A Multifaceted Equity Metric System for Transportation Electrification

Transportation electrification offers societal benefits like reduced emissions and decreased dependence on fossil fuels. Understanding the deployment of electric vehicles (EVs) and electric vehicle supply equipment (EVSE) has been a popular focus, however, achieving their equitable distribution in the transportation system remains a challenge for successful electrification. To address this issue, this paper proposes a multi-dimensional equity metric system that assesses the equity status in the impacts of EV and EVSE deployment across different socio-demographic groups. Four types of equity are considered in the equity metric system: a fair share of resources and external costs that are grouped into horizontal equity, as well as inclusivity and affordability that refer to vertical equity. This paper performs a case study to examine equity concerns regarding the adoption of EVs and EVSE in Los Angeles County in 2035 by leveraging the proposed equity metric system. The results reveal disparities in the adoption of EVs and public chargers, as well as variations in EV trips and economic status across different socio-demographic groups. These disparities underscore the urgency to address equity issues during electrification. Building upon the results, this study puts forth recommendations to tackle these equity challenges to provide valuable insights for local agencies.


I. INTRODUCTION
T RANSPORTATION systems are undergoing fundamen- tal changes, including transition towards electrification, implementation of new technologies represented by connected and automated vehicle, and adoption of renewable energy.This paradigm shift brings considerable benefits to society, such as climate change adaptation, air pollution reduction, safety improvement, road capacity enhancement, and human well-being improvement [1], [2], [3], [4], [5], [6], [7], [8], [9], [10], [11], [12], [13], [14].However, regarding the transportation electrification, the electric vehicle supply equipment (EVSE, refers to public chargers in this study) for enabling the EV mobility, is insufficient at this stage, making the EV charging inconvenient and the coast-to-coast EVSE network difficult to implement [15].This gap in The review of this article was arranged by Associate Editor Jia Hu.
EVSE deployment has inspired plenty of research focusing on statewide and nationwide charger need predictions and distributions, either through approaches based on the real-life data from test EV fleets and charger stations or based on people's travel behavior derived from travel survey or transportation simulations [1], [16], [17], [18].However, the existing literature seeks to maximize the efficient charging service for existing and candidate EV drivers, paying ultra-attention to locations with a high-charging need.This neglects the underserved areas that have low-charging need due to the limited EV adoption.Such an issue can induce an unfair distribution of EVSE in the transportation system, which reduces people's willingness to purchase EVs at underserved areas and undermines the overall societal benefits of electrification.
Equity evaluation measuring the distribution impact of resources and investments is a necessary component for gaining deeper insight into the implementation of EVSE deployment decisions, reducing the vulnerability to neglect the underserved areas [19], [20].In this study, we propose a multifaceted equity metric system for the electrified transportation system to quantify the geographical equity issues in the socio-demographic context.First, existing research developed by us is introduced for generating the public charger deployment plan for large-scale transportation networks according to people's travel trajectory and charging demand in time and space [15].Second, we assess inequities regarding the deployment plan via EV-or EVSE-related performance indicators and the widely applied equity index that is the Gini index [21] in the four equity types: a fair share of resources and external costs that are grouped into horizontal equity, as well as inclusivity and affordability that referred to as vertical equity [20].Finally, inequitable properties across socio-demographic groups in the four equity types are investigated, to help planners understand the potential negative impacts of candidate EVSE deployment decisions and provide suggestions for improvement.Contributions of this study are outlined below.
• In this study, an equity assessment framework is proposed with our existing EV mobility and charging need analyses to inform the EV-user-centric EVSE deployment decision-making to consider the balance between fairness and utilitarianism.• The four-dimensional equity indicators (two horizontally and two vertically) and the comparison across diverse population groups comprehensively quantify how EVSE deployment decisions affect residents' social well-beings.• The proposed equity metric system is demonstrated through a large-scale transportation system in LA County of California.It can guide planners to pursue equitable EVSE deployment at the planning stage, maximizing the execution effect of suggested improvements given the results in equity.The rest of the paper is organized as follows: Section II presents literature reviews on equity evaluation indices, equity types, and EV-and EVSE-related performance indicators.Section III introduces the framework and components of the equity metric system.Section IV describes the specification and the assumption of the case study.Sections V and VI discuss the results of the case study and propose the corresponding improvements of equity.The final section draws conclusions and the extension of research interests.

II. LITERATURE REVIEW A. EQUITY EVALUATION INDICES
Equity evaluation on the distribution of benefits and impacts of transportation is necessary for determining the target of investment and estimating the effects of projects or policies [19], [20], [21], [22].To quantify how social equity is achieved among different socio-demographic groups, several indices were investigated in the previous studies.Reference [21] found that the Gini index was the most popular index for equity evaluation due to its ease of interpretability and communication.Meanwhile, [23] revealed that the Theil index was common for equity evaluation because of its ability to describe the equity both within and across socio-demographic groups.For another study, [24] investigated the spatial equality in taxi-hailing service in China via the Gini index and the Theil index.Reference [25] estimated the change of income, disparity, and inequality caused by the development of high-speed railway by using the Theil index.Besides these two indices, [26] presented mean log deviation, relative mean deviation, coefficient of variation, and Atkinson index to investigate the spatial equity improvement in accessibility by road capacity enhancement.Reference [27] also measured the equity of transportation investment in China via indices including Theil index, mean log deviation, Hoover index, and Atkinson index and found that equity performances under the same investment were different among the four indices.Based on the literature review of historical social equity theories including above six equity indices and their application, [28] recommended that planners have the flexibility to choose either one or multiple indices for equity evaluation in transportation network design problems depending on specific conditions and needs.

