Beyond Adoption: Examining Electric Vehicle Miles Traveled in Households with Zero-Emission Vehicles

Abstract

The environmental benefits of zero-emission vehicles (ZEVs) are affected by both consumer adoption and usage patterns. While numerous studies examine consumers’ stated or revealed preferences for ZEV adoption, ZEV usage patterns have received less attention. Based on the 2019 California vehicle survey data, this paper analyzes the annual mileage of three ZEV types: battery electric vehicle (BEV), plug-in hybrid electric vehicle (PHEV), and fuel cell electric vehicle (FCEV). Results show that ZEVs are driven as much as or more than internal combustion engine vehicles (ICEVs). Furthermore, focusing on households with one ZEV and one or more ICEVs, factors that influence household electric vehicle miles traveled (eVMT) are explored using multiple linear regression models. Greater battery range, home charging capability (regardless of charger type), and provision of special electricity rates for home charging are found to be positively correlated with the eVMT of PHEV households. The eVMT of BEV households is positively associated with Level 2 home charging capability, solar panel installation, access to workplace DC fast charging, and access to public Level 2 and DC fast charging stations. The number of routinely-used public hydrogen refueling stations is associated with higher FCEV household eVMT. Lastly, when high-occupancy vehicle lane access is rated as extremely important in the ZEV purchase decision, greater eVMT is found for both BEV and FCEV households, but not for PHEV households. Results of this study inform policies to encourage eVMT over vehicle miles traveled by ICEV in a household, achieving greater environmental benefits from ZEVs.

Keywords

zero-emission vehicles electric vehicles eVMT battery range charging infrastructure

The transportation sector is the largest source of greenhouse gas (GHG) emissions in the U.S., accounting for 28.2% of total GHG emissions in 2018. Passenger cars and light-duty trucks contribute to over half of the GHG emissions from the transportation sector ( 1 ). Replacing conventional gasoline or diesel-powered internal combustion engine vehicles (ICEVs) with zero-emission vehicles (ZEVs) is a promising path to reduce GHG emissions associated with transportation. Battery electric vehicles (BEVs), plug-in hybrid electric vehicles (PHEVs), and fuel cell electric vehicles (FCEVs) represent the three most common ZEV types on roadways today ( 2 ). When considering full lifecycle (vehicle cycle plus well-to-wheels fuel cycle) emissions, a global average mid-sized BEV, PHEV, and FCEV reduce GHG emissions by 25%, 28.6%, and 21.4%, respectively, compared with a global average ICEV ( 3 ). With continued advances in automotive technology and increasing shares of renewable sources for electricity generation and hydrogen production, future ZEVs will decrease GHG emissions even more compared with their gasoline-powered counterparts.

Numerous studies have examined early adopters and potential adopters of ZEVs, as detailed in the literature review papers ( 4 – 6 ). Given the limited battery range of plug-in electric vehicles (PEVs, which include both PHEVs and BEVs) and limited charging/refueling infrastructure, current ZEVs are more likely to be adopted by multi-car households ( 7 – 11 ). Kurani et al. ( 12 ) conducted a study of 454 multi-car households in California and proposed the “hybrid household hypothesis.” Their findings back up the theory that consumers who adopt ZEVs diversify their vehicle portfolios to take advantage of the strength of various vehicle powertrain types, highlighting the issue of ZEV usage patterns in those households. However, it remains unclear how ZEVs are used in lieu of ICEVs in households that own both, and little is known about the factors that contribute to different ZEV usage patterns in the existing literature. Moving beyond adoption and looking into ZEV usage patterns will expand the understanding of ZEVs’ environmental implications: the more household vehicle miles traveled (VMT) shift to electric vehicle miles traveled (eVMT), the greater the environmental benefits.

This study aims to examine ZEV usage based on the 2019 California Vehicle Survey (CVS) data ( 13 ) with two specific objectives. The first objective is to compare the total usage (i.e., annual mileage) of ZEVs and their ICEV counterparts in a household. The second objective is to investigate factors associated with the variation in eVMT across households. The results offer insights for policy measures to achieve more emission reduction benefits during ZEVs’ usage phase.

Literature Review

Many studies have examined consumer adoption preferences for ZEVs. Compared with ICEV buyers, ZEV early adopters tend to be male, middle or older aged, with high income and high educational attainment, and from multi-car households ( 14 – 21 ). The stated preferences of mainstream consumers in relation to the potential purchase of ZEVs are also extensively studied ( 4 , 6 , 22 –26). These consumer preference studies provide important insights for policymakers to encourage the adoption of ZEVs. Beyond adoption, ZEV usage patterns are also relevant to the emission reduction benefits, particularly when ZEVs are adopted by households that also own ICEVs and make usage choices between the different powertrain types. This literature review section mainly summarizes studies that focus on the usage patterns of ZEVs.

Given the small ZEV market share, the simulation approach has been used by earlier studies to evaluate the potential of ZEV usage because of the difficulty of collecting sufficient real-world usage data. Simulation studies often focus on households with multiple ICEVs. The feasibility of replacing one or more ICEVs with PEVs is examined based on logged GPS trajectory data ( 9 , 27 –29) or household travel survey data ( 30 , 31 ). Although simulation studies show the potential of using PEVs to fulfill household trips, they cannot uncover how these vehicles are used in real-world conditions ( 32 ). Another branch of studies provides empirical evidence on real-world ZEV usage patterns. These empirical studies are summarized below based on data collection methods: (i) telematics devices or on-board recorders; (ii) respondents’ self-reporting in surveys; (iii) household electricity consumption.

Automakers’ Telematics or On-Board GPS Recorders

Automakers’ telematic systems provide accurate data of vehicle operation conditions, which have been used to examine ZEV driving patterns ( 33 – 35 ). Specifically, Plötz et al. ( 33 ) analyzed the utility factors (the ratio of eVMT to total VMT) of PHEVs from Germany and the U.S. They found that PHEVs with battery ranges of 20 km, 40 km, and 60 km showed utility factors of 15%–35%, 40%–50%, and 75%, respectively. Also focusing on utility factors, Geobel and Plötz ( 34 ) analyzed 1,768 Chevrolet Volt PHEV drivers from North America. They found that the most important variables in estimating utility factors according to a linear regression model were the variance and skewness of the daily VMT distributions, as well as the frequency of long-distance driving. Ahmed and Kapadia analyzed usage patterns of Ford BEVs and PHEVs based on voluntary consumers’ in-vehicle operation data. The study found that BEVs showed lower daily driving distances than PHEVs, with the latter having usage patterns similar to the overall vehicle population in Atlanta ( 35 ).

