A New Approach to Understanding the Impact of Automobile Ownership on Transportation Equity

Abstract

Transportation affordability, the ability of a household to comfortably bear necessary transportation expenses, is a pressing challenge to the development of sustainable and equitable places. Transportation planners have historically identified transit user groups as choice riders, those with access to other modes or the ability to purchase access, and captive riders, transit-dependent populations who must use transit regardless of service quality. However, this characterization is incomplete, disregarding built-environment pressures that compel a much larger population of households to own and use vehicles regardless of their ability to afford the very high price of vehicle ownership. We propose a new theoretical construct, illuminated using the Bureau of Labor Statistics Consumer Expenditure Survey Public Use Microdata—the transportation choice/captivity matrix—which examines transportation user groups via vehicle ownership and income variables.

Keywords

transportation equity transportation burden transportation affordability captive drivers transit dependency forced vehicle ownership

Transportation in the United States is dominated by the automobile. National prioritization of personal vehicle use has led to the degradation or outright removal of other transportation facilities that would allow people to accomplish daily travel safely and easily without a car ( 1 ). In addition to the tens of thousands of annual traffic deaths, the high toll of greenhouse gas emissions, exacerbation of social isolation, and many other external costs of automobile-prioritization, direct household spending on transportation presents a steep burden for many ( 2 – 5 ). Rising in step with vehicle ownership rates, the average cost of getting around has climbed from less than 5% of total household spending at the beginning of the 20th century to nearly 20% in the present day ( 6 , 7 ).

Though a substantial and rapidly growing cost to households, transportation spending rarely features prominently in public discourse beyond excitement surrounding tech novelties, like ridesharing, or outrage when gas prices increase as a result of exogenous factors. Outside of these anomalous spending events, the very high fixed costs of transportation, particularly transportation that requires vehicle use, are not presented as a pressing issue in need of immediate intervention via existing, actionable solutions. King et al. draw attention to the problem faced by households who must pay the transportation costs associated with vehicle ownership or risk the potential cost of lost economic opportunity presented by reduced mobility ( 6 ). The burden of high transportation costs in the United States, a country of extreme income inequality, is heaviest on those with limited financial resources, presenting a problem of transportation equity ( 8 ). Equitable transit service delivery has not historically been prioritized over service that is perceived to increase ridership ( 9 ). Instead, the characterization of “captive” and “choice” users has deprioritized the needs of carless transit users, known as captive riders, who must use transit regardless of the quality of service, while not considering the much larger parallel cohort of vehicle users referred to by Schmitt as captive drivers, that is, those who own a vehicle regardless of their ability to comfortably afford its use, and choice drivers, those who own a vehicle and are less likely to be financially burdened by the associated costs ( 9 , 10 ). The examination of transportation burden through the framework of captive/choice riders is limiting, treating vehicle ownership as an ideal state, and overlooking circumstances that encourage individual participation in automobility, such as poor transit service, lack of sidewalk connectivity, or sprawl and single use development patterns. Our ability to measure the experience of equity in transportation affordability is further complicated by a lack of metrics that can be easily replicated by transportation practitioners at a local scale. We propose broadening the discussion of different classes of transportation users by introducing a matrix that examines transit user groups together with choice and captive drivers.

We examined the literature surrounding captive transit riders and forced car ownership to form a rigorous definition of transportation captivity. This study used data collected from the Bureau of Labor Statistics’ (BLS) Consumer Expenditure Survey (CE) to estimate the proportion of U.S. households who fall into each of our proposed categories of transportation users.We then analyzed state-level variables using ordinary least squares regression to examine the extent to which prevalence of the choice rider group correlated to state and local transit expenditure and to population density. We discuss the importance of this broader classification, and the impact of improved knowledge surrounding transportation diversity, defined by Litman as the availability of multiple transportation mode options, as a poverty reduction strategy ( 11 ).

Literature Review

Transportation Affordability

The experience of transportation equity, the just distribution of transportation burdens and benefits, has for the first time been identified as a strategic goal of the U.S. Department of Transportation ( 12 ). Transportation affordability is one element of this emerging subfield, in which a household is considered to be burdened by transportation costs when their total transportation expenditures exceed what they can comfortably afford, or 15% of total household spending when using a standard threshold ( 13 ). Litman estimates as many as 20% to 40% of U.S. households experience issues of transportation affordability, a situation exacerbated by automobility in a country where the average annual cost of vehicle ownership and use can range from $3,000 to $10,000 per car ( 13 – 15 ). These wide ranges reflect shifting definitions of poverty and transportation burden, fluctuations in fuel prices, and a lack of data on direct household spending. There are built-environment factors that support lower-cost transportation modes, such as increased activity density, mixed-use development, and improved street connectivity ( 16 ). However, there has been some debate about the utility of existing measures of location affordability, largely via critiques of the Center for Neighborhood Technology’s Housing + Transportation Index which question methods used to calculate transportation spending, such as the application of modeled automobile use assumptions ( 17 , 18 ).