B. EQUITY TYPE IN TRANSPORTATION
In the transportation field, multiple types of equity exist and multi-dimensional understanding of equity provides a comprehensive insight for the decision-making on transportation policies and projects.Reference [20] defined social equity as allocation of benefits and cost and classified equity in transportation into two categories: horizontal and vertical equity.Horizontal equity means that people with similar socio-demographic background should equally share resources and is represented by two types of equity: a fair share of resources and external costs.A fair share of resources addresses whether public resources such as transportation budget are equally distributed to people and external costs concern the benefits or risks that people provide other individuals.Vertical equity means that people with travel restrictions should be treated well and is comprised of two types of equity: inclusivity and affordability.Inclusivity discusses how people with disabilities and special mobility needs are served in transportation and affordability addresses how low-income people can afford to receive transportation services.Table 1 gives a summary of classification and definition of each equity type that this study's equity metric system references.

C. EV-AND EVSE-RELATED PERFORMANCE INDICATORS
For the equity evaluation in transportation electrification, quantitative indicators to show the performance of EV and EVSE are required for understanding the degree of the existing inequities in the system and to what extent those inequities can be improved.This section reviews EV-and    In the aspect of the fair share of resources, accessibility is commonly evaluated in transportation equity evaluation [21], [29].Accessibility is defined as people's ability to reach the destinations and the potential opportunity for human activities generated by transportation [28], [30].As an indicator for EV and EVSE performance, the charging accessibility defined as the number of charging stations or chargers available within a certain distance from an EV-user's residence is an important factor to assess the service level of charging facilities and evaluate the people's willingness to adopt EVs [7], [31], [32].

EVSE-related performance indicators by referring to the four equity types in
Regarding external costs, [20] evaluated the costs of noise, air pollution, crash damages, and traffic congestion per user by mode and revealed that automobile has larger external impacts than other modes.In other studies, emissions of air pollutants such as carbon monoxide (CO), nitrogen oxides (NO X ), particulate matter (PM), sulfur oxides (SO X ), and volatile organic compounds (VOCs) as well as emissions of greenhouse gas (GHG) such as carbon dioxide (CO 2 ), methane, and nitrous oxide during the electricity generation or EV trip have been often assessed [33], [34].
For inclusivity, [20] and [35] presented it via travel mode share and showed that people with travel restrictions tend to drive less and travel more by walking, taxi, and public transit than those without travel restrictions.For instance, [35] analyzed inclusivity with individuals' average travel distance and travel time by using 2017 National Household Travel Survey data and presented that people with disabilities tend to travel for shorter time and smaller distance than able-bodied people.
In affordability, [20] investigated it by the share of household spending on transport and found that the ratio is considerably high in lower income groups.For the evaluation of affordability in transportation electrification, the energy consumption cost is an important factor to evaluate the performance of EV travels and determine the location or the number of EVSE.Reference [36] calculated the daily power consumption cost of EV users to propose a strategy of orderly charging EVs to shave the peak and fill the valley of a power grid operation in a day.
While previous studies have introduced various indicators for equity evaluation, none of them have addressed equity issues specifically related to EV and EVSE performance across all four equity classes.This study aims to fill this gap by presenting a multifaceted equity metric system that incorporates specific performance indicators for each equity type, allowing for a comprehensive evaluation of equity in transportation electrification.

III. METHODOLOGY A. OVERVIEW OF THE EQUITY METRIC SYSTEM
The equity metric system proposed in this study follows the equity type classification by [20] to enable a multifaceted equity evaluation from both horizontal and vertical aspects in transportation electrification.Table 2 summarizes the scope of the equity metric system by describing the four equity types, the potential equity performance indicators addressed in the literature review section, and EV-and EVSE-related performance indicators.Note that considering the data limitation and the ability of each indicator for quantification and statistical analysis, indicators shown in the third column of Table 2 are the ones adopted by the equity metric system of this study, which are explained in depth with the following paragraphs.
In terms of horizontal equity and the fair allocation of resources, [7] and [32] found that the distribution of public chargers significantly influences people's willingness to adopt EVs.A higher density of public chargers reduces individuals' concerns about running out of energy while traveling.Therefore, the number of public chargers in a specific area, known as public charger density, serves as a crucial indicator in this study for assessing the fairness of resource allocation for EV operation.
External costs, measuring the impact of travel activities on others, has been extensively discussed in previous studies, particularly in relation to environmental concerns, as outlined in the literature review.Reference [37] also emphasized to consider safety as a critical factor in transportation cost and benefit analysis.Therefore, external crash cost per person and the cost reduction associated with air pollution and GHG emissions were chosen as indicators for assessing horizontal equity in this study.
As a type of vertical equity, inclusivity focuses on ensuring the participation of individuals with disabilities or travel restrictions in transportation activities.Reference [35] exanimated and determined that the average travel distance and travel time for individuals were significant variables for quantifying inclusivity among individuals.Since the average travel distance will be thoroughly discussed in the external costs, this study adopts the average EV travel time as an evaluation indicator of inclusivity.
Regarding affordability, [20] emphasized the importance of considering user costs for different travel modes as they directly impact travelers' mode choices.In the context of EVs, the cost of public charging plays a substantial role in daily travel expenses.Therefore, this study considers the cost burden of public charging to each individual's income as a vital indicator for assessing affordability.Fig. 1 shows the overall research flow of the equity evaluation by the developed equity metric system, indicating in which steps the outputs, indicators, and variables are processed.First, the EVSE deployment decision-making model in our previous study are employed in the equity metric system to generate people's travel trajectories, public charging demand profile, and public charger deployment plan, see the introduction in Section III-B.Second, each EV-and EVSE-related performance indicator associated with the four equity classes is calculated based on the outputs in the first step, and the outcomes are compared concerning people's socio-demographic background to statistically show the distribution of EV-and EVSE-related impacts and its disparity across different socio-demographic statuses, which are carefully delineated in Section III-C.Finally, proposals for the improvements of equity are drawn based on the results of the equity evaluation.

