Three Early Takeaways from the 2017 National Household Travel Survey

How do people get around?

This seemingly simple question belies the vastness and complexity of the transportation system that underpins almost every aspect of our lives: daily commutes to and from work; accessing vital services like child care, education, and medical appointments; social and recreational activities; and many others. Understanding how, where, and why people travel over time is necessary for transportation agencies and urban planners to manage the use of infrastructure effectively, and identify when it’s not meeting a community’s needs. Because of this, the Federal Highway Administration periodically conducts a large-scale study known as the National Household Travel Survey (NHTS).

The most recent NHTS was completed in 2017 and released earlier this year; members of over 100,000 households across the country completed detailed travel diaries of every trip they took on an assigned day, along with the trip purpose, mode of travel, and many other factors. Our research team is particularly excited because this is the first time NHTS has included data on ridesharing trips and usage (the last one was in 2009, the year before Uber first launched). After an initial look at the data, we’ve highlighted three findings below that we think are particularly compelling.

Takeaway #1: Private autos still reign supreme, while ridesharing is a drop in the bucket

Anyone who’s spent time outside the largest city centers can tell you that private vehicles (and the road networks and parking lots that support them) dominate the landscape. But NHTS data reveal just how much this is still the case: across the US, less than 0.5% of person-miles traveled (PMT) are by ridesharing, taxi, or limo.¹ Personally-owned vehicles account for about 200 times more travel, at 92% of the total, while mass transit (bus, subway, and commuter rail) and active modes (walking and biking) account for just 3% and 1% respectively.

Of course, this is a nationwide average which includes small towns and rural areas, so perhaps it’s not surprising that relatively few people use ridesharing and transit (although Uber does increasingly serve areas like these). Even if we consider only households in dense parts of large cities, private autos are still used for the vast majority (76%) of miles traveled by ground transportation. Transit use is far more prevalent, at 16%, but taxis and ridesharing still only account for about 1.7% of travel.

Percentages of person-miles traveled in each mode, excluding air and water transportation. The right-hand plot only includes households in medium- to high-density census tracts (greater than 10,000 people per square mile) of metro areas with populations higher than 3 million — examples of places at this density threshold include parts of Evanston, IL; Hackensack, NJ, and Anaheim, CA.

We’ve known that ridesharing comprises a relatively small percentage of transportation, but it’s remarkable just how small this percentage is — particularly in the largest cities.² Even considering growth in ridesharing in the year since the survey was conducted, the fact remains: the vast majority of US travel is by private automobile. While Uber is committed to making it easier than ever to get around without owning a car, NHTS data helps us understand just how much our work is cut out for us.

Takeaway #2: Households without cars are far more multimodal

In our own data we’ve found evidence that Uber is often used for “one-way” journeys that wouldn’t be possible for someone driving their own car — for instance, taking Uber out to a restaurant and the train home afterwards. This illustrates how on-demand ridesharing can facilitate multimodality, where an individual selects the best mode of transportation for each trip over the course of their day; conversely, when you leave home in your personal car, you’re pretty much stuck with it until you bring it back home.

The NHTS data support this narrative: in urban households³ without cars, no single mode of travel substantially dominates. Walking and biking account for about half of trips, 24% were on public transit, and a relatively small fraction were by ridesharing and taxi (2.9%). Not surprisingly, vehicle usage was much lower, but still significant (20%, likely people riding with friends or family).⁴ Taking a slightly different look at the same data, one-third of rideshare and taxi trips (and about half of transit trips) are taken by the 6% of people without access to a car at home.

Distribution of person-trips in four mode categories, for urban households with and without cars.

On the other hand, to state the obvious: people who own cars use them. Perhaps less obvious is the sheer magnitude of this use: people who live in urban households with at least one car took 85% of their trips by private automobile. All other modes we looked at were used far less often: e.g., 0.4% of their trips were by ridesharing or taxi, and 1.7% by public transit. The difference between households with and without cars is illustrated in the figure above. All together, the NHTS data further support the idea that, when a household doesn’t own a vehicle, their travel shifts to a mix of public transit, ridesharing, taxi, walking, biking, and other available modes to fit their daily needs.

Takeaway #3: “Deadheading” personal vehicle trips add up

Since cars need a driver, it’s often assumed that personally-owned vehicles are 100% utilized while on the road (in other words, always carrying at least one occupant for a productive purpose). We know this isn’t necessarily true — for instance, previous research has found that cruising for street parking can account for significant traffic. Furthermore, some types of trips (like shopping or picking up a family member at the airport) effectively have an “unproductive” leg that might not be necessary with efficient, networked delivery, public transit, and ridesharing services. Even if these are a small fraction of trips, they could account for a large amount of travel given the magnitude of personal vehicle travel in the US.

Since participants in the NHTS record the reason they took each trip, it’s possible to get a sense of how often these types of trips occur. Three trip types in particular (“Drop off / pick up someone,” “Buy goods,” and “Buy services”) are likely to involve one productive leg where the person or purchased good is actually being transported, and some level of “deadheading.” Since this won’t be the case for all such trips, we assume for illustrative purposes that the unproductive legs account for 25% of the trip miles.⁵

With these assumptions, “deadheading” personal cars add 100 billion vehicle miles per year to US roadways — or about 5% of total personal vehicle travel. For comparison, this is the same as all person miles traveled on public transit combined, and over eight times higher than the miles contributed by ridesharing and taxis (about 12 billion). This further serves to illustrate how efficient public transit alongside optimized delivery networks and online services can take even more vehicle miles off the road.

Conclusion

Although the transportation landscape in the US is largely dominated by personal vehicles, there is an incredible potential for public transit, bicycles, ridesharing, carsharing, and other modes to entice people away from car ownership. We are excited to see the insights that researchers derive from these data (both on ridesharing and on transportation in general) in the coming years, and to find new ways that we can leverage our own data to supplement surveys like NHTS and help cities plan for the future.

Notes

1. NHTS groups all of these “for hire” services into a single category (TRPTRANS=17). Throughout this post, we define private vehicle trips as those with TRPTRANS categories 03, 04, 05, 06, 08, 09, or 18, and mass transit as categories 11, 15, or 16. Airplane and boat trips (19 and 20) are excluded from all of our calculations.
2. As a sanity check, we compared the NHTS trip numbers against our internal data from the same period and they were generally in the right ballpark.
3. For this calculation, urban households are selected through the NHTS variable URBRUR=01.
4. To check whether this is a purely income-based effect (i.e. households that can’t afford cars using more public transit), we re-ran this for households with annual incomes over $75,000, which are more likely to be car-free by choice. The results ended up being similar; in fact, these higher-income car-free households used transit and TNCs somewhat more often.
5. 25% is assumed purely hypothetically and as an order-of-magnitude estimate — the fraction of deadhead miles on these types of trips ranges from nearly zero (where a trip is perfectly integrated into a daily routine) to 50% (where a journey is taken solely to pick up someone or something). The number of deadhead miles found simply scales with this assumption.