The strategies employed by low-income households to minimize the burden of their transportation costs are varied and creative. Households have been observed limiting the number and duration of trips, keeping detailed accounting of transportation metrics, opting for initial lower-quality vehicle purchases, forgoing maintenance, relying on interpersonal and informal social networks for vehicle maintenance and ridesharing, reducing all other spending, and rarely, moving to locations that are less vehicle dependent ( 19 – 21 ). Even with such strategies, car ownership can be a temporary and tenuous status for those in poverty, who may cycle in and out of vehicle ownership with new or reduced opportunity and the changing of household composition through events such as marriage, childbirth, retirement, or death ( 22 , 23 ). Low-income households find themselves choosing between burdensome transportation costs or limited mobility, which can reinforce poverty by reducing access to higher-paying job opportunities ( 24 ). In a system with limited transportation options, households with fewer financial resources must make complex spending decisions, weighing need and immediacy against safety and continuity.

Automobility and the Built Environment

Regardless of the costly experience of vehicle ownership, transit use in the United States is very low. Only 14 of 3,143 counties report a transit commute mode-share higher than 20%, representing the metropolitan areas of New York City, Washington, D.C., San Francisco, Philadelphia, and Boston ( 25 ). The decline of transit use in the United States is frequently interpreted as a natural market shift for a car-obsessed nation ( 1 ). Researchers have questioned this narrative, instead proposing that the mid-century removal of fixed transit lines in favor of buses, paired with the development of highly prescriptive and vehicle-oriented street design and segregated zoning standards have created a transportation ecosystem uniquely devoid of diversity (1, 26 –28). The sprawling places of contemporary America are difficult to navigate without a personal vehicle, and often dangerous for those who make the attempt ( 29 ). Though the external costs of such a system are well-established, the development of these places is still the norm, further embedding automobility in the United States. The use of an automobile in America is presented as a freedom of flexibility, but a peek behind the curtain reveals a society entrenched in automobility with no viable alternatives for millions of households ( 1 ).

Transportation Choice/Captivity Definition

The assignment of transit users to the binary characterizations of choice or captive riders is a common technique employed primarily by transportation planners. This is largely motivated by a desire to increase transit ridership through the lens of customer loyalty, identifying choice riders as a subsection of the market who may be more willing to use transit if certain quality and reliability standards are met, while designating captive riders as the minimum level of ridership, independent of service quality ( 30 – 32 ). The concerns of the captive rider market can in this framing be dismissed, as these users must use transit to meet their transportation needs, having no other viable modal options or ability to purchase personal vehicle use ( 9 ). The use of captive rider identification to exclude user perspectives and a tendency by some to stereotype captive transit users has led to calls for the retirement of these terms. Guerra rejected the traditional understanding of rider captivity, finding that high-quality transit attracts users of all backgrounds and that many transit-dependent users still regularly use vehicles ( 9 ).

Researchers in the UK and Australia consider whether vehicle ownership in lower-income or otherwise financially distressed households is an expression of “forced car ownership” ( 33 – 35 ). Krizek and El-Geneidy approach this issue from the perspective of transit market capture: in which captivity is defined not by income but by preference, describing those who do not use transit but are amenable to the idea as “potential transit users,” and “auto-captives,” those who do not have transit available to them or are ideologically opposed to transit use ( 36 ). Automobile ownership as a condition coerced by land-use and transportation policy has regularly been discussed in regard to transportation affordability, though direct, formal study of this condition and its potential impact on financially distressed households in the United States is sparse. Brown found variation in households without vehicles, describing riders who do not own vehicles though they can afford one as “carfree,” and those who do not have access to vehicle use because of financial limitations as “carless” ( 37 ). Riders in Brown’s study list multiple reasons for not using a personal vehicle, but the most common response was the financial burden that cars place on households, with 48% of rider households forgoing vehicle ownership because of cost ( 37 ). As financial limitations are a significant cause of transportation captivity and our study aimed to assess costs, we selected income for use in our definition, though there are many households who may find their transportation choices limited by age, health, legal, or other circumstances. We used this understanding as the basis for our transportation classifications, asking whether a household owned a vehicle and whether that household was in poverty.

The definition of poverty in transportation studies is variable, as strict income thresholds fail to capture the complexity of households, such as those with nonworking dependents or those with high income but burdensome recurring financial obligations. A nonstandardized approach to defining poverty or financial distress is not necessarily a problem as the definition should be determined by the limitations of the dataset. If a more locally relevant, nuanced definition of the circumstances that define financial distress is possible, this should be implemented. For this work, we used a similar method to that of Blumenberg and Pierce, who considered the bottom quintile of income earners as low-income households when studying transportation poverty ( 24 ), a term used to describe those suffering from issues of transportation affordability often prompted by vehicle ownership costs ( 38 ). As elucidated by Guerra, the terms captive and choice have a somewhat contentious history in transportation planning, but their familiarity is valuable ( 9 ). They have been repurposed for our transportation choice/captivity matrix. Choice/captive riders, carless or carfree households, forced vehicle ownership, autocaptives, all present evidence of the coerced experiences of those embedded in an imperfect transportation–land use arrangement. In bringing these terms together under the mantle of the transportation choice/captivity matrix, we provide a more encompassing metric that provides greater understanding of each user category and their relationship to one another.

