Uncovering substitution patterns in new car sales using a cross nested logit model

In the context of sales of new cars it is important to understand and model the consumers substitution patterns as well as their price elasticity towards different types of cars. To do so we develop (i) a multinomial logit model (MNL) and (ii) a cross-nested logit model (CNL) and compare the obtained results. The evolution of the market shares following an increase of the price of one of the alternatives is studied. This is, to the best of our knowledge, the first time that a cross-nested logit model is used to model car-type choice. For modeling purposes, a choice of car is considered to be a choice of market segment (small, medium, full, luxury, o road or multi-purpose vehicle) and a fuel type (gas, diesel, hybrid or electric). For the imputation of the attributes of the unchosen alternatives bootstrapping techniques are used. The model includes attributes of the car such as price and power and socioeconomic characteristics of the respondent such as gender, age, income level, occupation and education.