The Strange Economics of HOT Lanes
Upward-sloping demand, prices in excess of profit-maximization, and more
My house is a 5-minute walk from I-66, so when I have guests, I often take them to “see the economics.” Specifically, I take them to the center of a pedestrian bridge over this ten-lane highway. Six lanes are “free,” gratis. The center four lanes, however, are HOT — “high-occupancy toll”
How exactly do HOT lanes work, at least in my area?
Signs prominently display the current toll, and give you ample time to weigh whether you want to pay.
If you have one or two passengers in your car, the system charges you a toll that flexibly adjusts to keep congestion low. The toll averages around $10 for a segment roughly 10 miles long, and varies from a low of about $3 to a high of nearly $20.
Rough rule of thumb: You pay $3 per minute of travel time saved.
If you have more than two passengers, the toll is always $0.
Notice: The toll contains strong information about traffic conditions in the gratis lanes. Low tolls tell you: “Traffic is minimal in both free and paid lanes.” High tolls tell you: “Traffic is high in free lanes, but still minimal in paid lanes.”
Upshot: If your vehicle carries less than three passengers, your demand for the paid lanes has a standard downward-sloping demand curve. Though it’s probably unusually flat, because price indicates (relative) product quality. If the price is $20, a strategic driver with less than three passengers thinks, “A steep price, but $20 avoids a full-blown traffic jam.”
If your vehicle carries more than two passengers, however, your demand for the paid lanes is actually upward-sloping! No matter how high the toll gets, you pay zero to use the toll lanes. But the higher the toll, the worse the congestion in the gratis lanes — and the greater the marginal benefit of using the toll lanes. If the toll sign reads $100, for example, anyone in a high-occupancy vehicle would be a fool not to use the HOT lanes.
Added oddity: At least on I-66, I’m convinced that tolls are set above the profit-maximizing level.
During off-peak times, the price is about $3, and there are roughly 20 cars in the gratis lanes for every car in the HOT lanes. As traffic worsens, this ratio rises along with the price. (How do I know? I’ve counted the cars a dozen times at all hours of the day). The HOT lanes are so empty that they could easily triple their traffic without any noticeable reduction in safe driving speeds. (How do I know? Again, because I’ve looked many times. Check out the video I made yesterday).
Now consider: When the HOT toll is $3, at least half of the HOT lane cars are high-occupancy vehicles paying zero. So the ratio of gratis drivers to paying customers is at least 40:1. Upshot: If halving the HOT price from $3 to $1.5 enticed a mere 5% of the gratis drivers to switch, paying customers would triple, and revenue would rise by 50%. Since almost all of the costs of the road are fixed, that’s a huge gain in profits for the toll collectors.
As tolls rise, what happens to the share of paying customers? The incentive of high-occupancy vehicles to choose the HOT lanes unambiguously gets stronger as tolls rise. For other drivers, the effect is ambiguous: When congestion rises in the gratis lanes and price rises in the HOT lanes, the “best” option heavily depends on your value of time (plus sheer resentment of traffic). In the limit, if the toll were $1000, the share of paying customers would approximately equal zero.
Upshot: Existing I-66 HOT tolls probably exceed the profit-maximizing level under virtually all traffic conditions. I know that’s weird — governments’ default is to underprice, even for ultra-congestible products like roads. I am aware that this is a public-private partnership, but why would that lead to over-pricing? Yet the more I stand on that bridge and do the math in my head, the more convinced I am that whoever sets the tolls is losing money and inconveniencing DC motorists at the same time.
P.S. If anyone with access to the I-66 toll data is eager to prove me wrong, I’ll happily let them write a guest post to enlighten us all. There are some vaguely-relevant pieces on Google Scholar, but I failed to find any piece claiming to show that the current toll formula sets prices at or below the profit-maximizing level. If I’m missed something directly on point, please share in the comments.
I could see this being a case of behavioural economics where there are a lot of people who steadfastly refuse to pay for a much better traffic experience, even if the price is very low. And among the people who do properly recognize the value of their own time, they're already happy to pay $3 and lowering to $1.50 won't entice many more. We saw from the recent congestion tax implementation in New York that there were quite a few people happy to wait ages in traffic when it was free, but once a small fee was added, a significant amount switched to using the subway.
There might be better returns from making an advertising campaign to concince people to value their time better, instead of just lowering price.
I recognized that bridge right away, even before hearing you mention it in the video. We live about a mile and a half away from it as the crow flies and occasionally watch 66 traffic from the Blake Lane overpass (or the Vienna Metro footbridges).
Regarding the 66 HOT lane tolls: if I remember correctly, the contract between VDOT and the HOT lane operator (Transurban) requires the latter to maintain traffic free-flowing at speeds greater than 45 mph and penalizes them if that condition isn't met.
This is sensible on its face, as transportation engineers have found that drivers are more willing to pay for free-flow traffic conditions than for any other controllable variable. But it also gives the operator a goal that's something other than "continuously maximize the toll revenue from the HOT lanes", since they can't make it so attractive that it starts to get congested and lose its free-flow condition.