By design, express lanes present users of a corridor with a simple choice: choose the generally slower-moving General Purpose Lanes (GPLs) or pay a toll to travel on the generally faster Managed Lanes (MLs). Understanding how drivers value, compare, and choose between these two alternatives now and in the future is a critical consideration in ML investment decisions, pricing strategies, and policy.

Research into travelers’ use of MLs to date has focused on the individual’s willingness to pay a toll to realize travel time savings. ML choice models focus on the trade-off between price (the toll) and value (usually, reduced travel time). More recently, researchers often include travel time reliability, arguing that all else equal, travelers will pay for a more consistent travel experience. These hypotheses were confirmed by surveys of travelers who would, on paper, pay reasonable toll amounts to save time and improve their travel time reliability.

This basic model is built on logical, sensible assumptions and is workable in a limited data environment where planners need few inputs beyond demographics, speed measurements, and traffic counts. However, there now exist a few freeways where trip data is available from both the GPLs and MLs, capturing individual traveler choices over a long time horizon. Surprisingly, these revealed choices do not conform to earlier hypotheses regarding lane choice. This research uses empirical trip data from both Katy Freeway in Houston, TX and North Tarrant Express Freeway in Ft. Worth, TX to investigate lane choice on freeways with MLs. Novel findings include:

• Most travelers were not really choosing between lanes — they always used the same lane regardless of travel time and toll.
• Willingness to pay for travel time savings (expressed as toll paid divided by time saved on the MLs) was consistent across traveler groups when broken down by how frequently they traveled on the MLs.
• Travelers that did use both sets of lanes (termed “choosers”) often made choices that appeared counter intuitive based on travel time savings and toll rate.

A Sample Size of 24 Million Trips

Most travel behavior studies consist of a sample size of a few hundred to a few thousand travelers. They are often conducted surveys (stated preference) or examinations of a sample of real-world data (revealed preference). Our dataset includes nearly 2 million unique travelers (unique transponders) making over 24 million trips (the series of transponder reads along the highway) across a three-month period. Therefore, the size of our dataset is orders of magnitude larger than previous studies and includes the majority of vehicles who paid to use the MLs. It uniquely considers data from two different roadways in two different cities.

If the observed travel behavior is similar on both facilities then observed deviations from long-held assumptions must be, at the very least, not attributable to the unique design, driver sentiment, small sample size, or the specifics of a single urban corridor.

Finding ‘Choosers’

Using the Katy Freeway as an example, we examined frequent travelers who averaged over 30 one-way trips per month on the freeway. The data show that 56% of those frequent travelers did not use the MLs at all, while 44% were seen on both sets of lanes with only 1% exclusively used the MLs.

When we look at the share of choosers — those drivers that selected both lane types at least once, we see that the frequency with which the travelers used the freeway did not appear to substantially influence their likelihood of being a chooser. With the exception of the infrequent group, travelers tended to exhibit similar likelihood of being choosers at each facility. These data are only transponder reads and there is no information on the users, their trip purposes, trip urgency, or even if it was the same driver (for example, it could be different family members driving on different trips) each time the transponder was identified.

A Choosers Choices

The chart below summarizes the PM rush hour, workday lane choices of a randomly-selected commuter from the NTE corridor from September 2015. Each mark represents a trip with the choice of lane indicated by the shape.

What is remarkable about this pattern, which we also observed among choosers in aggregate, is that the price and time savings, even across a fairly wide range, offers little help in predicting the individual traveler’s lane preference for a specific trip. This particular driver was presented with prices between $3.75 and $6.10 and time savings between a few minutes and about 10 minutes. If the prevailing normative theory held at the individual level, this driver would have a higher affinity for the ML for trips on the right hand side of the figure (similar price, higher time saved).

Without additional data we, again, can only posit hypotheses about this observation. One is that the driver’s revealed preference for the MLs — in this specific case, 8 out of 18 trips — is not the result of eighteen discrete choices, carefully informed by trip purpose, price, value of reliability, and travel time, but rather an unobservable factor that outweighs the classic VOT assumptions. For example, sometimes there is an event (for example, their turn to cook dinner) so the user must be on time while other days there is flexibility in arrival time. Or, possibly, it could simply be the mood of the traveler. Some days they do not feel like dealing with GPL traffic or they feel like giving themselves the luxury of a free-flow trip home (we assume home based on the regular PM occurrence of this trip).

Conclusions & What’s Next

“I call it theory-induced blindness: once you have accepted a theory and used it as a tool in your thinking, it is extraordinarily difficult to notice its flaws. If you come upon an observation that does not seem to fit the model, you assume that there must be a perfectly good explanation that you are somehow missing… Disbelieving is hard work.”