Automakers’ telematics data have high reliability but as they are passively collected, researchers have little control in data collection. The datasets are often limited to a few automakers and contain limited information on users. In contrast, other studies actively recruited owners of electric vehicles and installed on-board GPS devices on the vehicles to obtain their detailed driving patterns ( 32 , 36 ). For example, using data from participants in an electric vehicle trial in Denmark in which drivers had access to a BEV for three months, Jensen and Mabit ( 36 ) examined vehicle choices between BEVs and ICEVs for household daily trips. Results showed that BEVs were preferentially used for well-planned home–work trips (relative to flexible weekend journeys) in comfortable weather conditions. Raghavan and Tal ( 32 ) installed on-board GPS loggers on 153 PHEVs in California. They found disparities between the real-world PHEV utility factors and the theoretical utility factors calculated by the Society of Automotive Engineers J2841 standard. While the vehicle on-board recorder data are also highly reliable, they are often limited by sample size because of time and budget constraints.

Questionnaire Surveys

Despite the potential self-reporting bias, the questionnaire survey approach has frequently been used to study ZEV usage patterns because of the convenience of implementation and the ability to obtain detailed information about ZEV owners. The driving mileage of ICEVs and ZEVs have been compared to explore whether ZEVs can be a good substitute for ICEVs. For example, Davis compared the mileage of PHEVs and BEVs with that of ICEVs based on the 2017 National Household Travel Survey (NHTS) data. The study found that the annual mileage of both PHEVs and BEVs was considerably lower than ICEVs, for both the national sample and the California add-on sample ( 37 ). Also using the 2017 NHTS data, Li et al. ( 38 ) found that the mean and 85th percentile daily trip distances of PHEVs and hybrid electric vehicles were significantly longer than ICEVs, while BEVs showed the shortest trip distances. In contrast, Tal et al. ( 39 ) surveyed 11,269 California ZEV owners. Results showed that BEV owners drove an average of 11,352 mi annually, and PHEV owners 13,028 mi, which is higher than 9,104 to 11,485 mi for ICEVs. Lastly, Hardman and Tal compared the travel patterns of 470 FCEV and 1,550 BEV adopters who applied for the California Clean Vehicle Rebate Project. Results showed no significant differences between the two vehicle types in commute distances or the number of household trips greater than 200 mi ( 10 ).

Other studies further explore influential factors associated with eVMT. Nicholas and Tal examined eVMT in California households with BEVs. The study found that, generally, the longer the BEV range, the more travel was shifted to it as a percent of total household VMT ( 40 ). Tal et al. ( 39 ) examined the usage of both PHEVs and BEVs at the household level in California. This study found that households with longer-range PHEVs or BEVs showed greater eVMT, and households with longer-range BEVs displaced the use of their ICEVs on longer trips. Chakraborty et al. ( 41 ) used repeat survey data of Californian PEV owners and found that battery range and access to Level 2 home chargers had a major impact on PEV VMT. Tal et al. ( 42 ) analyzed eVMT using the charging and driving behaviors of PHEV owners in California. The study found that PHEVs with a longer battery range showed greater eVMT than PHEVs with a shorter battery range. Furthermore, some Toyota Prius PHEV drivers seldom charged their vehicles because of the short battery range and high price of electricity. Without plugging in, PHEVs are used mainly in hybrid mode, limiting their environmental benefits. Chakraborty et al. ( 43 ) further explored the phenomena of PHEV owners who do not charge, using data of 5,418 PHEV owners from California. Results showed that higher home electricity rate, shorter battery range, lower electric motor power-to-vehicle weight ratios, lower potential cost saving from charging, and living in an apartment or condo were contributing factors to PHEV owners’ choice not to charge.

Indirect Electricity Consumption

Lastly, indirect measures such as electricity consumption have also been used to estimate PEV usage. For example, Burlig merged billions of residential electricity meter measurements with address-level PEV registration records in California from 2014 to 2017 ( 44 ). The results suggested that PEVs were driven only 5,300 mi annually, which was about half of the average mileage for ICEVs. These eVMT estimates are much lower than the results from individual surveys ( 39 ). Although this approach is free from the survey approach’s self-reporting bias, it requires many assumptions about non-home charging patterns, which may in turn induce bias for the calculated eVMT.

The literature review identifies several research gaps in existing ZEV usage studies. First, the existing mileage comparison between ZEVs and ICEVs is mainly conducted at the vehicle level without controlling for household effects. The vehicle-level analysis may present self-selection bias. It also fails to examine the role of ZEVs in fulfilling household VMT. Second, most studies rely on early generations of ZEV models (with shorter battery ranges) which may not represent the most recent ZEV usage patterns. Third, few studies have explored influential factors (other than battery range) associated with eVMT. Motivated by these gaps, this study conducts a systematic analysis on the usage patterns of BEVs, PHEVs, and FCEVs, compared with their ICEV counterparts at the household level using the latest 2019 CVS data. Furthermore, the influential factors that contribute to variations in eVMT across households are explored using multiple linear regression models.

Data Preparation

This study is mainly based on the 2019 CVS dataset. The CVS is conducted by California Energy Commission and consists of surveys of both residential vehicle and commercial fleet owners. This paper focuses only on the residential survey sample, which includes 4,248 households, 8,365 people, and 8,049 vehicles. The CVS residential survey collected detailed information about respondents’ socio-economic and household characteristics, as well as household vehicle inventory. A unique feature of this dataset is that it oversamples owners of BEV, PHEVs, and FCEVs, making it feasible to analyze usage patterns of ZEVs.

