Multinomial Logit: A Limited Dependent Variable Technique

Many econometric analyses include dependent variables which are constrained to the interval between zero and one. Under such circumstances simple regression procedures break down. The simple logit model is extended to the multi-factor case. Two alternative models are defined depending on the error structure. The generalized least squares approach assumes the share specification is an accurate representation of the underlying input demand structure. The multinomial maximum likelihood model treats the dependent variable as a probability with a multinomial density. Either model provides a wide array of functional forms while maintaining a straight-forward estimating procedure