Dear all.
I would like to estimate a multinomial logit model. My dependant variable y1 is categorial.
I have two endogenous explanatory variables that I want to plug seperately in the model.
One is binary (y2) and the other (y3) is categorial with 3 alternatives.
I thougth about control fonction approach and identified two instruments z1 and z2.
With x representing exogenous variables of the model, a have two questions.
First, is the following approach appropriate for the binary EEV?
Second, how should i handle endogeneity for my categorial EEV (y3)?
Thanks.
I would like to estimate a multinomial logit model. My dependant variable y1 is categorial.
I have two endogenous explanatory variables that I want to plug seperately in the model.
One is binary (y2) and the other (y3) is categorial with 3 alternatives.
I thougth about control fonction approach and identified two instruments z1 and z2.
With x representing exogenous variables of the model, a have two questions.
First, is the following approach appropriate for the binary EEV?
Code:
Probit y2 x z1 z2 predict v2hat, stdp mlogit y1 x y2 v2hat
Thanks.