B. EVSE DEPLOYMENT DECISION-MAKING MODEL
The EVSE deployment decision-making model developed by our previous study derives the distribution plan of public chargers for future EV fleets at the system and zone/community levels [15].This model is closely dependent on the understanding of people's daily travel trajectories and the spatially varying EV adoption in the system, the learning of EV attributes and public charging behaviors, and the exploration of possible future electrification levels.Fig. 2 shows this decision-making model with three critical modules, including travel profile simulation module, charging demand profile generation module, and charger deployment planning module.The ultimate outcome of the module is in Output 3 as the distribution plan of public chargers.The core modules are briefly introduced as follows.
The travel profile simulation module aims at simulating the daily travel trajectory baseline for people via multiple vehicle modes without application of charging events.This module employs the activity-based travel demand and agentbased traffic assignment models to capture the stochasticity of every individual's daily travel.First, the used activitybased traffic demand model is a regional transportation plan model to forecast travel needs associated with the survey of everyone's daily activities [38], [39].It gives people's travel plans based on their activity schedules.Next, the agent-based transportation simulation model assigns people's travel plans by simulating the movement of people and vehicles in the network [40], [41], [42].Finally, Output 1 shows the daily trip chains of the population.
The charging demand profile generation module learns the charging activities by i) understanding the possible future electrification levels such as the projected fleet size of EVs including battery electric vehicles (BEVs) and plug-in hybrid electric vehicles (PHEVs) as well as the future installation scale of private residential and workplace chargers, ii) defining comprehensive and explicit charging strategies to allow the diverse real-life charging habits of EV users, and iii) considering the significant impacts of traffic conditions (e.g., traffic speeds and stops) on vehicles' energy consumption and the time-sensitive charging price on charger location selection [43], [44], [45], [46].Then, this module applies the future electrification configurations, EV user habits, and driving and charging characteristics to the trip chain table in Output 1.Output 2 is the prediction result to list the relevant individual, vehicle, location, and time of each charging event in tabular format.
The charger deployment planning module aggregates the daily public charging activities based on the charging event prediction table in Output 2 to obtain the peak charging demand at the system and zone/community levels and the relative temporal attributes.Next, this module considers the shareability of public chargers to transform the spatiotemporal public charging need into the public charger deployment of the system by allowing multiple charging sessions on each charger throughout the day in zone/community.The zone/community defined in this study is census tract, commonly used to describe the socio-demographic area.Finally, Output 3 is the EVSE deployment plan to provide the census-tract-level public charger numbers for the electrified transportation system.This employed EVSE deployment decision-making model can predict and meet zone/community-and system-level charging needs by considering individuals from various socio-demographic backgrounds within the large-scale electrified transportation system.The model predicts the charging needs from the individual travel trajectories; simultaneously, the model fulfills potential charging demands by either assigning dedicated chargers or enabling users to share chargers at different times of the day.

Socio-demographic Group Classification
For statistical analyses in the equity evaluation, the census tracts (CTs) are grouped into three sets of bins based on socio-demographic variables.
where N mfhu,i refers to the number of MHFU units in the ith CT and N i refers to the total number of units in the ith CT.
• Zonal major race-ethnicity.Five racial-ethnic groups are considered: non-Hispanic Asian, non-Hispanic Black, non-Hispanic White, Hispanic, and no majority.CTs are grouped into these five groups based on the major race-ethnicity, which is defined as the raceethnicity that accounts for more than 50% of the CT's population.

EV-and EVSE-related Performance Indicators:
This study focuses on how the transportation electrification will serve people and communities, and thus five EV-and EVSE-related performance indicators are used for evaluating equity (as mentioned in Section III-A).
• A fair share of resources indicator -Public charger density (PCD).It is defined as the average number of public chargers per person of a socio-demographic group as in (2).

• External costs indicator -External crash cost per person (ECC).
It is given as the crash cost including medical care, lost wages and future earnings, and property damage or loss imposed on another person of a socio-demographic group as in (3) [37].

