WSDOT, flickr
How much are people actually willing to pay to avoid being stuck in rush-hour traffic? Data from a Washington Department of Transportation pilot project offers some tantalizing clues—and suggests that drivers are a lot less willing to pay to avoid congestion than many experts think.
In fact, the evidence suggests that the large majority of motorists—including many well-off ones—opt to sit in traffic rather than pay a modest toll for speedier and more reliable travel. The implications are huge, and raise serious questions about how our transportation agencies plan and justify road projects.
The evidence comes from Washington State Route 167, just south of Seattle, where WSDOT operates the Pacific Northwest’s only “High Occupancy /Toll” (or “HOT”) lanes. Carpools, vanpools, and buses can always get into the HOT lanes for free. But solo drivers can use the HOT lanes too—provided that they’re willing to pay a toll. A computer adjusts the tolls in real-time to keep traffic flowing: when HOT lane traffic starts getting too heavy, the computer boosts the tolls, which discourages additional drivers from entering the HOT lane and bogging down traffic. In effect, the HOT lane serves as an “escape valve” for congestion—both by drawing some vehicles out of the general lanes, and by giving toll-paying drivers a quicker, more reliable trip.
But in addition to easing congestion, the SR-167 HOT lanes serve another important function: they provide solid data on how much drivers are willing to pay to avoid traffic delays. Drivers who can’t stand rush hour congestion will hop in the free-flowing HOT lane—and the toll rate at that moment, coupled with the time saved by traveling in the HOT lanes rather than the general lanes, reveals how much people are willing to pay to save time during rush hour.
So far, the HOT lanes have attracted far fewer toll-payers, and much less revenue, than the early forecasts had predicted. During the first year of operation, the HOT lane tolls collected just $316,600—over half a million dollars shy of the projections (pdf, see Exhibit 11 on p. 15) that were made before the HOT lanes were opened. By the end of 2011 the gap between the projections and reality had exceeded $1.4 million, with toll revenues more than 60 percent below the early forecasts. (Click on the chart to the left for details.)
So why were those projections so over-optimistic about the HOT lanes’ appeal? It isn’t because the HOT lanes haven’t been working. In fact, according to WSDOT, the HOT lanes consistently offer a faster, more reliable commute than the regular lanes. The HOT lanes are working pretty much as advertised.
Instead, we suspect that there are two reasons for the unduly rosy forecasts. First, the early projections overestimated how quickly traffic would grow over time. Second, planners may have overestimated how much drivers are actually willing to pay for a quicker trip.
Let’s run the numbers, shall we?
The first problem—overestimated traffic growth—is easy to spot. The state’s 2006 traffic and toll revenue study assumed that traffic volumes on SR 167 would grow at about 2.5 percent per year. But as WSDOT data shows, traffic volumes on SR 167 fell for three consecutive years, starting in 2006—including a drop of almost 5 percent in 2008, when gas prices spiked and the economy cratered. Traffic volumes have picked up since then, but are still well behind forecasts. With far fewer cars than anticipated, there’s been less congestion than predicted—and, as a result, less incentive for solo drivers to pay to use the HOT lanes.
The second problem—how much drivers are willing to pay to save time—requires some deeper analysis.
According to WSDOT figures for 2011, northbound drivers during peak morning hours paid an average toll of $1.75 to enter the HOT lane, saving about nine minutes in the process. Southbound evening peak-hour travelers paid $1.25 for about six minutes of time savings. Given those values, peak hour HOT lane toll payers apparently are willing to spend about $12 for every hour they save in traffic.
Yet only a small fraction of peak hour drivers (13 percent in the morning, 6 percent in the afternoon) are paying to use the HOT lanes. So for solo drivers during peak hours on SR-167, the median value of travel time savings—the point at which half the drivers would pay more, and half would pay less—is by definition below $12 per hour. In fact, considering how few drivers use the HOT lanes, the median value of travel time savings could be far less than $12 per hour.
Traffic theorists believe that North American commuters are willing to pay at least half their hourly wage to avoid being stuck in rush-hour traffic. (See Todd Litman’s excellent literature review for more details of travel time valuation.) If that’s really correct, then with travel time savings valued at $12 per hour, commuters who earn about $48,000 per year ($24 dollars per hour, for 2,000 hours per year) should be willing to hop in the HOT lanes to save some time.
