Modeling consideration for electric vehicles and plug-in electric vehicles in car renewal

We propose a behaviorally sound model capable of simulating choice set generation for car renewal. It appears at an upper layer of a choice model of a new car. Our model sheds light on potential latent demand for electric cars. We develop a hierarchical latent variable approach to model consideration for electric and plug-in hybrid electric vehicles in car renewal. It drives the observed outcomes that regard how much an individual would account or not for electric cars when planning to renew one. We model it as a latent factor that is defined as a linear combination of attitudes and perceptions. In our application, respondents are asked to whether they would consider or not a 100% electric and a plug-in hybrid electric vehicle. Attitudes and perceptions are also latent factors. These are additional variables that we model. They here are of three types: range anxiety, environmental concerns, and perception about compatibility of use in everyday life. They are measured through a series of questions that concern barriers and motivations to adoption of electric cars. For each of them, respondents are asked to rate on a 5 points Likert scale. We use data from the 2012 Nissan-Renault Alliance survey for application. It covers 5 European countries (France, Germany, Italia, Spain, UK). The sample size is about 5000 observations. In addition to observed outcomes that we model, we have information about individuals, structures of their households, the current cars they own and the way they use them (driven mileages). The measurement model takes the form of a mixed multivariate ordered Logit model. We discuss identification conditions to uncover the structural parameters of our system. It is estimated by maximization of the associated simulated log-likelihood function. Estimates live up to our expectations and are in line with existing literature.