• External costs indicator -Reduction in air pollution and GHG emission cost (CR).
It is presented as the reduction in the following environmental cost per person in a socio-demographic group as in (4).Air pollutants are CO, NO X , PM, SO X , and VOCs from tailpipe emissions and GHGs are CO 2 , methane, and nitrous oxide.GHGs are quantified as lifecycle emissions which include emissions from fuel production, car manufacture and maintenance, as well as road infrastructure construction and maintenance [37].The cost represents the control cost that is required for the reduction of emissions, defense against risks, carbon retention, and compensating individuals affected by the impacts [49].
Since the dataset used in this study focuses on EV trips, the cost reduction is calculated by comparing the air pollution and GHG emission cost of the EV fleet and that of comparable non-EVs.

• Inclusivity indicator -Average EV travel time (ATT):
It is provided as the travel time of an EV in each trip in a socio-demographic group as in (5).

• Affordability indicator -Charging cost burden (CCB):
It is defined as the share of daily public charging cost against the median daily household income of a person in a socio-demographic group as in (6).
where i is the ith socio-demographic group; N i is the number of CTs in group i; PC j is the number of public chargers in the jth CT of group i; and P j is the population of CT j in group i. ECC ev is the external crash cost of EV that is $0.081 per vehicle miles traveled (VMT) [37]; and VMT j is the total daily VMT of EVs owned in the jth CT of group i. C pol gas is the air pollution cost of non-EV as $0.059 per VMT; C pol ev is the air pollution cost of EV as $0.015 per VMT; C ghg gas is the GHG emission cost of non-EV as $0.025 per VMT; and C ghg ev is the GHG emission cost of EV as $0.006 per VMT [37].TT j is the total daily travel time of EVs owned in the jth CT of group i; and n j is the total trip number of EVs owned in the jth CT of group i. ADCC j is the average daily public charging cost per EV user living in the jth CT of group i; and MDHI j is the median daily household income of the jth CT of group i.All costs provided are referenced at [50] and are as of February 2023.Since it is difficult to estimate future inflation rates and obtain the data to speculate transition of money value, this study adopts the value of 2023 and assumes that it will remain the same.Equity Evaluation Index: This study adopts the Gini index in (7) to identify the disparity in the performance indicators across different socio-demographic groups due to its wide implementation on equity evaluation and distributions of accessibility [15], [21], [26], [28], [51].The Gini index is determined among 0 (a completely equal distribution) and 1 (the most unequal distribution), indicating that the larger the Gini index, the more unequal distribution.Reference [26] states that the Gini index is sensitive to small change of values of an equity performance indicator, as well as [21] argues that the Theil index and Atkinson index are not suitable for policy-makings on equity because of their low sensitivity.Since this study investigates the difference in EV and EVSE performance across diverse socio-demographic groups, the Gini index is considered to be the most appropriate index for the equity evaluation among several indices.
where n is the number of socio-demographic groups; X i is the value of the selected equity performance indicator for group i; and X is the mean of all X i .Control Variable: Control variable helps provide deep insights on the statistical distribution of performance indicators and their disparities across socio-demographic groups in equity evaluation.This study treats income as the control variable and divides CTs into four income groups via the income ratio (IR) given by ( 8) [52]: [0%, 60%] for the lowest-income group, (60%, 80%] for the low-income group, (80%, 140%] for the middle-income group, and greater than 140% for the high-income group. where MAHI i is the median annual household income of the ith CT; and AMI is the area median income of the region which the CT belongs to.For example, in the case study of this paper, the area median income of LA County in 2022 ($91,100) is used [53].

IV. CASE STUDY AND DATA SPECIFICATION A. SOCIO-DEMOGRAPHIC DATA SPECIFICATION AND ASSUMPTION
California is attempting to achieve the goal of the executive order by governors that requires all new passenger cars sold in California to be zero-emission vehicles by 2035 [54].
Given this, our previous study projected the public charging demand profile via EVSE deployment decision-making model in Section III-B for LA County in 2035 [15].Given this projection, this paper carries out an equity evaluation in a large-scale electrified transportation system in LA County in 2035.The transportation system includes 2,342 CTs, 38% of which are DACs, all primary and secondary roads, more than 10 million population, and over six million vehicles with 1.23% market share of EVs in 2022 [15].The resident income data is obtained from [55].To simplify the analysis, this study assumes that people's travel behavior, median household income, MFHU rate, and race-ethnicity share of each CT will stay the same from 2022 to 2035.Fig. 3 shows CTs with classification of DAC and non-DAC with road network, and with a sample of the projected public charger distribution of 2035 LA County.

