A Logit Model With Endogenous Explanatory Variables and Network Externalities

Abstract

A novel logit model is presented that explicitly includes endogeneity in explanatory variables whose values depend on individual choice decisions that involve network externalities or social interactions such as those impacting road congestion or public transport comfort and convenience. The proposed specification corrects for this particular type of endogeneity. The model is derived from a linearly constrained maximum entropy optimization problem that incorporates the network externalities or social interactions causing the endogeneity. It is validated through simulations and an application to a case of transport mode choice in a Chilean city using real data.