The US transportation system is characterized by inequitable access to opportunity, leading to lower levels of economic and physical mobility for people of color and families with low incomes. The introduction of autonomous vehicles (AVs) could redress or exacerbate these inequities, depending on how they are rolled out and what policies are implemented to regulate them. In this report, we explore one aspect of how ride-hailed AVs might affect equity. Specifically, we model how ride-hailed AV services could affect access to jobs and public amenities compared with privately owned human-driven vehicles, human-driven ride-hailed vehicles, and existing public transit services in two major cities: Atlanta and Los Angeles. We investigate how results vary based on estimated transportation service costs, the spending power of potential riders, the distribution of jobs, and neighborhood location. Our analysis also looked at other influencing factors, such as possible negative bias of human ride-hail drivers toward Black people. We find that, under most scenarios and comparisons, AV ride-hailing services can meaningfully increase access to jobs and public amenities only if they offer fares at less than half the typical levels of current human-driven ride-hailing options—rates that are also much cheaper than current ride-hailed AV fares. And this increased access is unlikely to be affordable to people with low incomes, even under the most optimistic pricing scenarios.

Findings

Ride-Hailed AVs Compared with Privately Owned, Human-Driven Cars

In both Atlanta and Los Angeles, we project that ride-hailed AVs would provide access to fewer jobs within a 30-minute commute than would privately owned human-driven cars across most trip budgets and pricing scenarios. If ride-hailed AV services were offered at our lowest modeled fares and private cars charged at the high end of our estimates, ride-hailed AV services could provide a higher level of access to employment in both cities for individuals willing to spend between $5.65 and $11.30 per trip (a possible budget for people with incomes at twice the poverty level spending between 15 and 30 percent of their incomes on transportation). Under our middle-fare ride-hailed AV cost scenario, none of the tracts in either city would see higher access to employment via ride-hailed AVs compared with privately owned cars, even if the cost of using those cars was at the higher price point.

We find similar effects for the potential impact of ride-hailed AV service on access to libraries (a proxy for public amenities); ride-hailed AVs would only offer accessibility gains if they operated in the low-fare scenario and cars in the highest cost scenario. It is worth noting, however, that even if less costly on average (including purchase, maintenance, and use), car ownership is not possible for many people with low and moderate incomes because of high initial purchase costs. Financing a car also may be more challenging or costly for people with lower incomes since they are more likely to have lower credit scores.

Ride-Hailed AVs Compared with Human-Driven Ride Hail

Compared with current human-driven ride-hail services, we estimate that AV services at the same fares could offer a small increase in job accessibility compared with human-driven options under current human-driven ride-hailing fares if they exhibit less bias than human-driven ride hailing. This is because Black passengers currently experience longer wait times for human-driven services, which may not be present for ride-hailed AVs. Our comparisons assume that racial bias only occurs with human-driven ride hailing, as our model assumes that the scheduling algorithms used by AV providers to assign vehicles to requesting passengers would avoid racial bias. Yet we have seen in other sectors that, absent effective oversight and intervention, algorithmic systems can learn and replicate human racial biases.

Of course, if AV ride hailing is offered at lower fares than human-driven ride hailing, access increases substantially. However, as noted, it is unclear how much ride-hailed AV trips will cost in the future compared with human-driven ride hailing. Current AV ride hailing is more expensive than human-driven services, but AVs show promise for reducing fares over time, given the cost efficiencies resulting from economies of scale in a large fleet, the lack of driver labor costs, and no need for tips. However, other costs or profit motivation may make ride-hailed AV fares similar to or just below those of human-driven ride hailing. And, if AV ride hailing is more comfortable, as suggested by AV experts, it may actually become more expensive than human-driven ride hailing since passengers may be willing to pay more for a more comfortable and private ride.

Ride-Hailed AVs Compared with Public Transit

Compared with public transit, ride-hailed AVs provide equal or lesser access to jobs, on average, when compared with the high- or middle-fare AV scenarios we model (with a trip budget of $5.65). Therefore, employment accessibility would only expand dramatically if riders are willing to pay much more for a trip, which may be unrealistic for people with low incomes. Ride-hailed AVs offered at lower fares could expand accessibility tremendously compared with transit, but that would depend on AV services having substantially lower fares than current human-driven ride-hail options. In Atlanta, this expansion in access would most benefit higher-income neighborhoods rather than tracts with the lowest-income residents, largely because employment is clustered closer to higher-income areas.

Similarly, for public amenities like libraries, access via ride-hailed AVs would likely be less effective than transit unless ride-hailed AV fares are significantly cheaper than current ride-hail prices—or if individual trip budgets are substantially higher than the price of a transit fare.

Conclusion

Our analysis shows that accessibility benefits generated by ride-hailed AVs are heavily dependent on the costs of ride-hailed AV services and what alternative travel options people can afford to use. At current ride-hail fares, AVs are unlikely to offer meaningful access improvements for most low- and moderate-income households. Many of our results suggest that the increased accessibility produced by AV service depends on ability to pay, which means that households with higher incomes would most benefit. If riders have to spend $8 or more per trip to benefit from access improvements, they will find it more affordable to use public transit for most trips—or to purchase their own cars if they have the means to do so. If ride-hail AV fares decline substantially, this may no longer be the case, but we do not yet have evidence to demonstrate that this will occur. Another possibility is that many people will choose to occasionally supplement transit, walking, and biking on long or inconvenient trips with AV ride hailing—a pattern we have already seen with human-driven systems.