B. ELECTRIFICATION LEVEL, AND CHARGING CONDITION
The case study is based on the transportation system in a typical traffic-demand day.To incorporate the influence of varying electrification statuses in the future, two electrification levels in 2035 shown in Table 3 are adopted based on the possible futures discussed in the Los Angeles 100% Renewable Energy Study: the high electrification level and the moderate electrification level [43].The total EV numbers described in Table 3 are estimated based on the assumption that they will linearly increase with the EV market share of 2022 across all CTs.Other assumptions on public chargers and battery charging conditions are outlined in the following bullets and refer to our previous study which has been well developed and discussed [15].The public charger type assumed in this study is Level 2 (L2) charger since it is common for EV charging and has a large share (approximately 90%-95%) in the public charger number in the U.S. [45], [56].Furthermore, our previous study simulated a large variance in charging needs in a day by choosing different initial SOC distributions since the initial SOC of an EV differs depending on the charging frequency of the EV user [45].Reference [45] showed that the average charging frequency of BEV and PHEV are 1.5 and 1.2 days respectively, which indicates that an EV is recharged a day and a half after the last charge.Thus, this study assumes two EV (BEV and PHEV) initial SOC states: • The low-charging-demand day where a given EV starts its trip with a healthy and high SOC status (80%) on the first day after charging.
• The high-charging-demand day where a given EV begins its travel with half of the healthy SOC status (40%) on the second day after the energy consumption on the first day without any public recharging.
To sum up, EV users having home chargers are deemed to always begin their trip with 80% of the full SOC due to the convenience of home charging.80% of the full SOC is the maximum healthy state for battery without operation degradation [57], [58], [59], [60], [61], [62].For EV residents without home chargers, two SOC levels are considered for their initial trips of the day.A value of 80% is assigned for the low-charging-demand day, representing the healthy initial SOC level that allows for day-long trips after the last charge.Alternatively, a value of 40% is assigned for the highcharging-demand day, approximately reflecting the average energy consumption of the first day, as the high-chargingdemand day occurs on the second day of trips after the last charge.40% SOC is in the range of 20% -80% SOC that is a range with which EV batteries can be operated without capacity degradation and accidents such as explosion, structural degradation, and dissolution of active materials [57], [58], [59], [60], [61], [62].

V. RESULTS AND DISCUSSION
All equity evaluations were performed for the four scenarios including the high-and low-charging-demand days in the high and moderate electrification levels in 2035.However, since the electrification level primarily determines future transportation electrification, this study principally compares the results of the two electrification levels.Besides, the general trends between the high-and low-charging-demand days in the results of the fair share of resources and affordability are similar.While, the trends of the high-and low-chargingdemand days are the same for external costs and inclusivity since whether there are charging events in people's daily travel does not impact their EV travel distance and time, which induces no difference in external costs and inclusivity.Therefore, most of the discussions in this section are focused on the results of the high-charging-demand day, which are similar with or the same as the results of the low-charging-demand day.

A. FUNDAMENTAL STATISTICAL DATA
This section describes all the fundamental statistical data for each socio-demographic group in LA County to support the discussion of the results of each equity type.Fig. 4 provides the percentage of CT numbers and Fig. 5 shows the home and workplace charger adoption scale for each socio-demographic group against income level in the high and moderate electrification levels in 2035.Since DACs are CTs rare in the high-income group and there are no CTs for non-Hispanic Black population in the high-income group as shown in Fig. 4 (a) and (c), DACs and non-Hispanic Black population don't have values in the high-income group for all equity performance indicators.Table 4 provides the number of public chargers in each socio-demographic group for the high-charging-demand day in the high and moderate electrification levels in 2035 and Table 5 describes EV number per capita in each socio-demographic group in the high and moderate electrification levels in 2035.Table 6 shows the average total daily EV travel distance per CT in the high and moderate electrification levels in 2035.Table 7 describes the average population per CT in 2035 and the median annual household income in each socio-demographic group.Table 8 provides the trip number share and the average travel time by purpose of all EV trips.

B. RESULTS OF EQUITY EVALUATION IN A FAIR SHARE OF RESOURCES
Fig. 6 and 7 show the public charger density that is the performance indicator of the fair share of resources.As a key finding, DACs, MFHU rate < 25%, and non-Hispanic    White and Hispanic population have higher public charger density than other socio-demographic groups in general.That is because DACs, MFHU rate < 25%, and Hispanic population have lower home and workplace charger density than other groups in Fig. 5, and non-Hispanic White population have the large EV number per capita in Table 5.

FIGURE 6. Public charger density for the high-charging-demand day by DAC/non-DAC (a), MFHU rate (b), and race-ethnicity (c).
Meanwhile, non-Hispanic Asian population generally have the low values due to the relatively high home and workplace charger adoption scale.Non-Hispanic Black population have high values in the lowest-and middle-income groups, but since the percentage of their CTs is small across all income groups in Fig 4 (c), the results of the public charger density are subject to the extremely large values and might be unreliable.
Fig. 8 provides the disparities in public charger density across different socio-demographic groups via the Gini index.Major findings are that the disparity is large between DACs and non-DACs in the middle-income group for the high electrification level, it is high across different MFHU rates in the high-income group for both electrification levels, and it is high among different race-ethnicity especially in the lowest-and middle-income groups.The reasons are that non-DACs have a big need in the public charger in the middle-income group as in Table 4, the gap in the public charger density among different MFHU rates is large in the high-income group in Fig. 6 (b) and Fig. 7 (b), and the gap in the public charger number across different race-ethnicity is big in Table 4. Since there are no values of the public charger density for DACs and non-Hispanic Black population in the high-income group, the Gini index becomes large in this income group in Fig. 8 (a) and (c).