- Daniel Kahneman

A favorite interview question of your author is “Imagine that a person made a highway trip yesterday and I want your help guessing if they chose the ML or GPL. You can ask me or the driver for any information to help you guess. What would you ask?” The reader may not be surprised to learn that the vast majority of students of transportation, without blinking, ask about the level of congestion yesterday and ask the driver how much money they make, if they had a passenger in the car, the purpose of their trip, and how many children they have. These are among the typical inputs to a model of ML use. It is a case of theory-induced blindness. Compared to the depth of information within the student’s grasp, his proposed solution is shallow. He sticks with what he knows and passes up the opportunity to ask the questions that non-transportation students tend to come up with first, like “Do you often use the express lanes?”.

After this preliminary review of this dataset, we have arrived at a few conclusions. First, many travelers on priced ML corridors are not choosing between the lanes. This finding on its own is not inconsistent with the classic VOT model or even more modern bounded rational behavior models, but it is notable that a substantial number of apparently informed, frequent drivers are sticking to one set of lanes across diverse circumstances. Several hypotheses were posited, including a developing awareness of the MLs and decision aversion. Better understanding of how to predict if a traveler is a “chooser” or “non-chooser” would be a significant step forward in predicting the use of MLs.

Second, predicting when a chooser will select MLs versus GPLs is not straightforward. We set aside the question of the price-time savings trade off and observe that frequent drivers of the highway do not exhibit behavior that suggests they are ‘optimizing’ their lane choice any more than the very rare user. Third and finally, individual lane preferences observed over time suggests at least some and possibly most choosers do not seek to ‘optimize’ each trip, but rather select a mix of lane choices that does not correspond to travel time savings and toll paid. How such decisions are made should be an area of intense future research and would be critical in developing an alternative, cohort-centric approach to understanding and forecasting managed lane use over time.

John Brady is an Austin-based transportation analyst. You can reach him at johnbrady.jb (at) gmail.com or find him on LinkedIn.

How new new thinking about travel behavior can improve investment in priced express lanes

This is an expanded transcript of the talk “Pocketbook Impacts: Social ROI for Managed Lanes” given at the Transportation Research Board’s 2018 Annual Meeting symposium entitled Managed Lanes: Driving Tomorrow’s Regional Economic and Mobility Goals.

Urban planners are given the Sisyphean task of reducing congestion and improving mobility against perpetual growth. With autonomous vehicles poised to increase demand for auto travel, priced express lanes are positioned to be a mobility solution that can keep the biggest U.S. cities moving. When designed well, executed fully, and operated with the political will to provide a high level of service, priced express lanes are the most future-proof solution to urban mobility.

Given the large price tag and limited resources, a lot of effort is put into prioritizing, designing, and developing express lane projects. The way these projects are evaluated and ultimately chosen for development — and how that process can be improved using the latest data and thinking about travel behavior — is our topic today.

The Cost-Benefit Ledger, Expanded

Minnesota’s Department of Transportation recently developed a new Cost-Benefit framework that it will use to prioritize express lane investment on various corridors in Minneapolis-St. Paul, already home to some of the longest-operating express lanes in the U.S. The new thinking captures primary benefits like increased throughput, time savings, and trip reliability as well as societal benefits like reduced crash rates and vehicle operating costs. They’ve also updated their estimates of construction and operation costs using data from like projects.

Look closely and you’ll notice that the impact on drivers’ wallets isn’t included. This is understandable for all sorts of reasons, notably the lack of complete travel behavior data, familiarity with the status quo, and the higher political priority assigned to the task of managing urban infrastructure. Fortunately, some of the newest research on travel behavior is coming from private operators comparing notes with academics and publicly-operated facilities to get a holistic picture of how an entire city of drivers interacts with express lanes.

Private sector operators have the unique opportunity to focus on the performance of a single asset and so invest heavily in collecting and synthesizing data to better understand traffic patterns and drivers’ thinking about how and when they drive. And this is not an esoteric exercise: when a road fails to meet its contractual performance level, the state imposes stiff financial penalties and a formal reprimand that can harm opportunities to develop new projects. A thoughtful operator will innovate, looking for better ways of operating roads where it matters to drivers. Dallas’ LBJ TEXpress and Ft. Worth’s NTE TEXpress are operated by the private sector. Both have been moving traffic for about three years and so sit at the critical inflection point between the early rapid-growth years and the steadier, long-term growth phase. Now is a good time to reflect on their headline performance and share the insights that can help other citizens and agencies think more clearly about express lanes.

Infrastructure or Product?

