Introduction

Two elements of the paradigm of choice proposed in chapter 1 are central to the development of a basic choice model. These elements are the function that relates the probability of an outcome to the utility associated with each alternative, and the function that relates the utility of each alternative to a set of attributes that, together with suitable utility parameters, determine the level of utility of each alternative. In this chapter, we develop the basic choice model known as the multinomial logit (MNL) model. Beginning with this basic form, making a detailed examination and extending it to accommodate richer behavioural issues is an effective way to understand discrete-choice models in general, and provides a useful vehicle to introduce a wide range of relevant issues.

In section 3.3 the conventional microeconomic demand model with continuous commodities is outlined and used to demonstrate its inadequacy when commodities are discrete. A general theory of discrete-choice is developed around the notion of the existence of population choice behaviour defined by a set of individual behaviour rules, and an indirect utility function that contains a random component. The random component does not suggest that individuals make choices in some random fashion; rather, it implies that important but unobserved influences on choice exist and can be characterised by a distribution in the sampled population, though we do not know where any particular individual is located on the distribution.