So the fact that so few drivers use HOT lanes during rush hour suggests two possibilities:
- Only a tiny fraction of SR-167 drivers earn $48,000 per year or more, or
- Peak-hour SR-167 drivers aren’t willing to pay half their hourly income to save time in the HOT lanes.
The first possibility seems unlikely. Data from the Census and WSDOT suggests, quite clearly, that plenty of SR-167 drivers earn more than $48,000 per year, yet aren’t hopping in the HOT lanes during rush hour! The SR-167 “travelshed,” as defined by WSDOT, actually has a higher share of higher-income residents than the state as a whole. (See method notes at the end of this post for more details.)
So that leaves us with the second possibility—namely, that in the one place in the Pacific Northwest where drivers are offered a choice to pay out of pocket for a quicker trip, they’re simply not willing to pay very much. Most drivers, even those from high-income households, would simply prefer to sit in traffic, rather than pay for a little speed.
The implications could be huge. Transportation planners typically justify big urban road projects based on drivers’ economic interest in saving time. The projects are costly, and the time savings may be illusory; but that’s how big roads are sold to the public. Yet if drivers themselves aren’t willing to pay much for quicker trips, shouldn’t we expect WSDOT to reconsider how much it should be investing in wider roads?
Method Notes
High-income drivers opting out of the HOT lanes. At the northernmost point of the northbound HOT lane, the average per-lane general purpose lane volume for the morning peak-hour is 1,143, or 2,286 trips in both lanes. At the same time, 1,008 cars moved through the HOT lane, only 420 of which paid the toll. Assuming that all the GP lane traffic is solo drivers, then, there are 2,706 peak-hour solo drivers each morning.
Within the SR-167 “travelshed” (as defined by WSDOT – pdf, see pages 2-3 – 2-5), data from the US Census shows that about half of all households have total incomes above $60,000 per year—with 23 percent of households earning more than $100,000 per year.
If SOV drivers reflect the same Census demographics of the travelshed established by WSDOT, than 622 (2,706 * 23%) come from households earning at least $100,000 per year. However, according to a survey fielded by WSDOT, 45 percent of tolled HOT lane drivers come from households in this income bracket. So that translates to 189 (420 * 45%) drivers from households earning at least $100,000 per year during the peak-hour northbound commute. With 189 high-income SOV travelers out of an expected 622, roughly 70 percent are opting out of the HOT lanes.
In short, real-world experience suggests that less than one-third of drivers from high-income households are willing to pay $12 per hour for a quicker trip during rush hour.
Our estimates are rough, and there are confounders. Not everyone on the road during rush hour is a commuter (and traffic theorists say that non-commuters pay as little as one-quarter of their hourly wage to save time). On the other hand, solo drivers tend to be better-off than average—simply because poorer folks drive less. To complicate things still further, some drivers may be uncertain about potential time savings: WSDOT posts toll rates at the entry points to the HOT lanes, but not how much time toll-payers can save. These factors add uncertainty to any analysis—so there’s no precise data to tell us the actual hourly earnings of the folks on SR-167, or what share of their income they’ll pay to avoid congestion.
Sources:
WSDOT 5-Minute Traffic Data Archive, SR 167 2008 – 2011, mileposts 24.61 northbound and 20.68 southbound.
WSDOT SR 167 HOT Lanes Pilot Project, Third Annual Performance Summary, May 2008 – April 2011 (PDF)
US Census American Community Survey 2006 – 2010, Table DP03, 5-year estimates, applied to Census tracts in the SR 167 travelshed identified by WSDOT in “SR 167 HOT Lanes: Social, Economic, and Environmental Justice Report,” 2007, 2-3 – 2-5. (PDF)
Clark Williams-Derry
I just wanted to clarify one thing: NONE of the analysis above suggests that HOT lanes are a bad idea. They really do seem to be working. It’s just that the mobility “product” hasn’t sparked the demand, either in numbers of travelers or in dollars, that planners had expected.
One lesson here is that tolling may be more effective as a traffic management strategy than as a way to generate lots of revenue.
Joe Cortright
Brilliant work, Sightline!