In this study, the analysis of ZEV usage pattern is mainly conducted at the household level and focuses only on households with one ZEV and one or more ICEVs, which are further divided into two household types: (i) two-car households (one ZEV and one ICEV), (ii) households with three or more vehicles (one ZEV and two or more ICEVs). Other ZEV household types (i.e., with one ZEV and no ICEVs, or with multiple ZEVs) are excluded from the analysis because of their small sample size in the 2019 CVS. Additionally, observations with reported annual VMT greater than 75,000 are removed, based on the same criteria as the VMT analysis in the California 2017 vehicle survey ( 45 ). Furthermore, some PHEV owners’ records are removed because of self-reported unrealistic battery ranges (five reported battery range greater than 100 mi and one reported 4 mi). Finally, 126 PHEV, 184 BEV, and 208 FCEV households that also own ICEVs remain in the analysis (out of 755 ZEV-owning households in the original dataset).

The ZEV makes (brands) in the sample are shown in Table 1. For the PHEV sample, Chevrolet, Toyota, and Ford account for 82% of PHEVs in the dataset. The top three BEV makes in the sample are Tesla, Nissan, and Chevrolet, respectively, which together account for 76% of BEVs in the dataset. The FCEV sample is dominated by the Toyota Mirai and the Honda Clarity, which together represent a 99% share of FCEVs in the dataset. Moreover, the distribution of model years is shown in Figure 1a. The mean ages (as of 2019) of PHEVs, BEVs, and FCEVs are four, three, and three years, respectively. The battery range of the BEV sample shows two clusters (centered around 80 mi and 280 mi, respectively), as shown in Figure 1b. Figure 1c shows the distribution of the PHEV battery range, which ranges from nine to 66 mi. Lastly, as shown in Figure 1d, the PHEV utility factors show a bimodal distribution (centered around 25% and 80%), possibly corresponding to PHEVs with low and high battery ranges, respectively.

Table 1.

Distribution of Zero-Emission Vehicle Makes in the Sample

Plug-in hybrid electric vehicles (PHEVs) sample			Battery electric vehicles (BEVs) sample			Fuel cell electric vehicles (FCEV) sample
Make	Count	Percentage	Make	Count	Percentage	Make	Count	Percentage
Chevrolet	46	37%	Tesla	74	40%	Toyota	133	64%
Toyota	31	25%	Nissan	38	21%	Honda	73	35%
Ford	26	21%	Chevrolet	27	15%	Hyundai	2	1%
Other	23	18%	Other	45	24%	Other	0	0%
Total	126	100%	Total	184	100%	Total	208	100%

The socio-economic characteristics of the sampled ZEV owners are shown in Table 2. Notably, ZEV owners are more likely to be male, middle-aged, with higher educational attainment and higher household income compared with ICEV owners. In addition to the CVS data, county-level variables (population density and public transport mode share) are also included which are gathered from the U.S. Census Bureau American Community Survey.

Table 2.

Socio-Economic Characteristics of the Sampled Zero-Emission Vehicle (ZEV) Owners

	Plug-in hybrid electric vehicle (PHEV) owners (n = 126) (%)	Battery electric vehicle (BEV) owners (n = 184) (%)	Fuel cell electric vehicle (FCEV) owners (n = 208) (%)	Internal combustion engine vehicle (ICEV) owners (n = 6,147) (%)	California population (%)
Male	61	63	64	52	50
Age: 18 or younger	na	na	na	na	23
Age: 18–34	10	9	7	12	24
Age: 35–64	60	70	71	52	38
Age: 65 or older	30	21	22	35	15
Associate degree or lower	16	18	14	43	65
Bachelor	33	35	34	29	22
Graduate	51	47	52	27	13
Full-time employment	55	59	67	42	49
Annual household income
Less than $49,999	4	2	5	17	32
$50,000 to $99,999	16	20	10	30	28
$100,000 to $199,999	40	34	36	31	27
$200,000 or more	28	35	39	12	14
Prefer not to answer	13	9	10	9	0
Household size: 1	3	5	8	17	24
Household size: 2	48	49	46	47	30
Household size: 3 or more	48	46	47	36	46
Number of household vehicles: 1^*	0	0	0	23	31
Number of household vehicles: 2	67	60	57	43	37
Number of household vehicles: 3 or more	33	40	43	34	32

Note:^*The ZEV owner sample used in this analysis includes only households with one ZEV and one or more ICEVs.na = not appliable.

Methods

First, the annual mileage values of ZEVs and ICEVs are compared. Considering the non-normality of annual mileage probability distribution which has a long right tail, the Wilcoxon signed-rank test is used to examine whether the differences between ZEV mileage and ICEV mileage are statistically significant.

Influential factors associated with household eVMT are then explored using multiple linear regression. The eVMT regression models are developed for BEV, PHEV, and FCEV households separately, considering that different ZEV types have different range and charging/refueling requirements. Note that the eVMT for BEV and FCEV households is the total mileage of BEVs and FCEVs, respectively, while the eVMT for PHEV households is equal to PHEV’s electric-only mileage, which is generally lower than the total PHEV mileage. In the 2019 CVS, PHEV owners were asked about their utility factors with the following question, “What percentage of these miles are in electric-only mode?” Although these self-reported utility factors (as shown in Figure 1d) may have bias and errors, the utility factors in the sample are generally in line with existing studies using automakers’ telematics data ( 33 ).

The selection of independent variables consists of two steps. First, the control variables that are relevant to household VMT are included by referencing conventional VMT studies ( 46 , 47 ). Specifically, the model controls for socio-economic variables, household characteristics, and county-level built environment attributes. Then, ZEV-specific variables that can be associated with eVMT are examined, such as battery range and charging/refueling infrastructure availability.

The regression models are shown in the equation below:

Log (eVMT) = α_{0} + α_{1} x_{zev} + α_{2} x_{gas} + α_{3} x_{charge} + α_{4} x_{driver} + α_{5} x_{household} + α_{6} x_{coun} + μ

(1)

where the eVMT takes the natural logarithm form; $x_{zev}$ , $x_{gas}$ , and $x_{charge}$ are independent variables (representing the ZEV characteristics, household ICEV characteristics, and charging infrastructure attributes, respectively); $x_{driver}$ , $x_{household}$ , and $x_{coun}$ are control variables representing ZEV driver, household, and county-level characteristics; $μ$ is the error term; $α_{0}$ to $α_{6}$ are coefficients that are estimated using the Ordinary Least Squares (OLS) estimator. Lastly, variance inflation factor (VIF) scores are calculated for each variable to check for the presence of multicollinearity. None of the VIF scores exceed three, suggesting no serious multicollinearity in the regression model.