C. RESULTS OF EQUITY EVALUATION IN EXTERNAL COSTS
This section involves the outputs of the equity evaluation in external crash cost and air pollution and GHG emission cost reduction that are the performance indicators of external costs.
The results of daily external crash cost and air pollution and GHG emission cost reduction are presented with Fig. 9 and 10 respectively.For both results, major implications are that DACs have smaller values than non-DACs, while MFHU < 25% and non-Hispanic and Hispanic population keep low positions.The results can be explained that these socio-demographic groups have shorter EV travel distance than other groups as in Table 6, which makes potential crash risk and environmental benefit smaller.
Fig. 11 provides the disparities in the values of external costs across different socio-demographic groups.Since the overall trends and values of the disparities were significantly similar between the external crash cost and the air pollution and GHG emission cost reduction, Fig. 11 represents the results of both indicators.For both outputs, key findings are that the disparity is relatively high between DACs and non-DACs in the middle-income group, it is large across different   MFHU rates in the lowest-and low-income groups, and it keeps a relatively high level around 0.2 to 0.3 across different race-ethnicity for both electrification levels.The possible reasons are that DACs and MFHU rate < 25% have shorter EV travel distance than other groups while they have the similarscale population with other groups as in Table 6 and 7, which makes the gap in the external costs large between DACs and non-DACs, and across different MFHU rates.Also, the gap in the travel distance is large among different race-ethnicity in Table 6, making the disparity of their external costs big.There is a rise of the Gini value in the high-income group in Fig. 11 (a) and (c), which is caused by the absence of the external cost values in DACs and non-Hispanic Black population.statistics, it can be said that the trips whose purpose is home or work, which account for approximately 50% of all EV trips in number majorly impact the general trend of average EV travel time in each result.For more details, non-DACs have a slightly high share of trips heading to home (35.4% for non-DACs and 35.3% for DACs) and work (13.0% for non-DACs and 11.9% for DACs).Also, MFHU rate > 75% has a little high share of trips heading to home (35.6% for MFHU rate > 75% and 35.4% in average for other rates) and work (13.8% for MFHU rate > 75% and 12.6% in average for other rates).Besides, non-Hispanic Black and White population have rather large shares of trips heading to home (35.4% for non-Hispanic Black, 35.6% for non-Hispanic White, and 35.1% in average for others) and work (15.7% for non-Hispanic Black, 14.0% for non-Hispanic White, and 11.4% in average for others).

D. RESULTS OF EQUITY EVALUATION IN INCLUSIVITY
The disparities in the average EV travel time are provided in Fig. 13.Key implications are that there are not significant disparities between DACs and non-DACs and across different MFHU rates, while the disparity increases with income across different race-ethnicity.The reason is that the Gini index is calculated based on the values of the average EV travel time in Fig. 12, which indicates that the gap of the values across different race-ethnicity increases with income.In Fig. 13 (a) and (c) the Gini index rises in the high-income group due to no values of travel time for DACs and non-Hispanic Black in this income group.

E. RESULTS OF EQUITY EVALUATION IN AFFORDABILITY
This section summarizes the results of charging cost burden per EV user via Fig. 14 and 15 that is the performance indicator of affordability.Fundamental implications are that non-DACs, MFHU rate > 75%, and non-Hispanic White population have higher values than other socio-demographic groups.That is because the travel distance is large in these groups as in Table 6, which induces high energy consumption and thus high charging demand, while the median household income is similar with other groups at the same income level as in Table 7.Although non-Hispanic Black population have the highest values in the lowest income group, their results might be impacted from supreme values and thus not be reliable because they have the small share of CTs as shown in Fig. 4 (c).Fig. 14 and 15 also show that the charging cost burden decreases with the growth of income, which indicates that people with high income can afford to often charge EVs.
The disparities in the charging cost burden across different socio-demographic groups are shown in Fig. 16.As major findings, there are almost no disparities between DACs and non-DACs, the disparity across different MFHU rate  is relatively high in the low-income group for the moderate electrification level, and the disparity across different race-ethnicity is large in the lowest-income group for the moderate electrification level.The possible reason for the disparity across different MFHU rate is the extremely long travel distance of MFHU rate > 75% against similar annual household income among all MHFU rates in Table 6 and 7, which makes the gap across different MFHU rates large.The cause of the disparity among race-ethnicity is the large difference of charging cost burden in Fig. 14 (c) due to the varying values of travel distance under every income level in Table 6.Fig. 16 (a) and (c) show that the Gini index becomes significantly high in the high-income group, which is because DACs and non-Hispanic Black population have no values of charging cost burden in this income group.

VI. PROPOSAL FOR THE IMPROVEMENTS OF INEQUITIES
From the results of the multifaceted equity evaluation in the previous section, it was found that disparities exist in the transition of transportation electrification through EV and EVSE across different socio-demographic groups.The disparities were quantitatively revealed by the Gini index ranging from 0.02 to 0.31 in the public charger density, 0.05 to 0.27 in the external crash cost and air pollution and GHG emissions cost reduction, 0.01 to 0.05 in the average EV travel time, and 0.002 to 0.10 in the public charging cost burden except for the high values in the high-income group due to the absence of performance indicator values in DACs and non-Hispanic Black population.As described above, the larger the Gini value (closer to 1), the greater the disparity among equity performance indicators.The Gini index calculated in this study was low in general relative to the results of existing studies [24], [51], which indicates that our existing study of EV and EVSE deployment projection takes equity into account to some extent.While, the results also revealed that the disparities were larger in the fair share of resources and external costs grouped into horizontal equity than in inclusivity and affordability classified as vertical equity.In other words, the inadequacy of horizontal equity should be emphasized in transportation electrification.Based on these results, the following improvements are proposed to achieve equity in the future transportation electrification.