Theoretical Framework

The existing framework to interpret the experience of transportation captivity is limited. Transit users are considered captive by their lack of a vehicle and by their relative poverty, which would limit their ability to purchase mobility in any form. Drivers exist as a separate class of people who have invested in an automobile, the supposed ideal mode of transportation, and would therefore have no reason to rely on other modes of transportation (Figure 1). This fails to consider the existence of the large cohort of drivers struggling to afford vehicle use.

Figure 1.

Existing framework of transportation captivity as interpreted by the authors.

In our updated transportation choice/captivity matrix, we broaden the existing interpretation to include the dependency experiences of households who own vehicles (Figure 2). Captive driver households own vehicles even if they may struggle to afford them, whereas choice driver households own cars and are less likely to be burdened by the cost of the experience. Households are considered to be captive if they are in the lowest two income deciles, and riders if they do not own a vehicle. It is much more common for households in the United States to own vehicles than to not. More than 90% of households have access to a vehicle, so although captive riders made up 4.5% of households in our survey, 15.5% of households fit the captive driver definition ( 39 ). The same held true for choice households: 6% did not have vehicles, and 74% did. Lower income households who do not have a vehicle may suffer from a lack of mobility, but there are many more households who pay the very high costs of vehicle ownership regardless of their ability to do so comfortably. The transportation choice/captivity matrix knits together two threads of understanding in transportation planning, that of the choice/captive rider, and that of forced car ownership. The concepts are natural extensions of one another, both used to explore the potentially coercive nature of the transportation–land use patterns typical of automobility, with emphasis on the experiences of certain groups identified using income and vehicle ownership status. Whereas there is weight to Guerra’s argument that the choice/captivity model is not precise enough to describe the breadth of travel behavior and preference, it is an appropriate lens through which to examine transportation affordability, as vehicle ownership and financial distress are the keystones of transportation poverty. Assessments finding no evidence of disparate transportation spending across income deciles do not presently account for variation in vehicle ownership between these groups ( 40 ). Most transportation spending is related to vehicle ownership and use, so although deciles average a relatively static annual spending of 17% to 20%, there is more than a 20% difference in vehicle ownership rates between very low-income households and other income groups ( 39 ). This suggests low-income households who own vehicles may be more likely to be burdened by vehicle costs than other groups as the average is depressed by extremely low spending in no-car households ( 40 ).

Figure 2.

Transportation choice/captivity matrix.

Other interpretations of transportation equity may include additional nuance in modal variation, such as households who own a vehicle though regularly commute via bicycle, transit, or carpooling. We did not find additional categorization to be necessary in this context of the initial transportation choice/captivity matrix definition as these households participate in vehicle ownership regardless of individual strategies used to minimize transportation costs. Identifying households who engage in these travel behaviors may be useful in future studies to expand the definition of captive drivers to include those who exceed our low-income thresholds, but still struggle to afford transportation costs. It may also be beneficial to envision the matrix as more of a continuum, where the experience of car-light households, those who own fewer vehicles than there are adult household members, can be more fully explored. Though limited in its definition and by our data, this framework allows us to explore trends in household spending via the distinct experience of each group and their prevalence within a study area using an easily replicable metric.

Data and Methodology

The BLS CE Survey captures quarterly income and expenditure data, along with critical household characteristics. Though trends in transportation spending are influenced by existing physical infrastructure, spending can vary widely between neighbors with similar household characterizations ( 38 , 40 , 41 ). Therefore, we selected the unweighted BLS CE data from 2017 to 2019 for our investigation, as it provides directly surveyed spending trends from households. This survey is not without its limitations. The sample size of approximately 52,300 is small compared with the 122.4 million households in the United States. Data are self-reported, rural areas are under sampled, and variable names and definitions change over time ( 42 ). Income is understood to be inaccurate as some people do not feel comfortable reporting this number or report high losses on investments that might result in a misrepresentation of available household resources ( 42 ). Retirees, students, or people living on savings might be interpreted as resource restricted owing to having no income, when, potentially, they could be living comfortably regardless. Spending may be higher than income in some deciles as many households spend their savings or incur debt. The BLS has also taken measures to protect user privacy as is required of public agencies. Identifiable information has been scrubbed, bracketed, or imputed in some places. As income reporting hesitancy, misrepresentation of financial status, and data manipulation for privacy purposes are fairly widespread issues, we used income only to define our decile brackets. All other analyses considered direct spending by households.

The CE data were collected by the U.S. Census Bureau on behalf of the BLS, both as a spending diary, where smaller, more frequent purchases are tracked, and as a household interview, which details larger and more stable household spending, such as utilities, rent, or car insurance ( 42 ). The foundation for our analysis was the Public Use Microdata interview files, which had been annualized for all quarters in 2017, 2018, and 2019, following procedures outlined by the BLS. All years were considered together to increase the number of records available for analysis. Each record represents a consumer unit, allowing for multiple economic units to exist within a household, creating a more realistic depiction of those who cohabitate but do not share assets or income. We will use household to describe consumer units throughout this paper as it is a more commonly understood term.