Results

ZEV Mileage versus ICEV Mileage

Figure 2a pools all vehicles in the dataset and compares the annual mileage for ICEVs, PHEVs, BEVs, and FCEVs. Their median annual mileage is 8,000 mi, 10,000 mi, 10,000 mi, and 12,000 mi, respectively, suggesting that the mileages of the three types of ZEVs are higher than for ICEVs. However, on average, the age of ICEVs is greater than ZEVs. Thus, Figure 2b further compares annual mileage while controlling for vehicle age effects. For model years earlier than 2010, only ICEVs are available. These older ICEVs show a median annual mileage of 6,500 mi. For model years between 2010 to 2015, the median annual mileage of ICEVs, PHEVs, and BEVs are 10,000 mi, 10,000 mi, 9,000 mi, respectively. Note that FCEVs are not available in this model year range. For model years between 2016 to 2019, the median annual mileage of ICEVs, PHEVs, BEVs, and FCEVs are 10,000, 10,000 mi, 10,000 mi, and 12,000 mi, respectively.

The comparison in Figure 2 does not control for household effects, which may have self-selection bias, since the relationship between annual mileage and ZEV adoption is not clear. Figure 3 shows the mileage comparison within the same household, which helps mitigate self-selection bias. For two-car households (one ZEV and one ICEV), the mileage difference between ZEV and ICEV can be compared directly. For households with three or more vehicles (one ZEV and two or more ICEVs), the ICEV with the highest annual mileage is compared with the ZEV. Paired Wilcoxon signed-rank test is conducted to examine whether the mileage differences are statistically significant.

As shown in Figure 3, the annual mileage of PHEVs and ICEVs does not show a significant difference at the 5% significance level for both household types. The annual mileage differences between BEVs and ICEVs are not significant for households with three or more vehicles, but are significant for the two-car households. For the two-car BEV households (n = 110), BEVs, on average, are driven 750 mi more than the ICEVs annually. Lastly, the differences in annual mileage between FCEVs and ICEVs are highly significant. For both household types, the annual mileage of FCEVs is 2,000 mi higher than ICEVs.

Factors Influencing Household eVMT

Table 3 shows the regression model results for the factors associated with eVMT of PHEV, BEV, and FCEV households, respectively. Considering the relatively small sample size, variables with a large p-value are still kept in the model if they provide interesting insights. The discussion of the model results focuses on both statistical and practical significance. The semi-elasticity is calculated using the formula of $100 \times [\exp (β) - 1]$ , where $β$ is the coefficient estimates associated with the independent variables. For independent variables with p-values higher than the usual levels (10%, 5%, or 1%) and yet large semi-elasticity, the authors acknowledge that the large estimates could be the result of sampling error.

Table 3.

Multiple Linear Regression Model Results for the eVMT of PHEV, BEV, and FCEV Households

Variables	PHEV households: logarithm of eVMT			BEV households: logarithm of eVMT			FCEV households: logarithm of eVMT
	Estimate	Standard error	p-value	Estimate	Standard error	p-value	Estimate	Standard error	p-value
(Intercept)	1.0865	1.38	0.43	–0.2121	0.46	0.65	2.6971	0.48	0.00
Logarithm of household total mileage	0.5473	0.14	0.00	0.9216	0.05	0.00	0.6636	0.05	0.00
Battery range (miles)	0.0305	0.00	0.00	0.0005	0.00	0.11	ns	ns	ns
Average age of internal combustion engine vehicles (ICEVs)	ns	ns	ns	0.0171	0.00	0.00	0.0086	0.00	0.02
Home charging: availability	0.4840	0.27	0.07	ns	ns	ns	ns	ns	ns
Home charging: Level 2 (240 V)	ns	ns	ns	0.0927	0.06	0.11	ns	ns	ns
Special electricity rate for charging PEVs at home	0.2377	0.16	0.15	ns	ns	ns	ns	ns	ns
PEV parks at workplace at least once a week	ns	ns	ns	0.1979	0.08	0.01	ns	ns	ns
Workplace DC fast charger availability	ns	ns	ns	0.1663	0.08	0.03	ns	ns	ns
Public chargers: available at one or more frequent destinations	0.1936	0.17	0.24	ns	ns	ns	ns	ns	ns
Public chargers: Level 2 (240 V) charger	ns	ns	ns	0.1286	0.06	0.03	ns	ns	ns
Public chargers: DC fast charger	0.3520	0.27	0.20	0.1105	0.06	0.09	ns	ns	ns
Number of routinely visited hydrogen refueling locations	ns	ns	ns	ns	ns	ns	0.0435	0.03	0.11
Distance from home to the most used hydrogen refueling station	ns	ns	ns	ns	ns	ns	0.0035	0.00	0.08
High-occupancy vehicle (HOV) lane access is extremely important in purchase decision	ns	ns	ns	0.1104	0.06	0.07	0.0844	0.04	0.06
Driver: male	0.2298	0.13	0.09	0.1213	0.06	0.03	ns	ns	ns
Driver: drives daily	0.4798	0.24	0.05	0.2184	0.09	0.01	0.1416	0.09	0.12
Household: number of vehicles	−0.1554	0.08	0.07	−0.2261	0.03	0.00	−0.1543	0.03	0.00
Household size	ns	ns	ns	−0.0563	0.03	0.04	ns	ns	ns
Solar panel installation	ns	ns	ns	0.1021	0.06	0.07	ns	ns	ns
Number of observations							208
Multiple R-squared							0.525
Adjusted R-squared							0.508

Note: eVMT = electric vehicle miles traveled; PEV = plug-in electric vehicle; PHEV = plug-in hybrid electric vehicle; BEV = battery electric vehicle; FCEV = fuel cell electric vehicle, ns = not significant.