A. ENHANCEMENT OF CHARGER ADOPTION
People's willingness to adopt EVs is closely dependent on public charger distribution because high charger density provides a more reliable charging environment for people.From the results of the equity evaluation in the fair share of resources, disparities in public charger density are obvious across different socio-demographic groups.To achieve this kind of equity, it is important to deploy more public chargers in the lowest-and low-income groups, non-DACs, and non-Hispanic Asian groups, by prioritizing or increasing the budget to allocate EVSE.However, since DACs, MHFU rate < 25%, and Hispanic population have the low adoption scale of home and workplace chargers, it is also necessary to enhance the home and workplace charger adoption level in a supportive way such as subsidies for EV purchase.

B. IMPROVEMENT OF EV ADOPTION
The results of equity evaluation in the fair share of resources also revealed the disparity in the adopted EV number per capita between the two electrification levels and across different socio-demographic groups as in Table 5.This implies an improvement space to develop the EV fleet in DACs, MFHU rate < 25%, and Black and Hispanic.The improvement will reduce the gap in EV adoption level and thus result in the acceleration of transportation electrification.The possible methods are development of more public chargers, which encourages people to adopt EVs by reducing their anxiety about charging opportunity, subsidies for EV purchase, and discussion with or subsidies for manufacturers to promote EVs.

C. INCREASE OF TRAVEL BY EV
According to the results of equity evaluation in the external costs, the environmental benefit increases with income because people with high income prefer to travel for longer distance.The results also show that the environmental benefit is relatively low for MFHU rate < 25%, non-Hispanic Black, and Hispanic population due to the relatively short travel distance.To improve equity in the environmental benefit, subsidies for EV purchase are recommended and more public, home, and workplace chargers should be installed in communities with higher proportions of lower income groups, MFHU < 25%, non-Hispanic Black, and Hispanic population since a reliable charger accessibility will reduce EV-users' anxiety about losing energy and thus encourage them to adopt EVs in the aforementioned ways [7], [34], [45].
Moreover, although the external crash cost of EV and non-EV introduced by [37] are the same, [63] reported that EVs had considerably lower insurance claim frequency than conventional counterparts, which indicates that potential crash risk of EV was lower than that of non-EV.Therefore, the transition to EVs in all groups may lead to the reduction of total external crash cost by reducing the potential crashes by non-EVs.

D. REDUCTION OF PUBLIC CHARGING COST BURDEN
From the results of equity evaluation in affordability, the public charging cost burden differs by income and sociodemographic status due to the gap in EV travel distance and household income.To achieve the equity in affordability, the reduction of public charging cost burden is important, particularly for the lower income groups, non-DACs, MFHU rate > 75%, and non-Hispanic White communities, whose EV travel distance is significantly large.The possible approaches can be providing subsidies for EV purchase; developing more energy-efficient EVs by manufacturers; and increasing the share of renewable energy such as bioenergy, geothermal energy, hydrogen, hydropower, marine energy, solar energy, and wind energy that is expected to reduce the electricity generation cost compared with traditional energy [64].However, since most of the results of the equity evaluation indicate that the lower income groups, DACs, MFHU rate < 25%, and non-Hispanic Asian, non-Hispanic Black, and Hispanic population are underserved as discussed above, the efforts to reduce the public charging fee for these underprivileged groups will also be effective and recommended for acquiring equity.

