12: Hybrid Choice Models with Logit Kernel: Applicability to Large Scale Models¹

Traditionally, an economic analysis of micro-level data views individual agents (household members, firms) as entities that optimize in order to take actions or decisions that are best for them. The widely used discrete choice models represent the decision process through a simple conceptual framework depicted as a black box for which alternative attributes and individual characteristics are inputs and where the observed choice corresponds to the output of the system. Its simplicity has left the door open to critics arguing that it often neglects important idiosyncratic aspects and that it cannot deal with decisions that look irrational. The Hybrid Choice Model’s conceptual framework possesses many ingredients to enhance the behavioural representation of the choice process and therefore address the problems raised above. As a direct result, the choice model specification is improved and it gains in predictive power.