st: signficant cmp update

["David Roodman (droodman@cgdev.org)" <DRoodman@cgdev.org>]

I've added a significant new feature to cmp. You can now fit models that include references to the hypothesized latent linear variables behind observed, censored variables. And these references need not have a recursive structure: there can be simultaneous causation. The references are made by suffixing with a "#".

I think this is easier to explain with examples.

This models foreign as probit. It then models price as depending not on the observed binary outcome foreign, but on the latent linear variable that might be written foreign*:

cmp setup
sysuse auto, clear
cmp (price = foreign#) (foreign = mpg), ind($cmp_cont $cmp_probit) nolr

This estimates the same simultaneous linear equation model with reg3 and cmp:

webuse klein, clear
reg3 (consump wagepriv wagegovt) (wagepriv consump govt capital1), ireg3
cmp (consump = wagepriv# wagegovt) (wagepriv = consump# govt capital1), ind($cmp_cont $cmp_cont) nolr tech(dfp)


The new version can be installed with "ssc install cmp, replace".
--David

*
*   For searches and help try:
*   http://www.stata.com/help.cgi?search
*   http://www.stata.com/support/faqs/resources/statalist-faq/
*   http://www.ats.ucla.edu/stat/stata/