This is what we economists call “revealed preference.” Traffic forecasters go to great lengths to guess what the average value is that we attach to travel time. They rely heavily on making heroic and often unverifiable assumptions based on “stated preferences” taken from surveys. The beauty of the HOT lanes is that they provide a real-world empirical test of how much people will actually pay to save a few minutes on a peak hour trip. This is “best evidence” and trumps the convoluted survey-based guesstimates that the DOTs routinely use.
And what does it show: Bottom line–roughly 90% (87% AM, 94% PM) of drivers value their time at less than $12 per hour. And this is for a mix of travelers that tilt pretty heavily to higher income households. (Sightline’s methodology for estimating income in the corridor is very conservative; travelers, particularly peak-hour single occupancy car travelers, are likely to be from households with higher incomes than the average travelshed resident). The real median value of travel time must be a lot less than $12.
This has both financial and policy implications. The financial implication, as in the case of the SR 167 Hot Lanes, is that tolls are going to produce significantly less revenue than the DOTs had hoped. The policy and practical implication: there’s just no real demand for additional lanes to speed traffic. Most travelers don’t value the time savings enough to pay for the cost of the project. Sure, they’ll use additional capacity if it’s free–i.e. somebody else pays for it–but its not worth the time savings if they have to pay.
And finally, as Clark says, this is not an argument against road-pricing. Road pricing contributes to greater efficiency by making it possible for those who do value the time savings highly to get where they need to go quickly. The fact that they’re willing to pay the toll means that they get even more than $12/hour in benefits from the time savings. And for the rest of us, we can opt to save a bit of money by taking a few extra minutes on that trip.
Kudos to Sightline for showing how this works. But they’ve just scratched the surface. The HOT lanes are a terrific laboratory for constructing real world measures of the value of travel time. Is WSDOT going to use this data to inform future decisions? Or will it continue to rely on the kind of “stated preference” modelling that leads to chronic over-forecasts of traffic levels and demand? Stay tuned.
Brent Roraback
Interesting stuff. One thing I do wonder about is the degree to which confusion about the program plays a role, i.e., do HOT lane toll prospects understand how tolling for the lanes will work? Is there an education issue also at work? Awhile back I finally decided to get a ‘Good to Go’ sticker (mostly for semi-frequent toll bridge trips). But before I got it I know I’d look at the HOT lane signs and be a bit puzzled about how it would work. Did I need a sticker for it to work? If not, how would I be billed? If I missed the bill would that be some sort of traffic infraction?
I only half wondered about these questions at the time and now that I’ve gone through the GtG sticker and debiting process I’d have no concerns about paying for HOT lane use if I was stuck in traffic. But I’d be interested to hear whether factors like this (understanding of the program, GtG program usage/penetration, etc.) were considered.
Zachary Howard
Hi Brent, and thanks for your comment.
I think a lot of the confusion over using Good to Go on the SR 167 lanes comes from the varieties of stickers and transponders available.
WSDOT has indicated that motorists who use the HOT lanes with a license-plate mounted Good to Go pass or the windshield-mounted sticker will be charged the toll regardless of whether they have an additional passenger that would qualify them as high-occupancy.
To avoid this, users would need the moveable or switchable Good to Go passes that can be either removed from the windshield or switched off prior to HOT lane use.
There are state troopers along SR 167 monitoring for solo drivers without a pass. Unlike other tolled infrastructure in Washington, there is no “bill me later” option – just a ticket!
I agree this might be confusing to the point that it deters new users, but after 3 years and a steady (if lower-than-predicted) increase of tolled trips, I think it safe to assume that people who would benefit from the time savings of HOT lanes are gradually overcoming the knowledge barrier.
Clark Williams-Derry
I had the same thoughts, Brent. But I think Zack has it right: if you’re a *regular* commuter on SR-167, and you consistently see the HOT lanes moving much faster than the general lanes, you have an incentive to figure out the Good-to-Go pass. The cost of figuring out GtG is definitely a hurdle, but it doesn’t strike me as a huge one — which may show, like the tolling data, that there’s not a huge demand for paying, either in money or free time, for faster travel on SR-167.