Any mention of households in the remainder of this document refers to households with at least one earner, as we exclusively considered households with at least one working household member. This excludes households with retirees, nonworking student households, households with members who are looking for work, or households with members who do not work because of disability. Our purpose in this exclusion was to capture the trends in transportation for households that are actively generating income. Vehicle use is often necessary to access work opportunities, positioning the earner as a required driver who must invest in vehicle use to generate income. Households without income could potentially be more burdened by transportation costs than other households, so our conservative analytical approach might have downplayed the severity of the transportation burden, failing to represent the experiences of some of the most financially vulnerable populations. It is of critical importance that these populations be considered in any practical application of this work, ideally using a locally relevant, nuanced measure of financial health that may be more capable of isolating no-income households with a low standard of living. We additionally removed any households with a negative income value, which are sometimes reported because of losses incurred from rental property or self-employment investments. All other records were retained, including those with unintuitive, seemingly outlier spending habits, as suggested by the BLS from their rigorous prerelease data cleaning process. Income decile values used to define transportation choice/captivity groups were generated for this subset of the nationwide CE dataset, and again for state-level analysis to create groups within each state sensitive to local variation in income values. A summary of relevant spending via additional household characteristics is available from the corresponding author on request.

Results and Discussion

There were 52,280 households considered in our final analysis at the national level. As our data only included working households, decile breaks were are slightly higher than they would have been were they to include all households, such as those of nonworking students, retirees, or out of work jobseekers, with captive categories including all households reporting $33,600 or less in pretax income (Table 1). Captive driver households allocated an average of 16% of spending toward transportation, referred to here as the expenditure ratio, a much higher rate than captive rider households who spent just 7%. Households are considered burdened when they allocated more than 15% of total household spending to transportation. In our dataset, 39% of captive driver households were transportation burdened, whereas only 15% of captive rider households exceeded this limit. Though both captive categories are low-income, captive drivers were most likely to have their financial stress exacerbated by burdensome transportation costs. The use of income as a defining factor to our categories meant group membership reflected systemic patterns of income inequality. Most notably, households where the reference person was Black represented 15.26% of the total surveyed population, but 21.20% of the captive driver group and 32.81% of the captive rider group. Further exploration of sociodemographic and built-environment predictors of group membership, such as race, gender, household composition, and geographic location will be explored via application of this construct on the Panel Study of Income Dynamics, a longitudinal survey published by the University of Michigan spanning more than 50 years ( 43 ).

Table 1.

Average Transportation Spending by Transportation Choice/Captivity Matrix Categories

	Captive rider	Captive driver	Choice rider	Choice driver
Number of households	2,353	8,108	3,106	38,713
Percentage burdened*	15%	39%	27%	30%
Family size	1.86	2.25	2.36	2.86
Workers per household	1.13	1.22	1.57	1.80
Vehicles per household	0.00	1.57	0.00	2.37
Expenditure ratio**	7%	16%	10%	15%
Income after tax	$17,063	$21,701	$85,285	$94,731
Household spending	$23,070	$33,287	$61,456	$73,891
All transportation	$1,745	$6,922	$6,492	$13,146
Car (new)	$ 0	$1,043	$ 0	$2,612
Car (used)	$ 0	$1,808	$ 0	$2,966
Car (other)	$ 0	$21	$ 0	$81
Gas/motor oil	$481	$1,712	$1,204	$2,646
Insurance	$216	$1,157	$866	$1,817
Financing	$ 0	$133	$ 0	$332
Maintenance	$73	$622	$246	$1,062
Rental/lease/licenses/fees	$426	$196	$2,645	$658
Public transit (all)	$549	$231	$1,530	$971
Transit (trips)***	$154	$169	$844	$849
Transit (local, no trips)	$395	$62	$687	$122

Percentage burdened represents the percentage of households in each category that allocates more than 15% of spending to transportation.

Expenditure ratio is the average proportion of household spending dedicated to transportation costs.

***

Transit trips that are not part of regular daily transportation needs, may include vacation travel or airfare.

Captive Riders

Captive riders are low-income households without vehicles. Though captive riders had the lowest average income, they were least likely to be burdened by transportation costs and averaged just $1,745 in annual transportation spending. Of those who were burdened, most spending was directed toward vehicle access, such as rental cars with associated insurance and fuel costs, or out-of-town travel. Less than 0.2% of households experienced transportation burden caused by transit spending. Captive rider transportation spending may also include the cost of vehicle ownership for households who owned a vehicle for part of a year but no longer do, and fuel spending during informal arrangements, for example, a household pays for gas in exchange for a ride from a vehicle owner. The bulk of transportation spending, even in households that do not participate in vehicle ownership, was directed toward vehicle access and use.