VII. CONCLUSION
This paper performed an equity metric system with regard to the EV and EVSE deployment for the transportation electrification across different socio-demographic groups with the four equity aspects that are the fair share of resources, external costs, inclusivity, and affordability.A case study of equity evaluation in 2035 LA County is implemented with the developed equity metric system.The case study described the disparity in the EV-and EVSE-related performance across people by using Gini index.
The case study results present some equity issues: (1) The public charger density shows a general increasing trend with income while it varies across socio-demographic groups depending on the total EV and public charger adoption scale; (2) The external crash cost and the reduction in air pollution and GHG emission cost increase with income status due to the growing EV travel distance; (3) The average EV travel time also grows with income because of the growing share of trips heading to home and work while the variance can be observed across different socio-demographic groups due to the difference of trip purpose; and (4) The public charging cost burden decreases with the growth of income and varies among people with different socio-demographic background depending on the EV travel distance.Based on these results, the improvements to enhance the equity are proposed: i) enhancement of public, home, and workplace charger adoption; ii) improvement of EV adoption; iii) increase of travel by EV; and iv) reduction of public charging cost burden.These improvements can be implemented by developing policies of subsidies for EV purchase, prioritization of budget, further promotion of EVs, adoption of renewable energy, and increase of job opportunities in the currently underserved communities.The proposed equity metric system can be implemented by local agencies in conjunction with the EVSE deployment decision-making model from the team's previous study [15] and Equations ( 1)-( 7) whenever they make decisions or plans related to EV adoption and the development of public chargers.The EVSE deployment decision-making model generates outputs that encompass individual travel trajectories, charging behaviors, and the distribution of public chargers within the system.This achievement is facilitated by utilizing comprehensive data, including socio-geographical characteristics of the synthetic population, individual travel demands, EV market share, battery parameters, EV energy consumption patterns, and more.Subsequently, the equity metric system in this study aggregates the high-fidelity outputs of travel and charging behaviors from the previous model by transitioning them from the individual level to the zone/community level, while considering attributes of communities such as income levels, race-ethnicity, and multifamily housing unit (MFHU) statuses.Through this process, insights into inequities among communities during the electrification of transportation are acquired and quantified.These insights serve as the foundation for targeted recommendations to address specific disparities.
However, this study also has limitations.One is the use of the linear projection of EV adoption for future.Our previous study predicted the number of EVs based on the EV population in 2022, but [7] and [32] pointed out that people's intention to purchase EVs is dependent on charger accessibility, which indicates that the more public, home, and workplace chargers, the more EVs adopted by people.A more enhanced method can be used to predict the number of EVs considering the increase in the charger density in further study.Another limitation is related to the indicator of the environmental benefit of EVs.This study addresses the reduction in air pollution cost that derives from tailpipe emissions and lifecycle GHG emission cost based on the assumption that the air pollution cost from upstream emissions such as electricity generation and fuel production is significantly large and hard to quantify [37].However, [65] shows that the air pollution and GHG emissions including upstream emissions of non-EVs is larger than that of EVs.A more sophisticated equity metric system will be able to estimate the lifecycle environmental benefit of EVs in the future study.In addition, one of the challenges of the proposed equity metric system is that it adopts one equity evaluation index (the Gini index) while [28] suggested that policy-makers should use one or several indices to evaluate equity issues in transportation planning based on their particular conditions as explained in the literature review.Adopting other indices would not only provide us deeper insights for the equity problems in the deployment of EV and EVSE but also enhance the applicability of the proposed system to diverse agencies with specific needs.
TAKAHIRO TSUKIJI received the B.S. and first M.S. degrees in civil engineering from Kyoto University, Japan, and the second M.S. degree with the Civil and Environmental Engineering Department, University of California at Los Angeles, Los Angeles.After receiving the first M.S. degree, he started working with Ministry of Land, Infrastructure, Transport and Tourism (MLIT), Japan.He has engaged in the research of intelligent transportation systems, the policy-making of bridge maintenance, and the planning and design of national highways at MLIT.NING ZHANG is a Postdoctoral Fellow with the Samueli School of Engineering, University of California at Los Angeles, Los Angeles.She has expertise in transportation electrification, risk assessment and mitigation, and resilience-oriented optimization of the transportation network under operational and weather risk conditions.She is currently serving as an Associate Member of the American Society of Civil Engineers (ASCE) T&DI Sustainable Transportation Committee and a younger member of the ASCE T&DI Active Transportation Committee.
QINHUA JIANG received the B.S. and M.S. degrees in civil engineering from Beijing Jiaotong University.He is currently pursuing the Ph.D. degree from the Civil and Environmental Engineering Department, University of California at Los Angeles, Los Angeles.His areas of expertise include microscopic traffic simulation, activity-based travel demand modeling, AI/machine learning based traffic forecasting, and large-scale intelligent transportation system analysis.
BRIAN YUESHUAI HE is a Project Scientist with the Samueli School of Engineering, University of California at Los Angeles.His research interests include transportation system modeling, travel behavior analysis, and transportation planning.He has led and managed many research projects funded by federal/state/local programs in areas of multi-scale transportation system modeling, urban mobility simulation, machine learning and artificial intelligence, and computational simulation.He

FIGURE 2 .
FIGURE 2. Research framework of the EVSE deployment decision-making model.

FIGURE 3 .
FIGURE 3. Map of LA County with roads (a) and public charger distribution (b).

Fig. 12
Fig. 12 provides the average EV travel time that is the performance indicator of inclusivity.The results reveal that non-DACs, MFHU rate > 75%, and non-Hispanic Black and White populations have high values.The reasons can be explained as follows.As shown in Table 8, home, work, maintenance, shop, and escort have relatively large share of trip number among all trip purposes, while university, home, and work have the top three long travel time.Based on these

TABLE 2 . Scope of the equity metric system. FIGURE 1. Research flow of the equity metric system.
[48]i-Family-Housing-Unit rate (MFHU rate).MFHU rate is given by (1)[15],[48].In CTs with high MFHU rate, many people need to share public resources and spaces.For classification of CTs, the MFHU rate • Disadvantaged/non-disadvantaged communities (DACs/ non-DACs).CTs are DACs in this study (particularly in the case study) if they meet the criteria of Senate Bill (SB) 535 Disadvantaged Communities, which designates DACs as communities most vulnerable to environmental pollution and facing economic challenges [47].•

TABLE 8 . Trip number share and average trip time by purpose.
is a member of the TRB Sub-Committee on Travel Time, Speed, and Reliability and the International Association for Transportation Behavior Research.JIAQI MA (Member, IEEE) is an Associate Professor with the Samueli School of Engineering, University of California at Los Angeles.He has led and managed many research projects funded by U.S. DOT, NSF, state DOTs, and other federal/state/local programs covering areas of smart transportation systems, such as vehicle-highway automation, intelligent transportation systems, connected vehicles, shared mobility, and large-scale smart system modeling and simulation, and artificial intelligence and advanced computing applications in transportation.He is an Associate Editor of the IEEE OPEN JOURNAL OF INTELLIGENT TRANSPORTATION SYSTEMS.He is a member of the TRB Standing Committee on Vehicle-Highway Automation, the TRB Standing Committee on Artificial Intelligence and Advanced Computing Applications, and the American Society of Civil Engineers Connected and Autonomous Vehicles Impacts Committee, and the Co-Chair of the IEEE ITS Society Technical Committee on Smart Mobility and Transportation 5.0.