Battery range is positively correlated with eVMT of BEV households, with low statistical and practical significance. With a 1-mi increase in battery range, the eVMT of BEV households is anticipated to increase by only 0.05%. On the other hand, the effect of battery range on the eVMT of PHEV households appears to be both statistically and practically significant: a 3.1% increase in household eVMT is predicted with a 1-mi increase in PHEV battery range. Relevant studies on PHEVs report similar results: PHEVs with greater battery range show higher utility factors ( 33 , 39 , 42 ). These results support the federal purchase incentive policies for PHEVs: greater federal tax credits are available for PHEVs with larger batteries, which have greater potential in replacing household ICEV VMT. Similarly, California requires a minimum battery range (30 mi) for a PHEV to be eligible for the state rebate since April 2021. These model results suggest that more flexible PHEV incentives tied to the battery range can lead to greater environmental benefits.

Characteristics of household ICEVs are also found to be relevant for the eVMT of BEV and FCEV households. The older the household ICEVs, the greater the household eVMT. Specifically, when the age of household ICEVs increases by one year, the anticipated eVMT of BEV and FCEV households increase by 1.7% and 0.9%, respectively. For PHEV households, however, the age of household ICEVs is found to be hardly significant for household eVMT. Additionally, the average fuel economy of household ICEVs has also been tested as an independent variable, and shows highly insignificant results for the eVMT of all three types of ZEV households. In other words, the age of household ICEVs, rather than the fuel economy, better predicts the household eVMT. Furthermore, removing the fuel economy variable does not change the coefficient estimate of ICEV age, thus the variable of ICEV fuel economy is excluded from the final model in Table 3.

Home charging infrastructure is found to be correlated with the eVMT of BEV and PHEV households, with some nuances based on charger type. Home charging capability (regardless of the type of charger) is positively correlated with the eVMT of PHEV households, with a 10% statistical significance level and a practically large impact: PHEV households with home charging infrastructure show 62% greater eVMT than those without. Note that the type of home charger appears to be irrelevant for the eVMT of PHEV households: access to Level 2 charging at home is found to be highly insignificant. In contrast, only having Level 2 charging at home is associated with greater eVMT of BEV households, with a marginal statistical significance (p-value = 0.11). According to the BEV model estimates, accessing at-home Level 2 charging is associated with a 9.7% increase in the eVMT of BEV households.

Furthermore, when PHEV households are offered special electricity rates for charging at home, they show a 27% increase in household eVMT. Interestingly, about half of respondents from BEV households mentioned that they were provided with a special electricity rate for charging at home. However, their household eVMT is not much different than those BEV households without special electricity rates. These results suggest that maximizing environmental benefits of PEVs necessitates different incentives for different powertrain types, such as providing incentives for the installation of Level 2 home charging for BEVs owners, and discount electricity rates for home charging for PHEVs owners.

Commuting and workplace charging characteristics are also found to be relevant for BEV household eVMT. BEV households commuting by BEV at least once a week show 21.9% higher eVMT than those who never use their BEVs for commuting. Furthermore, access to DC fast charging at the workplace is positively correlated with eVMT of BEV households, at a 5% statistical significance level. Practically, workplace DC fast charging provision shows an 18.1% increase in the eVMT of BEV households. For PHEV households, however, household eVMT appears to be irrelevant to whether PHEVs are used for commuting or not, and also to be irrelevant to the availability of workplace DC fast charging.

Public charging/refueling infrastructure is found to be correlated with household eVMT. Compared with those having no access to public charging at frequent destinations, PHEV households reporting public charging availability at one or more frequent destinations show 21.4% higher eVMT, but with low statistical significance. Practically, access to DC fast chargers shows a 42.2% increase in the eVMT of PHEV households. Despite low statistical significance (p-value = 0.20), the practically significant effect of DC fast chargers is unexpected since very few PHEV models have DC fast charging capabilities. It is possible that the DC fast charger variable in the PHEV eVMT model carries some unobserved factors which warrant further investigation. For BEV households, having access to public Level 2 and DC fast charging is associated with a 13.7% and 11.7% increase in the eVMT of BEV households, at 5% and 10% statistical significance levels, respectively. For FCEV households, the number of hydrogen refueling stations used routinely is positively correlated with household eVMT, with a p-value of 0.11. For each additionally available hydrogen refueling station, a 4.45% increase in eVMT of FCEV households is predicted. A recent study in Canada ( 48 ) showed that charging and refueling infrastructure deployment played minimal roles in consumers’ adoption of ZEVs. From the ZEV usage perspective, however, these results indicate that provision of charging/refueling infrastructure is important to achieve maximum household eVMT. Interestingly, the distance to the most frequently used hydrogen refueling station is positively correlated with the eVMT of FCEV households, despite the low practical significance. This is possibly because of FCEV households making dedicated trips or traveling out-of-the-way to refuel their FCEVs (rather than refueling along the route for other trip purposes).

When citing high-occupancy vehicle (HOV) lane access as extremely important in their purchase decision, both BEV and FCEV households show greater eVMT (11.7% and 8.8% increase, respectively) compared with those households that rate lower importance for HOV lane access. For PHEV households, however, the eVMT is irrelevant to the importance of HOV lane access in the purchase decision. A recent study in California by Chakraborty et al. ( 43 ) showed that consumers who indicated HOV lane access was important in their PHEV purchase decision were less likely to charge their vehicles and use the vehicles primarily in hybrid mode. These results encourage more research and discussion around the policy of providing HOV lane access for ZEVs. Although such an incentive can increase the adoption of PHEVs ( 49 ), it does not necessarily increase the eVMT of PHEV households. If the ultimate goal is to replace more ICEV VMT with eVMT, it would seem to be more effective to provide HOV lane access only to BEVs or FCEVs. However, abolishing the HOV lane access for PHEVs may be unfair to those owners of longer-range PHEVs, which have shown their potential to improve household eVMT. Perhaps, like the PHEV incentives at the adoption stage where federal tax credits are tiered by battery capacity, incentive policy at the usage stage (i.e., HOV lane access) should also be tied to the PHEV battery range.