Captive Drivers

Captive drivers are low-income households who own vehicles. Nationally, 15.5% of working households qualified as captive drivers, whereas only 4.5% qualified as captive riders. Captive categories allocated much more of their spending to transportation than their choice counterparts, but captive drivers stood out as spending the largest proportion of expenditure on transportation, even with fewer average vehicles per household. There were more captive drivers in our dataset than the two rider categories combined, unsurprising in a vehicle-oriented country. Captive drivers averaged less spending in all transportation categories than choice drivers. A larger proportion of their transportation spending went toward vehicle fuel, insurance, and repairs, regardless of their likelihood of having fewer vehicles. They spent more on used vehicles, and less on new vehicles, which may lead to higher maintenance costs, lower vehicle resale values, and an increased need for emergency towing and repairs. Both driver categories averaged less than 1% of total transportation spending on local transit. Captive drivers represented the larger of the two captive categories, with nearly 3.5 times as many households falling into the captive driver classification than the captive rider group. Figure 3 presents the distribution of annual spending toward transportation as a proportion of total annual spending in each category of the transportation choice/captivity matrix, in relation to the affordability threshold of 15%. A standard boxplot overlaying a violin graph is used to visualize kernel density and summary statistics. The violin graph reveals that many captive rider households had a higher likelihood of spending very little on transportation, with a density peak at or close to 0% of spending allocated toward transportation. The boxplot provides a more precise understanding of distributions, most notably showing median spending in the captive driver category approached the affordability threshold. Captive driver households made up a large portion of households in our data and represented a group of people who are more likely to suffer from transportation costs, as these households more frequently exceeded the affordability threshold (Figure 3).

Figure 3.

Transportation expenditure distribution by transportation choice/captivity matrix category.

Choice Riders

Choice riders are households that are not low-income but do not own a vehicle. These households are less likely to forgo vehicle ownership owing to the associated financial burden and may have some other reason for not owning a vehicle, such as environmentalism, inability to drive, or simply personal preference. This group represented the bulk of vehicle-free households, as the choice rider group was much larger than the captive rider group. Choice riders spent much more than captive riders on short-term vehicle use, in the form of vehicle rental, leases, and other charges, allowing them to increase mobility without incurring the expense of long-term vehicle ownership. Of all choice rider households, 63% directed more spending toward short-term vehicle use than to local public transit.

Choice Drivers

Choice drivers are households that are not low-income and own a vehicle. This constituency is the largest group in the United States, representing nearly 75% of households. Though they had an average income of $94,731, 30% experienced transportation spending higher than the affordability threshold. Choice driver households averaged $13,146 in transportation spending annually. They spent more than other groups on vehicle purchasing, fuel, insurance, and maintenance, and less on short-term vehicle use and local public transport. This group was most likely to have multiple vehicles. Average annual transportation spending when normalized to account for the number of vehicles was $6,047 in choice driver households. Although this spending itself may not be burdensome to more affluent households, those with multiple cars and dependents may still experience transportation cost-related financial stress. Transportation spending on vehicles was high, and all vehicle-owning households could benefit financially from a reduction in the number of vehicles necessary to accomplish daily trips.

Spatial Variation

We isolated and recalculated decile breaks for each state in our dataset, allowing us to relate statewide choice rider household prevalence to transit and built-environment variables. Of the 50 states and the District of Columbia, 40 of these entities were included in our comparative analysis, the remaining 11 states were either not surveyed in the dataset years we used or had geographic identifiers suppressed by the BLS to preserve anonymity as they are primarily low population states with dispersed settlement patterns, such as Wyoming, Montana, and North Dakota. Each state contains differing proportions of both car-dominant places that prioritize vehicle use, and transportation-diverse places that support multiple transportation modes. We used the choice rider variable to test the ability of this metric to identify potential spatial relationships, as the prevalence of these users in a transportation system may indicate successful policy that encourages the development of places where forgoing car-use is attractive to all household types. Choice ridership was selected over other categories as higher-income households are more likely to own a vehicle than lower-income households of a similar character, potentially positioning choice riders as a group whose travel needs are being met without investment in vehicle ownership, although they could afford to do so. Both New York state and the District of Columbia represented extreme cases in our dataset. The CE oversamples urban areas, so many households in the New York state-level data are likely to be located in New York City, potentially explaining why 24% of U.S. choice rider households were found in New York state. The distinctiveness of New York and D.C. in our data demonstrated the potential usefulness of the transportation choice/captivity matrix framing in the analysis of a more local dataset. Although the limitations of the data restricted our method to one that minimized the importance of transportation funding jurisdictions spanning state boundaries, we believe the results acted as a proof of concept, pointing to the type of analysis that should be done. If differences in choice rider prevalence are significant in a coarse, state-level analysis, the use of municipalities or neighborhoods may reveal more nuanced relationships between and within geographies. An example of a practical implementation of the transportation choice/captivity matrix would be to use the metric to identify a neighborhood with a high prevalence of captive drivers, but high-quality commuter transit service, suggesting that the commuter service is not fully meeting the travel needs of lower-income households, and should be further investigated by local transportation planning authorities. Local, regional, and state agencies could improve their understanding of household affordability by collecting income and vehicle ownership information in future travel surveys, or by creating a similar metric via vehicle registration data.