Paul Clapham
Up here in Vancouver a similar “experiment” just took place. We built a bridge, the Golden Ears Bridge, which replaced a slow ferry with a huge rush-hour backlog. And we put a toll on it which was about 3 or 4 dollars for a passenger car.
And then they found the traffic volumes were less than the forecasts.
Yes, they might have overestimated the traffic volumes for various reasons, but certainly one of the reasons was that a lot of people preferred to take a longer drive over a more congested bridge to avoid having to pay the toll.
I think that, psychologically speaking, comparing a toll to something which is free is a lot different to comparing two tolls of different amounts. In fact just charging for something which was free before causes much more resentment than increasing the cost of something which wasn’t free before.
Pamela
Don’t drivers need to sign up for and install a Good To Go device in their car in order to use the HOT lanes. I didn’t see any information about how those sign ups are going. Perhaps it’s too much of a barrier to allow for the kind of behavior the models assume?
Barb Chamberlain
I find this interesting for another reason in addition to those noted in other comments: If getting to work at a slow pace is acceptable (i.e., time isn’t money after all), how many of these drivers could switch to biking along alternate routes on which they could keep moving? Even if biking is slower than driving it represents constant forward progress, which is a far less frustrating commute than one spent mostly sitting in traffic.
The need for “speedier” transportation is often cited as a motor vehicle advantage over bikes, when in heavy traffic that’s illusory. A slower but happier and healthier commute could free up some asphalt for those who want to stay in their cars, and for a far smaller investment in infrastructure that keeps people moving.
Todd Litman
Well done, Zachary and Clark.
This finding is consistent with my study, “Changing Vehicle Travel Price Sensitivities: The Rebounding Rebound Effect (http://www.vtpi.org/VMT_Elasticities.pdf ), forthcoming in the journal Transport Policy, which indicates that automobile travel is much more price sensitive than most transport models assume.
Stephanie Gardner
What I have always found puzzling is the fact that people don’t consider the price of gas when choosing longer routes (either in terms of time or distance) to avoid tolls. You’ve left that out of your calculations as well.
A car traveling at 60 mph getting 30 mpg is spending $7.60 an hour on gas, if gas is priced at $3.80 a gallon. So saving 9 minutes of commute time equals saving $1.14 on gas. And that’s assuming 30 mpg fuel efficiency, which in traffic decreases significantly.
Paying the $1.75 thus might actually save people money.
Bill Popp, Sr.
Howard and Derry — great analysis of an otherwise smiley face spin job by WSDOT. Only fault I find is your conclusion that highways are justified based on what motorists are willing to pay thus wider highways should be rethought. I would agree, however, that future HOT lanes such as the current WSDOT I-405 corridor proposal for HOT lane widening only should be rethought, partly based on what you have observed about willingness to pay.
FYI highway widenings are primarily justified based on ability to accomodate projected vehicle capacity needs (after other practical modal choices are exhausted) that derive from serving desired regional and subregional land use plans, not from willingness to pay a toll assumption. Avoidance of future facility peak period breakdown is the primary objective. There are conversions of total person hours of delay to dollars and then cost benefit ratio comparisons between alternatives but these are more in the window dressing department.
There is an amazing paralell between VMT (vehicle miles of travel) and GDP and most business owners and even some economists get this — no one wants his customers or his employees (or his downtown) to be accessibilty disadvantaged. Unfortunately, the state’s elected and many of our regional policy makers do not get this and prefer to seriously constrain future VMT growth, in areas slated for major growth, through legislation currently on the books — the VMT Reduction Act. This in turn translates to “system management” with the principal weapon being HOT lanes — an incredibly blunt weapon that is also highly elitist. That is why the current I-405 HOT lanes plan will likely be a disaster for the corridor communities and the traveling public.
Toll projects such as SR 520 corridor widening are of course based on assumptions about willingness to pay and are appropriate ways to pay for very expensive projects. Fuel taxes and the former Motor Vehicle Excise tax have been excellent low collection cost methods of paying for road improvements, and the public has demonstrated support for hikes in the fuel tax a number of times. However, the full I-405 master plan will likely require additional assistance through tolling and that would be least onerous and most effective as a general low cost toll of say $1.
This is a little more than I wanted to say but the economic theorists in this chain touched a sore point.