The models also include control variables related to ZEV driver, household, and several county-level variables. For PHEV and BEV households, the gender effects are found to be both statistically and practically significant, particularly for PHEVs. PHEVs and BEVs which are mainly driven by males show 25.8% and 12.9% more eVMT, respectively, than those that are mainly driven by females. In contrast, the gender effect on eVMT is hardly significant for FCEV households. As expected, the number of household ICEVs is a highly significant predictor of household eVMT. For each additional ICEV in the household, PHEV, BEV, and FCEV households are predicted to have a 14.4%, 20.2%, and 14.3% decrease in eVMT, respectively. Solar panel installation at the household residence appears to be positively correlated with the eVMT of BEV households, at a 10% statistical significance level. Practically, households with solar panels at their residence show 10.7% more eVMT than those without. In contrast, the solar panel installation effect is not significant for the eVMT of PHEV households. Lastly, the county-level population density and public transport mode share are found to be insignificant determinants for the eVMT of all three ZEV households, after controlling for all other variables.

Conclusion and Future Work

Many studies have examined the consumer preferences for ZEV adoption. However, ZEV usage patterns, which affect their potential environmental benefits, are less studied. Using 2019 CVS data, this paper investigates the household usage patterns of PHEVs, BEVs, and FCEVs. Results show that all three types of ZEVs are driven at least as much as ICEVs. Furthermore, eVMT regression model results reveal nuances of determinants (e.g., battery range, charging availability, and HOV lane access) for eVMT of different ZEV types, highlighting the need to analyze ZEV usage patterns separately by powertrain types. To achieve greater environmental benefits, certain policies to encourage ZEV usage to replace more household ICEV mileage should therefore vary by powertrain types. For example, PHEV incentives (e.g., rebates, HOV lane access) should be more flexible and tied to battery range, considering PHEVs’ eVMT patterns. In addition, incentives for the installation of Level 2 home charging and discounted home charging electricity rates are effective in increasing the eVMT of BEV and PHEV households, respectively. Lastly, more public charging/refueling infrastructure is positively correlated with the eVMT for BEV and FCEV households.

This study has several limitations that warrant further research. First, the sample is limited to ZEV households in California, which has its unique set of electric vehicle policies, road networks, and user behavior. These local contexts should be kept in mind when interpreting model results. Additional studies from other regions are needed to acquire a more comprehensive understanding of eVMT patterns. Moreover, this study focuses only on households with one ZEV and one or more ICEVs. With the increasing adoption of ZEVs, future research can sample single-ZEV households or households owning multiple ZEVs, which inevitably present different usage patterns.

Second, the findings of this study, based on the 2019 CVS data, contradict those studies based on early adopters of ZEVs with a limited battery range ( 37 , 44 ). Still, this paper only uncovers a snapshot-in-time of ZEV usage patterns. The dynamics of ZEV usage patterns changing over time should be an avenue for future research for two reasons: (i) the ZEV technology keeps evolving, and (ii) the characteristics of current ZEV owners may differ from future mainstream ZEV owners ( 50 ). Such longitudinal research thus helps design dynamic ZEV regulations and policies that are based on the evolution of ZEV usage patterns.

Third, this study can be complemented with qualitative research methods such as semi-structured interviews to understand the underlying causes of specific ZEV usage patterns. For example, qualitative studies could uncover why male drivers, all else being equal, show higher eVMT for both PHEVs and BEVs when compared with female drivers, while the gender effect is not significant among FCEV owners. The mixed study approach would improve our understanding of ZEV usage patterns beyond the financial and practical factors, and delves into the role of lifestyles, motivations, and attitudinal elements ( 21 ).

Fourth, survey respondents self-report their car annual mileage, which may or may not reflect their true mileage. Although data collection would be more difficult and costly, using automakers’ telematics data or on-board GPS data would provide more reliable vehicle mileage. Finally, this research investigates determinants that contribute to household eVMT with the underlying assumption that higher household eVMT replaces ICEV VMT, implying greater environmental benefits. However, increased eVMT could also be induced VMT associated with ZEVs instead of replacement ICEV VMT, such as the situation with FCEVs traveling more miles on dedicated refueling station trips, which would negate the environmental benefit of increasing eVMT.

Footnotes

Author Contributions

The authors confirm contribution to the paper as follows: study conception and design: W. Jia, T. D. Chen; data collection: W. Jia, T. D. Chen; analysis and interpretation of results: W. Jia, T. D. Chen; draft manuscript preparation: W. Jia, T. D. Chen. All authors reviewed the results and approved the final version of the manuscript.

Declaration of Conflicting Interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This study is funded by Mid-Atlantic Transportation Sustainability University Center (MATS UTC) Grant No: 146221.

ORCID iDs

Wenjian Jia

T. Donna Chen

Data Accessibility Statement

The data used for this analysis are publicly available at Transportation Secure Data Center, .

References

United States Environmental Protection Agency. Inventory of U.S. Greenhouse Gas Emissions and Sinks: 1990-2018. EPA 430-R-20-002. https://www.epa.gov/ghgemissions/inventory-us-greenhouse-gas-emissions-and-sinks-1990-2018.

California Air Resource Board. Zero-Emission Vehicle Program. https://ww2.arb.ca.gov/our-work/programs/zero-emission-vehicle-program/about. Accessed January 18, 2022.

International Energy Agency. Global EV Outlook 2019: Scaling-up the Transition to Electric Mobility. International Energy Agency, 2019.

Liao

Molin

van Wee

Consumer Preferences for Electric Vehicles: A Literature Review. Transport Reviews, Vol. 37, No. 3, 2017, pp. 252–275. https://doi.org/10.1080/01441647.2016.1230794.

Hardman

Understanding the Impact of Reoccurring and Non-Financial Incentives on Plug-in Electric Vehicle Adoption – A Review. Transportation Research Part A: Policy and Practice, Vol. 119, 2019, pp. 1–14. https://doi.org/10.1016/j.tra.2018.11.002.

Singh

Vaibhav

A Review and Simple Meta-Analysis of Factors Influencing Adoption of Electric Vehicles. Transportation Research Part D: Transport and Environment, Vol. 86, 2020, p. 102436. https://doi.org/10.1016/j.trd.2020.102436.

