Suggested Citation:
Ji, Jean (2025)

Accelerate Plug-in Electric Vehicles Adoption via Understanding Household Adoption Decisions and Designing Sustainable Transportation Finance Policies

. Institute of Transportation Studies, University of California, Davis, Dissertation UCD-ITS-RR-25-62

Abstract:

The transportation sector has been a dominant contributor to greenhouse gas (GHG) emissions in the United States, contributing to 27% of GHG emissions in 2020 (EPA, 2022). In the last decade, vehicle electrification has rapidly taken place, especially in California. By the end of 2023, zero-emissions vehicles (ZEVs) consisted of 24% of all new car sales and accounted for around 5% of the total light-duty vehicles (LDVs) in the state (CEC, 2024). While vehicle electrification is an important strategy for reducing GHG emissions in the transportation sector, this fuel transition has implications for transportation infrastructure funding, which has traditionally been funded by motor fuel taxes. The transition from internal combustion engine vehicles (ICEVs) to ZEVs also invites research questions around the substitutions between these two technologies, especially at the household level. To explore these implications, this dissertation sets out to 1) explore the feasibility of implementing a per-mile road-usage charge (RUC) in replacement of the motor fuel taxes, 2) estimate the changes in vehicles-miles travelled (VMT) and the revenues generated from motor fuel taxes under the backdrop of vehicle electrification, and 3) quantify the effects of vehicle class and fuel type portfolios in ZEV-adopting households’ vehicle replacement decisions. In the first project, the feasibility of integrating RUC programs and tolling was explored to identify potential opportunities to reduce operating and administrative costs. Semi-structured interviews were conducted with experts from tolling programs across the U.S. to identify areas of overlap between tolling and RUC. The interview findings are leveraged to inform the criteria of a multi-criteria decision analysis (MCDA) to evaluate how well the state-level RUC pilots and programs can integrate with tolling systems. The results demonstrate that there are numerous mutual benefits of a RUC-tolling integration. Both the tolling industry and RUC implementations can benefit from the increased scale of operations and the spur of technical innovations, which would reduce administrative costs. RUC programs can also learn from the tolling industry on addressing data privacy and security issues. Lastly, an area that is highly relevant in the rate design of RUC is ensuring equity by alleviating financial burdens on low-income populations and ensuring that unbanked and underbanked populations have access to the system. In the second project, a time-series model is constructed to estimate state-level VMT in California from 2023 to 2030. The model is fitted on county-level sociodemographic and public roadway data, and the model specification which maximize the log-likelihood is an autoregressive integrated moving average (ARIMA) model with two lags of VMT, one-period error term and two lags of explanatory variables. The estimated VMT is leveraged to compute the gasoline excise tax revenues under three scenarios: 1) baseline scenario assuming a constant market share of ZEVs, 2) modest market growth of ZEVs, and 3) more aggressive market growth of ZEVs. The results demonstrate that under both scenarios of modest and aggressive ZEV market growths, the gap in gasoline excise tax would continue to widen between the present and 2030. By 2030, the revenue loss is expected to be around $0.23 billion. To address revenue shortfalls, re-creating the linkage between road usage and payment is a crucial policy solution. The per-mile LDV RUC rate is 3 cents/mile based on the results from this study. Future research will focus on refining the specification of the time-series regression model to explicitly estimate how cost of driving and potential technological innovation such as autonomous vehicle affect VMT. Finally, the last project employs a vehicle transaction model to account for vehicle portfolio preferences when households decide to replace one of their vehicles with a plug-in electric vehicle (PEV). A sample of two-vehicle and PEV-adopting Californian households’ vehicle transactions from 2017 to 2020 was examined via a mixed multinomial logistic regression. The results demonstrate that there are strong complementarities among certain vehicle classes in a two-vehicle household portfolio. Namely, when adopting a PEV, households are more likely to pair a car with a large truck or a SUV than pairing a car with a car. Fuel type complementarities are also observed as households prefer to own a PEV and an ICEV rather than a PEV-PEV portfolio. Our dataset did not include plug-in electric trucks as they became available for sales in the United States in 2021. Future research should collect data on households who have adopted an electric truck and to understand the changes in preference parameters for vehicle class portfolios once the availability of plug-in electric trucks has increased. The implications of this work are two-fold: to estimate PEV-adopting households’ preference parameters for vehicle portfolios and to highlight the importance of vehicle portfolio complementarities in projecting future household vehicle fleet compositions in California