Where Are the Electric Vehicles? A Spatial Model for Vehicle-Choice Count Data

摘要: Electric vehicles (EVs) are predicted to increase in market share as auto manufacturers introduce more fuel efficient meet stricter CAFE standards and driver concerns of increasing costs. Reflecting spatial autocorrelation while controlling for a variety demographic locational (e.g., built environment) attributes, this zone-level count model paper offers valuable information power providers charging station location decisions. By anticipating over 745,000 personal-vehicle registrations across sample 1000 census block groups the Philadelphia region, trivariate Poisson-lognormal conditional autoregressive (CAR) anticipates Prius hybrid EV, other conventional vehicle ownership levels. Initial results signal higher EV rates central zones with household incomes, along significant residual autocorrelation, suggesting that spatially-correlated latent variables and/or peer (neighbor) effects on purchase decisions present. Such data sets will become comprehensive informative shares rise. This work’s multivarate CAR modeling approach rigorous, behaviorally-defensible framework patterns choice behavior.