Klöckner

C. A.

Nayum

Mehmetoglu

Positive and Negative Spillover Effects from Electric Car Purchase to Car Use. Transportation Research Part D: Transport and Environment, Vol. 21, 2013, pp. 32–38. https://doi.org/10.1016/j.trd.2013.02.007.

Figenbaum

Kolbenstvedt

Electromobility in Norway: Experiences and Opportunities with Electric Vehicles. TØI Report 1281/2013. Institute of Transport Economics, Norwegian Center for Transport Research, Oslo, Norway, 2013.

Karlsson

What Are the Value and Implications of Two-Car Households for the Electric Car?

Transportation Research Part C: Emerging Technologies, Vol. 81, 2017, pp. 1–17. https://doi.org/10.1016/j.trc.2017.05.001.

10.

Hardman

Tal

Who Are the Early Adopters of Fuel Cell Vehicles?

International Journal of Hydrogen Energy, Vol. 43, No. 37, 2018, pp. 17857–17866. https://doi.org/10.1016/j.ijhydene.2018.08.006.

11.

Ferguson

Mohamed

Higgins

C. D.

Abotalebi

Kanaroglou

How Open Are Canadian Households to Electric Vehicles? A National Latent Class Choice Analysis with Willingness-to-Pay and Metropolitan Characterization. Transportation Research Part D: Transport and Environment, Vol. 58, 2018, pp. 208–224. https://doi.org/10.1016/j.trd.2017.12.006.

12.

Kurani

K. S.

Turrentine

Sperling

Testing Electric Vehicle Demand in “Hybrid Households” Using a Reflexive Survey. Transportation Research Part D: Transport and Environment, Vol. 1, No. 2, 1996, pp. 131–150. https://doi.org/10.1016/S1361-9209(96)00007-7.

13.

Transportation Secure Data Center. National Renewable Energy Laboratory, Golden, Colorado. www.nrel.gov/tsdc. Accessed June 8, 2020.

14.

Trommer

Jarass

Kolarova

Early Adopters of Electric Vehicles in Germany Unveiled. World Electric Vehicle Journal, Vol. 7, No. 4, 2015, pp. 722–732. https://doi.org/10.3390/wevj7040722.

15.

Figenbaum

Kolbenstvedt

Learning from Norwegian Battery Electric and Plug-in Hybrid Vehicle Users: Results from a Survey of Vehicle Owners. TØI report 1492/2016, Institute of Transport Economics, Norwegian Center for Transport Research, Oslo, Norway, 2016.

16.

Hardman

Shiu

Steinberger-Wilckens

Comparing High-End and Low-End Early Adopters of Battery Electric Vehicles. Transportation Research Part A: Policy and Practice, Vol. 88, 2016, pp. 40–57. https://doi.org/10.1016/j.tra.2016.03.010.

17.

Hardman

Tal

Exploring the Decision to Adopt a High-End Battery Electric Vehicle: Role of Financial and Nonfinancial Motivations. Transportation Research Record: Journal of the Transportation Research Board, 2016. 2572: 20–27. https://doi.org/10.3141/2572-03.

18.

Javid

R. J.

Nejat

A Comprehensive Model of Regional Electric Vehicle Adoption and Penetration. Transport Policy, Vol. 54, 2017, pp. 30–42. https://doi.org/10.1016/j.tranpol.2016.11.003.

19.

Vassileva

Campillo

Adoption Barriers for Electric Vehicles: Experiences from Early Adopters in Sweden. Energy, Vol. 120, 2017, pp. 632–641. https://doi.org/10.1016/j.energy.2016.11.119.

20.

Westin

Jansson

Nordlund

The Importance of Socio-Demographic Characteristics, Geographic Setting, and Attitudes for Adoption of Electric Vehicles in Sweden. Travel Behaviour and Society, Vol. 13, 2018, pp. 118–127. https://doi.org/10.1016/j.tbs.2018.07.004.

21.

Axsen

Cairns

Dusyk

Goldberg

What Drives the Pioneers? Applying Lifestyle Theory to Early Electric Vehicle Buyers in Canada. Energy Research & Social Science, Vol. 44, 2018, pp. 17–30. https://doi.org/10.1016/j.erss.2018.04.015.

22.

Rezvani

Jansson

Bodin

Advances in Consumer Electric Vehicle Adoption Research: A Review and Research Agenda. Transportation Research Part D: Transport and Environment, Vol. 34, 2015, pp. 122–136. https://doi.org/10.1016/j.trd.2014.10.010.

23.

Coffman

Bernstein

Wee

Electric Vehicles Revisited: A Review of Factors that Affect Adoption. Transport Reviews, Vol. 37, No. 1, 2017, pp. 79–93. https://doi.org/10.1080/01441647.2016.1217282.

24.

Long

Chen

Geng

A Review of Factors Influencing Consumer Intentions to Adopt Battery Electric Vehicles. Renewable and Sustainable Energy Reviews, Vol. 78, 2017, pp. 318–328. https://doi.org/10.1016/j.rser.2017.04.076.

25.

Kumar

R. R.

Alok

Adoption of Electric Vehicle: A Literature Review and Prospects for Sustainability. Journal of Cleaner Production, Vol. 253, 2020, p. 119911. https://doi.org/10.1016/j.jclepro.2019.119911.

26.

Jia

Chen

T. D.

Are Individuals’ Stated Preferences for Electric Vehicles (EVs) Consistent with Real-World EV Ownership Patterns?

Transportation Research Part D: Transport and Environment, Vol. 93, 2021, p. 102728. https://doi.org/10.1016/j.trd.2021.102728.

27.

Khan

Kockelman

K. M.

Predicting the Market Potential of Plug-in Electric Vehicles Using Multiday GPS Data. Energy Policy, Vol. 46, 2012, pp. 225–233. https://doi.org/10.1016/j.enpol.2012.03.055.

28.

Tamor

M. A.

Milačić

Electric Vehicles in Multi-Vehicle Households. Transportation Research Part C: Emerging Technologies, Vol. 56, 2015, pp. 52–60. https://doi.org/10.1016/j.trc.2015.02.023.

