[Limdep Nlogit List] Dynamic MNL (page E23-25/26) "Expected to find B matrix from previous LOGIT model"

[hielke buddelmeyer]

Hi,
 
Has anyone tried the dynamic multinomial logit model as describes in the manual on pages E23-25 and E23-26? When I follow the steps exactly as described on page E23-26 it always results in the following error message after running the very last bit on page E23-26:
 
"Error 697: Expected to find B matrix from previous LOGIT model."
 
I have tried running it with and without the question mark following Rh2 = z, because it struck me as a potential typo, but it made no difference. 
 
I can make it run by running "MLOGIT; Lhs = yn ; Rhs = x ; Rh2 = z $" just prior to the final MLOGIT on E23-26. This produces the "B matrix from previous LOGIT model" and so solves the error message. But following that sequence estimates a "regular" random effects mnl producing only the "betagama" vector in that case (plus the RPM induced coefficents of course), e.g. as described on E23-20 and E23-21, and not invoke the dynamic mnl version that would produce both the "delta" and "betagama".
 
Of course, if can also just be my data, but the error message struck me as doing something wrong and hence not properly invoking the dynamic mnl estimation. It seems to really expect/need B and not be content with "Start = delta, betagama".
 
Any feedback is appreciated.
 
Cheers,
Hielke
 
 
_________________________________________________________________
Play free games, earn tickets, get cool prizes! Join Live Search Club. 
http://club.live.com/home.aspx?icid=CLUB_wlmailtextlink