David Levinger
Drivers are not rational actors. My work has uncovered that most people dramatically underestimate the amount they spend annually on owning and operating a car. They tend to think they spend $3,000-4,000/year whereas the inclusive costs of owning, insuring, parking, maintaining, and fueling a car is on the order of $8,000-$10,000.
Our present transportation system is perpetuated by this incongruity and the irrational economic decisions. It is a bit amusing to me that highway construction proponents have put so much stock in tolling for financing these projects. It’s my view along with numerous others (some who have commented above) that if they actually increase the rationality of drivers decisions, highway proponents will actually drive reductions in driving.
Another take on this: people have grown accustomed to being comfortable in their cars. Despite the fact that many people hate congestion, they’ve also adapted to it. They talk on the phone, listen to the radio, and enjoy the refuge or solitude. Complaining about the traffic however may be more on the order of the weather — a conversation starter in which people talk about things beyond their control. Paying several bucks to speed up a portion of one’s drive along 167 isn’t going to change the fact that one still has to wait at all the intersections and bottlenecks at the other points on the route.
Nice work Sightline!
Jeff Dubrule
I recently added up what my car has cost me since I bought it… I bought a $25,000 Subaru 11 years ago, and counting gas, insurance, taxes, fees, repairs, etc., etc. I’ve paid ~$50,000. Which is kind of a lot.
This does not include any residual value my car has, were I to try to sell it, but at this point that can’t be much more than $5000 or so.
Clark Williams-Derry
Wow, that’s a lot. Smart of you to tally it all up! Of course, there might be additional costs too — e.g., the cost of parking your car at home. I’ve got a garage, and I assume that the cost for land and construction was built into the purchase price of my house.
Jeff Dubrule
Hmm. Good point. However, unlike my disposable car, I think I will probably get back what extra I paid to get a house in Wallingford with a garage & a driveway spot.
Jeff Dubrule
I think most drivers don’t think of the cost of driving on a per-trip basis. In fact, most of them don’t want to think of it at all, but they consider maintenance, gas, insurance, etc. as “committed” costs, mostly immutable. This is reinforced by the fact that gas tanks carry enough fuel for several trips, so buying gas every week or so is just a fact of life, like your trash-pickup bill.
The only time to consider it is when buying a new vehicle, but external factors (person/cargo capacity, offroad capability, sportiness, towing limit) usually trump gas-mileage and other per-trip costs.
Now that I’m biking/busing (with free employer-provided pass) more, when I do drive, I am rather conscious of the trip-cost… 30 miles to the arena tonight for my men’s-league hockey playoff = ~ 2.25 gallons of gas @ $4/gal for the good stuff = $10. The league fee (for ice time, referees, admin, trophies, etc.) works out to about $20/game, but I bet if I asked my teammates why hockey was expensive, not 1 would bring up that 30% of the cost is getting there.
However, while driving is perceived as free, however, tolls, including HOT lane-fees, are not perceived as free; each one is a buying decision. This leads to the same sort of penny-wise-pound-foolish sort of thinking that has a company buy the cheapest printer paper that causes paper-jams, which results in a $60/hr employee spending their time un-jamming the printer (instead of advancing the interests of the company: double whammy).
Prefer Not To Say
In looking at the methodology here, I wonder if there is a significant concern that should play a larger role in your future analyses – the perception of the driver.
This is essentially a debate over stated preferenece versus revealed preference. You have done some analyis of the revealed prefernce but you don’t really have the stated preference. You have analysis of an assumed preference on behalf of the planners. Additionally, the perception of the driver could provide inroads into other confounding factors:
– Do the drivers have an estimation of how much time the HOT lane would save them?
– Do they have an estimation of how much time they would have to stay in their lane if they did not avail themselves of the HOT?
– What are their thoughts on the economy?
I think the last one is another issues you may want to explore in future analyses. You said it yourself – drove less when the country went into recession. Isn’t it possible that right now, as people hear more stories about slow and steady recovery that their preference may change? I think the idea that the use of HOT is a function of the perception of the driver overall versus a function of the driver’s perception of traffic has larger implications to future policy development.
All in all though, a well-done analysis.
Prefer Not To Say
Please forgive the typos above.