxgs ggdp90
sd_xgrth bissht_t resimp ressht bnkcrdst pub priv) 
note: euro dropped due to collinearity

Iteration 0:   log likelihood = -5507.0896  (not concave)
Iteration 1:   log likelihood = -5507.0569  (not concave)
Iteration 2:   log likelihood = -5507.0436  (not concave)
Iteration 3:   log likelihood = -5506.9886  (not concave)
Iteration 4:   log likelihood = -5506.9761  (not concave)
Iteration 5:   log likelihood = -5506.9675  (not concave)
Iteration 6:   log likelihood = -5506.9584  (not concave)
Iteration 7:   log likelihood = -5506.9227  (not concave)
Iteration 8:   log likelihood = -5506.9012  (not concave)
Iteration 9:   log likelihood = -5506.8931  (not concave)
Iteration 10:  log likelihood = -5506.8855  (not concave)
Iteration 11:  log likelihood = -5506.8781  (not concave)
Iteration 12:  log likelihood = -5506.8702  (not concave)
Iteration 13:  log likelihood = -5506.8622  (not concave)
Iteration 14:  log likelihood = -5506.8542  (not concave)
Iteration 15:  log likelihood =  -5506.846  (not concave)
Iteration 16:  log likelihood = -5506.8376  (not concave)
Iteration 17:  log likelihood = -5506.8291  (not concave)
Iteration 18:  log likelihood = -5506.8205  (not concave)
Iteration 19:  log likelihood = -5506.8117  (not concave)
...

*
*   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