> I estimated the nested logit using the maximum likelihood estimation,
> and my problem is with the inclusive value parameter (lambda). I get
> a value of lambda = -27.9 (statistically different from one). I read
> that this parameter is between 0<lambda<1
You could restrict the space of possible parameter values of lambda by a
simple transformation like lambda = 1/(1+exp(-teta)) which falls into
the desired interval. The teta in this expression is the parameter which
enters your likelihood function. Standard errors for lambda can be
obtained by means of the delta method.
Regards,
Joachim Inkmann
Center of Finance and Econometrics
University of Konstanz
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
|