[Limdep Nlogit List] problems with estimating fixed effects frontier with rh2
Jaap Bos
jwb_bos at yahoo.com
Tue Apr 11 05:15:45 EST 2006
Dear all,
I try to estimate the following type of model:
skip$
frontier;lhs=y;rhs=one,x1,x2;model=t;rh2=one,str*$
frontier;lhs=y;rhs=one,x1,x2;model=t;str=_stratum;rh2=one,str*$
I have 128 fixed effects, recorded as 1,..,128 by _stratum. And str* is a set of dummies for the same 128 firms.
When I estimate this without the rh2 command, everything runs fine. But when I add the rh2 command, the output appears on my screen, but then Limdep shuts off.
I have been able, however, to estimate this for the case where str* has 15 dummy variables (I just can't go beyond that). Also, to see what happens I tried using a different variable to create str* (in case there was some type of multicollinearity problem, which in view should not be the case anyway), but that does not help.
In sum, this looks like a memory type of problem. I have adjusted the project settings, and now have over 600MB allocated. For a data set with 5,438 observations....
Anyone who can offer some help?
thanks, Jaap Bos
More information about the Limdep
mailing list