[R] Seemingly Unrelated Negative Binomial (SUNB) estimation

[Ben Bolker]

Elisa Lanzi <elisa.lanzi <at> unive.it> writes:

> 
> Dear all,
> I am trying to estimate a system of equations with a Seemingly Unrelated
> Regression. However because of the characteristics of the data I'd like to
> do it with a negative binomial estimation. Would anybody know how to
> implement a Seemingly Unrelated Negative Binomial (SUNB) estimation in R?
> Thank you in advance for your help.
> Best regards,
> Elisa Lanzi
> 

  This PhD thesis

http://etd.library.arizona.edu/etd/metadata.jsp?id=urn:etd:azu_etd_1605_1_m

gives a log-likelihood function (pp. 43-45) and GAUSS code (p. 112)
that wouldn't be too hard to translate into R (using optim or mle).
("Not too hard" implies you already have some reasonable grasp of R/
statistical programming; if you're looking for a black box solution
you will probably have to get some help ...)

  Ben Bolker