Built-environment characteristics are represented here using the average density of cities and towns in each state as calculated using 2017 American Community Survey data and definitions. This variable is a modified version of Zheng et al.’s density composite, where the proportion of a state population living in central cities and small urban clusters are used as weight to determine average population density ( 44 ). Urban definitions and structures vary over space, leading to urban areas with quite different characterizations, thus necessitating the weighted density variable. Transit quality was characterized using total state and local transit expenditures per capita, as published by the Bureau of Transportation Statistics in 2019. Transit expenditures included both operating and capital expenditures. An ordinary least squares multilinear regression model was compiled, yielding statistically significant relationships (Table 2).

Table 2.

Choice Rider Prevalence Versus State Variables

Dependent: Choice rider prevalenceAdjusted R²: 0.870N: 40
	Coefficient	P > \|t\|
Intercept	0.7428	0.151
Density average	0.0010	0.000
Transit expenditure per capita	0.0029	0.000

Our model revealed 87% of choice rider prevalence at the statewide level could be explained by average urban density and transit investment per capita, both of which were highly statistically significant. This gave credence to our proposal that some jurisdictions are doing better at supporting diverse transportation options. Places that exhibited higher density and higher per capita transit investment, saw a significantly higher proportion of choice riders. This relationship suggests that changes to land-use and modal funding prioritization are necessary to change the balance of transportation choice/captivity matrix categories in a place. New York and Washington, D.C. represented extreme cases that exemplified this trend. Although they are much denser and have a higher than typical per capita government transit expenditure, they were not outliers but representations of the trend at its utmost intensity in the U.S. context. We propose that choice rider prevalence indicates the diversity of transportation systems but could also be used in tandem with the other categories of the transportation choice/captivity matrix to assess the reliance of populations on automobility and its associated impacts on vulnerable populations. The development of more dense places where transit is well-funded may provide households at all income levels with the opportunity to forgo car use while maintaining access. The regression results suggest that both urban density and quality transit were important components of diversified transportation systems. Mixed-use, activity- and housing dense, transportation-diverse places create an environment where households may have more autonomy in transportation-related financial decisions. The transportation choice/captivity matrix could give planners insights into how well their transportation network is serving the needs of constituents.

Conclusions

The transportation choice/captivity matrix is a new theoretical construct created to describe transportation equity using vehicle ownership and annual household income, two prominent contributing factors to affordability. The resulting matrix consists of four categories: captive riders, those who do not own a vehicle and could not comfortably afford one; captive drivers, those who own a vehicle and are more likely to be burdened by its cost; choice riders, those who do not own a vehicle even though they could probably comfortably afford one; and choice drivers, those who do own a vehicle and are less likely to be burdened by its cost. Historic use of choice/captivity in reference to transit riders does not stem from an equity perspective, and the resulting discussions suggest that attaining vehicle ownership is an optimal, immutable state, in which the mobility provided by vehicle ownership is universally prioritized by households. An expanded understanding of transportation user categories deemphasizes the priority of vehicle use, refocusing on the inherent burden of inefficient transportation systems, while reframing captive transit riders, a historically underserved group, as equal, important participants. We operationalized the new categories offered by the transportation choice/captivity matrix to explore trends in household spending and relationships with transit quality and land-use.

Our analysis revealed distinctly different household spending trends between categories. Captive riders were least likely to experience transportation burden, averaging 7% of their total annual spending on transportation, whereas their counterparts, captive drivers, averaged in excess of 16%, a larger proportion of their spending than any other group. Of captive driver households, 39% breached the 15% affordability threshold, despite generally owning fewer vehicles than choice drivers. Captive drivers were more often burdened by transportation costs than the other lower-income group and may benefit from the alleviation of these costs. The share of each household characterization in an area is related to land-use and transit quality, represented in our analysis by a composite density and annual statewide transit expenditures. We found that 87% of choice rider prevalence in states could be explained by these variables, exemplified by the very high presence of this group in New York state, where 24.4% of all choice rider households nationwide are located. The development of these activity-dense, transportation-diverse places may be an important poverty reduction strategy, where residents, regardless of income level, could be offered the latent economic benefit of forgoing vehicle ownership without a mobility impediment. The building of these places may be achieved by improving the connectivity, availability, or quality of transportation modes that do not require personal vehicle use and allowing mixed-use zoning with minimal density restrictions wherever feasible.

The transportation choice/captivity matrix offers a method to conceptualize issues of transportation equity, useful in understanding the character of a place when planning transportation and land-use policy. Assessment of transportation affordability would be much improved by the collection and analysis of vehicle ownership and income variables in local and regional travel surveys. Transportation spending is multifaceted, yet many trends in household burden could be accounted for by asking whether the household owns a vehicle, and whether they have restricted financial resources. Although the transportation choice/captivity matrix considers only a small portion of the metrics needed to understand transportation costs, we suggest its use as a streamlined approach for interpreting transportation spending habits at the national scale and identifying localities for more detailed examination to promote greater equity.

Footnotes

Acknowledgements

The authors thank Scott Curtin and Bryan Rigg of the Bureau of Labor Statistics for their efforts in helping us to understand the appropriate uses and limitations of the Consumer Expenditure Survey. Thanks also to PhD student Jack Deschler of the Harvard Kennedy School who assisted with some particularly troublesome code.

