[R] gradually switching regression

[Benjamin Volland]

Hello,
I'm trying to find an algorithm to estimate a switching regression model 
based on the 1990 Economics Letters paper by Ohtani/Kakimoto/Abe or the 
earlier version from 1985 (Ohtani/Katayama, Economic Studies Quarterly; 
assuming as a transition path a polynomial of order 1).

I found an idea for using nls here: 
http://www.biostat.wustl.edu/archives/html/s-news/2000-04/msg00223.html. 
Unfortunately it's based on the 1971 Bacon&Watts model, which does not 
allow obtaining explicitly both start- and end-point of the transition 
phase. If anyone stumbled across such a function or has suggestions on 
how to do it, I would greatly appreciate hearing about them.

Thanks so much
Ben Volland