29.

Björnsson

L.-H.

Karlsson

Electrification of the Two-Car Household: PHEV or BEV?

Transportation Research Part C: Emerging Technologies, Vol. 85, 2017, pp. 363–376. https://doi.org/10.1016/j.trc.2017.09.021.

30.

Laberteaux

Hamza

Highlighting the Differential Benefit in Greenhouse Gas Reduction via Adoption of Plugin Hybrid Vehicles for Different Patterns of Real Driving. SAE Technical Paper 2017-01-1178. 2017. https://doi.org/10.4271/2017-01-1178.

31.

Khayati

Kang

J. E.

Comprehensive Scenario Analysis of Household Use of Battery Electric Vehicles. International Journal of Sustainable Transportation, Vol. 14, No. 2, 2020, pp. 85–100. https://doi.org/10.1080/15568318.2018.1529210.

32.

Raghavan

S. S.

Tal

Plug-in Hybrid Electric Vehicle Observed Utility Factor: Why the Observed Electrification Performance Differ from Expectations. International Journal of Sustainable Transportation, Vol. 16, No. 2, 2020, pp. 1–46. https://doi.org/10.1080/15568318.2020.1849469.

33.

Plötz

Funke

S. A.

Jochem

Wietschel

CO₂ Mitigation Potential of Plug-in Hybrid Electric Vehicles Larger than Expected. Scientific Reports, Vol. 7, No. 1, 2017, p. 16493. https://doi.org/10.1038/s41598-017-16684-9.

34.

Goebel

Plötz

Machine Learning Estimates of Plug-in Hybrid Electric Vehicle Utility Factors. Transportation Research Part D: Transport and Environment, Vol. 72, 2019, pp. 36–46. https://doi.org/10.1016/j.trd.2019.04.008.

35.

Ahmed

N. K.

Kapadia

Big Data Analytics: How Big Data Is Shaping Our Understanding of Electrified Vehicle Customers. SAE International Journal of Materials and Manufacturing, Vol. 10, No. 3, 2017, pp. 351–359.

36.

Jensen

A. F.

Mabit

S. L.

The Use of Electric Vehicles: A Case Study on Adding an Electric Car to a Household. Transportation Research Part A: Policy and Practice, Vol. 106, 2017, pp. 89–99. https://doi.org/10.1016/j.tra.2017.09.004.

37.

Davis

L. W.

How Much Are Electric Vehicles Driven?

Applied Economics Letters, Vol. 26, No. 18, 2019, pp. 1497–1502. https://doi.org/10.1080/13504851.2019.1582847.

38.

Liu

Jia

Ownership and Usage Analysis of Alternative Fuel Vehicles in the United States with the 2017 National Household Travel Survey Data. Sustainability, Vol. 11, No. 8, 2019, p. 2262. https://doi.org/10.3390/su11082262.

39.

Tal

Raghavan

Karanam

Favetti

Sutton

Lee

J. H.

Nitta

Chakraborty

Nicholas

Turrentine

Advanced Plug-In Electric Vehicle Travel and Charging Behavior Final Report. UC Davis, Plug-in Hybrid & Electric Vehicle Research Center of the Institute of Transportation Studies, 2020.

40.

Nicholas

Tal

eVMT in the Household Fleet: Integrating Battery Electric Vehicles into Household Travel. Presented at 95th Annual Meeting of the Transportation Research Board, Washington, D.C., 2016.

41.

Chakraborty

Hardman

Tal

Integrating Plug-in Electric Vehicles (PEVs) into Household Fleets - Factors Influencing Miles Traveled by PEV Owners in California. Travel Behaviour and Society, Vol. 26, 2022, pp. 67–83. https://doi.org/10.1016/j.tbs.2021.09.004.

42.

Tal

Nicholas

M. A.

Davies

Woodjack

Charging Behavior Impacts on Electric Vehicle Miles Traveled: Who Is Not Plugging In?

Transportation Research Record: Journal of Transportation Research Board, 2014. 2454: 53–60. https://doi.org/10.3141/2454-07.

43.

Chakraborty

Hardman

Tal

Why Do Some Consumers Not Charge Their Plug-in Hybrid Vehicles? Evidence from Californian Plug-in Hybrid Owners. Environmental Research Letters, Vol. 15, 2020, p. 084031. https://doi.org/10.1088/1748-9326/ab8ca5.

44.

Burlig

Bushnell

Rapson

Wolfram

Low Energy: Estimating Electric Vehicle Electricity Use. No. w28451. National Bureau of Economic Research, 2016.

45.

Fowler

Cherry

Adler

Bradley

Richard

2015–2017 California Vehicle Survey. California Energy Commission. CEC-200-2018-006, 2018.

46.

Brownstone

Key Relationships Between the Built Environment and VMT. Special Report 298. Transportation Research Board, 2008.

47.

Singh

A. C.

Astroza

Garikapati

V. M.

Pendyala

R. M.

Bhat

C. R.

Mokhtarian

P. L.

Quantifying the Relative Contribution of Factors to Household Vehicle Miles of Travel. Transportation Research Part D: Transport and Environment, Vol. 63, 2018, pp. 23–36. https://doi.org/10.1016/j.trd.2018.04.004.

48.

Miele

Axsen

Wolinetz

Maine

Long

The Role of Charging and Refuelling Infrastructure in Supporting Zero-Emission Vehicle Sales. Transportation Research Part D: Transport and Environment, Vol. 81, 2020, p. 102275. https://doi.org/10.1016/j.trd.2020.102275.

49.

Jenn

Lee

J. H.

Hardman

Tal

An In-Depth Examination of Electric Vehicle Incentives: Consumer Heterogeneity and Changing Response over Time. Transportation Research Part A: Policy and Practice, Vol. 132, 2020, pp. 97–109. https://doi.org/10.1016/j.tra.2019.11.004.

50.

Axsen

Goldberg

Bailey

How Might Potential Future Plug-in Electric Vehicle Buyers Differ from Current “Pioneer” Owners?

Transportation Research Part D: Transport and Environment, Vol. 47, 2016, pp. 357–370. https://doi.org/10.1016/j.trd.2016.05.015.