Author Contributions

The authors confirm contribution to the paper as follows: study conception and design: Q. Molloy, N. Garrick, C. Atkinson-Palombo; data collection: Q. Molloy; analysis and interpretation of results: Q. Molloy, N. Garrick, C. Atkinson-Palombo; draft manuscript preparation: Q. Molloy, C. Atkinson-Palombo, N. Garrick. All authors reviewed the results and approved the final version of the manuscript.

Declaration of Conflicting Interests

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

Funding

The authors disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This work was partially funded via the Dwight D. Eisenhower Transportation Fellowship Program grant number 693JJ32245058.

ORCID iDs

Quinn Molloy

Norman Garrick

Carol Atkinson-Palombo

References

Norton

P. D.

Fighting Traffic: The Dawn of the Motor Age in the American City. MIT Press, Cambridge, MA; London, 2008.

National Center for Statistics and Analysis. Early Estimate for Motor Vehicle Traffic Fatalitites in 2020. Publication DOT HS 813 115. National Highway Traffic Safety Administration, Washington, D.C., 2021.

EPA. Fast Facts on Transportation Greenhouse Gas Emissions. United States Environmental Protection Agency, Washington, D.C., 2022.

Mackett

R. L.

Thoreau

Transport, Social Exclusion and Health. Journal of Transport & Health, Vol. 2, No. 4, 2015, pp. 610–617. https://doi.org/10.1016/j.jth.2015.07.006.

Bureau of Transportation Statistics. Transportation Economic Trends. U.S. Department of Transportation, Washington, D.C., 2020.

King

D. A.

Smart

M.J.

Manville

The Poverty of the Carless: Toward Universal Auto Access. Journal of Planning Education and Research, Vol. 42, No. 3, 2022, pp. 464–481. https://doi.org/10.1177/0739456X18823252.

BLS. Consumer Expenditures Report2019: BLS Reports. U.S. Bureau of Labor Statistics. https://www.bls.gov/opub/reports/consumer-expenditures/2019/home.htm#BLStable_2020_12_18_13_36_footnotes.

Litman

Evaluating Transportation Equity: Guidance for Incorporating Distributional Impacts in Transportation Planning. Victoria Transport Policy Institute, Victoria, BC, 2022.

Guerra

What the Heck is a Choice Rider? A Theoretical Framework and Empirical Model. Journal of Transport and Land Use, Vol.15, No. 1, 2022, pp. 165–182. https://doi.org/10.5198/jtlu.2022.2096.

10.

Schmitt

How Many Americans are “Captive Drivers”?

Streetsblog USA, October10, 2016.

11.

Litman

Evaluating Transportation Diversity: Multimodal Planning for Efficient and Equitable Communities. Victoria Transport Policy Institute, Victoria, BC, 2022.

12.

US DOT. Strategic Plan FY2022-2026. U.S. Department of Transportation, Washington, D.C., 2022.

13.

Litman

Transportation Affordability: Evaluation and Improvement Strategies. Victoria Transport Policy Institute, Washington, D.C., 2021.

14.

Roberts

The Average Cost of Owning a Car in the US. Move.org. https://www.move.org/average-cost-owning-a-car/#data.

15.

Bureau of Transportation Statistics. Average Cost of Owning and Operating an Automobile. US Department of Transportation, Washington, D.C., 2021.

16.

Handy

Boarnet

M. G.

Ewing

Killingsworth

R. E.

How the Built Environment Affects Physical Activity: Views from Urban Planning. American Journal of Preventative Medicine, Vol. 23, No. 2, 2002, pp. 64–73. https://doi.org/10.1016/S0749-3797(02)00475-0.

17.

Smart

M. J.

Klein

N. J.

Complicating the Story of Location Affordability. Housing Policy Debate, Vol. 28, No. 3, 2018, pp. 393–410. https://doi.org/10.1080/10511482.2017.1371784.

18.

Makarewicz

Dantzler

Adkins

Another Look at Location Affordability: Understanding the Detailed Effects of Income and Urban Form on Housing and Transportation Expenditures. Housing Policy Debate, Vol. 30, No. 6, 2020, pp. 1033–1055. https://doi.org/10.1080/10511482.2020.1792528.

19.

Blumenberg

Agrawal

A. W.

Getting Around When You’re Just Getting By: Transportation Survival Strategies of the Poor. Journal of Poverty, Vol. 18, No. 4, 2014, pp. 355–378. https://doi.org/10.1080/10875549.2014.951905.

20.

Belton Chevallier

Motte-Baumvol

Fol

Jouffe

Coping with the Costs of Car Dependency: A System of Expedients Used by Low-Income Households on the Outskirts of Dijon and Paris. Transport Policy, Vol. 65, 2018, pp. 79–88. https://doi.org/10.1016/j.tranpol.2017.06.006.

21.

Pollack

Williams

Lopez

Luna

The Toll of Transportation. Northeastern University Dukakis Center for Regional and Urban Policy, Boston, MA, 2013.

22.