When managed lanes are built they are built with two kinds of promises. First, to politicians and planners, the improved link will sustainably increase mobility and keep our cities moving. To citizens, it will provide them with a reliable choice to bypass congestion and get to their destination more comfortably and quickly.

Ask a planner to tell you if an express lane is working well and their answer will be in the context of delivering on these promises: daily highway volume increases of X, the crash rate was reduced by Y, and the express lanes moved cars quickly for Z% of days. Together, these describe the lanes as a piece of infrastructure, a high-functioning sub-system within the urban transport network. In Dallas-Ft. Worth, the express lanes are performing on these counts: throughput is up, the express lanes provide an exceptional level of service, and local officials have touted accelerated economic development along these corridors.

This is where the success story usually ends. What the two Dallas express lanes show is that drivers treat the express lanes more like a product, with drivers playing the role of consumers. For the curious planner, this realization opens a new point of view. If express lanes are part product, part infrastructure, how should they be measured and explained? Online firms track monthly active users and total subscribers. Revenue in the consumer packaged goods industry is monitored by product lines and customers are clustered along multiple dimensions into actionable groups. An express lane operator might think about her different products (the congested rush-hour trip, the midday or weekend low-time savings trip) and customer types (the everyday commuter, the occasional cross-town driver, the neighbor).

What if we did that? What might we find? How might it help us prioritize and operate projects more intelligently?

Before I share some new data, I want to state that the privately operated Dallas express lanes cannot connect our users to their address and related demographic information because that information is exclusively maintained by the local tolling authority. When a transaction is generated, the tag and plate are read and validated by us, then sent to the local authority for cross-validation, and finally billed to the customer. Although this might seem like a constraint, our analytics team believes this has spurred our creativity because we are not able to lean on the metaphorical crutch provided by demographics and let our bias cut short valuable lines of inquiry.

The Myth of the Commuter

Building out the performance along more novel metrics first shows that the true impact of express lanes on mobility across a region is enormous. In a single day, tens of thousands more trips are made on these critical corridors than before development and each of those trips benefits from the new infrastructure. Whether a driver chooses the tolled or non-tolled lanes, she benefits from reduced travel times than before construction and newly added ramps and frontage roads benefit even the shortest trip.

But look across a single month and a wide cross-section of the Dallas-Ft. Worth metroplex will make a trip on the corridor that includes the NTE TEXpress. The figure below shows the share of trips made by different ‘kinds’ of drivers, binned by the number of trips they made along the highway (whether free or tolled )in September 2015.

Only 3% of all drivers seen in a month are everyday “full-time” commuters — and collectively, they make about 15% of trips. Overall, about one third of trips are made by drivers who are on the highway fewer than 8 times each month. You could say that the regular congestion on a highway is actually created by a substantially different group of drivers each day.

There is a false belief that the primary beneficiaries of improved mobility are drivers who take the same route everyday. We call this the “myth of the commuter.” In fact, the user base is so diverse that the archetypal ‘commuter’ is only a small piece of the puzzle.

Pocketbook Impacts

Considering the diversity of drivers on a managed lane corridor and that each driver can choose the express lanes when it suits them, it might not be surprising to know that the median monthly bill for an express lane user on the LBJ and NTE TEXpress lanes is $5–10/month. On the high end, the top fifth percentile of express lane users has a bill of $50/month. Express lane users’ trips are made at 50 mph over 99% of the time. Consider that your author travels seven miles to work via a normal toll road and pays a monthly bill of about $56, regardless the level of service.

Bringing these data together, it’s correct to say that the infrequent driver are among the primary beneficiaries of a new express lane. They make up a majority of all drivers who will use the road and a substantial share of all the trips. They can choose the express lanes for a high share of their trips while keeping their monthly bill low.

Think Local. Act Global.

So how can planners and elected officials fold these data into their thinking to make better decisions about where, when, and how to develop express lane projects?

First, planners should make the most of the wealth of transaction data and look at the diversity of drivers and how drivers are using their tolled networks across long time horizons. The insights derived could inform efforts in public outreach, signage, design, and pricing strategy. Second, elected officials should understand that their constituents, while perhaps not near a new express lane project, are likely to be among its primary beneficiaries. Third and finally, operating express lanes should be willing to invest in the data, analytic tools, and creative talent to explore and learn from their projects.

It’s a bright future for priced express lanes, as both Democrats and Republicans want to invest heavily in infrastructure and ensure our roads and transit services are positioned to meet our future mobility needs. Considering the end user when defining those needs can support sound investment in transportation.

John is a Dallas-based transportation analyst. You can reach him at jbrady@lbjexpress dot com or find him on LinkedIn.