Factors Affecting Electric Vehicle Adoption at the Zip Code Level
Abstract
It is widely recognized that a requisite aspect of addressing climate goals is to develop a more sustainable transportation sector. One initiative towards this is the federal administration’s stated goal that 50% of all new vehicle sales will be electric by the year 2030. However, it is a common consensus that this will not occur without significant changes in electric vehicle (EV) adoption trends. In order to meet this goal and significantly diminish transportation greenhouse gas emissions, it is critical to better understand EV adoption at scale. To do this, it is necessary to understand at the system level what the progression of adoption will look like and what factors influence that adoption. This problem requires a more granular analysis than has been previously performed. Adoption at the ZIP code level in four US states (CA, CO, NY, WA) is analyzed using historical data dating to 2011. To understand the progression of adoption, two adoption models (the logistic model and the Bass model) are considered to forecast future EV levels in ZIP codes. The logistic is better for the data that is currently publicly available. It is additionally found that EV forecasts must be decomposed into both battery electric vehicle (BEV) and plug-in hybrid electric vehicle (PHEV) forecasts. There is sufficient evidence that the adoption processes for these two types of EVs differ. This analysis is further extended to consider the factors influencing adoption. Utilizing the adoption forecasts, spatial regression analyses are performed on the parameters that define the forecast shapes. These examine how multiple sociodemographic, land use, and charging availability measures correlate with the rate of EV adoption and the lateral shift of early EV adoption. It is found that multiple measures of charging infrastructure availability correspond with increased adoption; of these, a variation on the distance to fast-charging stations is the most consistent metric across final models. Additionally land use type is found to be relevant to adoption. Finally, this work is able to corroborate at a granular spatial level numerous sociodemographic variables from the literature. Ultimately, this research can provide valuable insights into adoption trends at a local level and what factors may be best leveraged to promote adoption.
Degree
M.Sc.
Advisors
Gkritza, Purdue University.
Subject Area
Civil engineering|Demography|Economics|Marketing|Sociology|Statistics
Off-Campus Purdue Users:
To access this dissertation, please log in to our
proxy server.