Klein

N. J.

Smart

M. J.

Car Today, Gone Tomorrow: The Ephemeral Car in Low-Income, Immigrant and Minority Families. Transportation, Vol. 44, 2017, pp. 495–510. https://doi.org/10.1007/s11116-015-9664-4.

23.

Clark

Chatterjee

Melia

Changes in Level of Household Car Ownership: The Role of Life Events and Spatial Context. Transportation, Vol. 43, No. 4, 2016, pp. 565–599. https://doi.org/10.1007/s11116-015-9589-y.

24.

Blumenberg

Pierce

Automobile Ownership and Travel by the Poor: Evidence from the 2009 National Household Travel Survey. Transportation Research Record: Journal of the Transportation Research Board, 2012. 2320: 28–36.

25.

Bureau of Transportation Statistics. County Transportation Profiles. US Department of Transportation, Washington, D.C., 2021.

26.

Sheller

Urry

The City and the Car. International Journal of Urban and Regional Research, Vol. 24, No. 4, 2000, pp. 737–757. https://doi.org/10.1111/1468-2427.00276.

27.

Urry

The ‘System’ of Automobility. Theory, Culture & Society, Vol. 21, No. 4–5, 2004, pp. 25–39. https://doi.org/10.1177/0263276404046059.

28.

Zaccaro

Chafetz

Schonfeld

Dangerous by Design. Smart Growth America, National Complete Streets Coalition, Washington, D.C., 2020.

29.

Ewing

Schieber

R. A.

Zegeer

C. V.

Urban Sprawl as a Risk Factor in Motor Vehicle Occupant and Pedestrian Fatalities. American Journal of Public Health, Vol. 93, No. 9, 2003, pp. 1541–1545. https://doi.org/10.2105/AJPH.93.9.1541.

30.

Lai

W.-T.

Chen

C.-F.

Behavioral Intentions of Public Transit Passengers—The Roles of Service Quality, Perceived Value, Satisfaction and Involvement. Transport Policy, Vol. 18, No. 2, 2011, pp. 318–325. https://doi.org/10.1016/j.tranpol.2010.09.003.

31.

Zhao

Webb

Shah

Customer Loyalty Differences Between Captive and Choice Transit Riders. Transportation Research Record: Journal of the Transportation Research Board, 2014. 2415: 80–88.

32.

Taylor

B. D.

Morris

E. A.

Public Transportation Objectives and Rider Demographics: Are Transit’s Priorities Poor Public Policy?

Transportation, Vol. 42, No. 2, 2015, pp. 347–367. https://doi.org/10.1007/s11116-014-9547-0.

33.

Mattioli

‘Forced Car Ownership’ in the UK and Germany: Socio-Spatial Patterns and Potential Economic Stress Impacts. Social Inclusion, Vol. 5, No. 4, 2017, pp. 147–160. https://doi.org/10.17645/si.v5i4.1081.

34.

Curl

Clark

Kearns

Household Car Adoption and Financial Distress in Deprived Urban Communities: A Case of Forced Car Ownership?

Transport Policy, Vol. 65, 2018, pp. 61–71. https://doi.org/10.1016/j.tranpol.2017.01.002.

35.

Currie

Delbosc

Pavkova

Alarming Trends in the Growth of Forced Car Ownership in Melbourne. Proc., 40th Australian Transport Research Forum, Darwin, Australia, 2018.

36.

Krizek

El-Geneidy

Segmenting Preferences and Habits of Transit Users and Non-Users. Journal of Public Transportation, Vol. 10, No. 3, 2007, pp. 71–94. https://doi.org/10.5038/2375-0901.10.3.5.

37.

Brown

A. E.

Car-Less or Car-Free? Socioeconomic and Mobility Differences Among Zero-Car Households. Transport Policy, Vol. 60, 2017, pp. 152–159. https://doi.org/10.1016/j.tranpol.2017.09.016.

38.

Mattioli

Lucas

Marsden

Transport Poverty and Fuel Poverty in the UK: From Analogy to Comparison. Transport Policy, Vol. 59, 2017, pp. 93–105. https://doi.org/10.1016/j.tranpol.2017.07.007.

39.

FHWA. National Household Travel Survey. US Department of Transportation, Washington, D.C., 2017.

40.

Blumenburg

Transportation Costs and Economic Opportunity Among the Poor. ACCESS Magazine, Vol. 1, No. 23, 2003, pp. 40–41.

41.

Deka

Relationship Between Households’ Housing and Transportation Expenditures: Examination from Lifestyle Perspective. Transportation Research Record: Journal of the Transportation Research Board, 2015. 2531: 26–35.

42.

Bureau of Labor Statistics. Consumer Expenditures and Income Methodology. Bureau of Labor Statistics, Washington, D.C., 2018.

43.

Institute for Social Research. PSID Main Interview User Manual: Release 2021. University of Michigan, Ann Arbor, 2021.

44.

Zheng

Atkinson-Palombo

McCahill

O’Hara

Garrick

N. W.

Quantifying the Economic Domain of Transportation Sustainability. Transportation Research Record: Journal of the Transportation Research Board, 2011. 2242: 19–28.