Congestion Pricing Questions

Congestion pricing for freeway capacity is a hot topic. The basic implementation of price-managed lanes known as high-occupancy toll (HOT) lanes has been rolled out in many cities, including the new lanes on the Capitol Beltway in Virginia and the retrofit of existing HOV lanes on the 110 and the 10. These lanes operate on a simple principle: when traffic increases in the lane, prices (tolls) are increased to decrease the number of people using the lane and prevent congestion.

Beyond that, though, a wide range of people have called for congestion pricing on all lanes of freeways. This ranges from libertarians who favor user fees, like Randal O’Toole, to urbanists that want to decrease the amount of driving by increasing costs, to cities and states that see potential revenues.

Theoretically, it is easy to extend the concept of HOT lanes to the entire freeway. However, it seems to me that to do so, you have to make a major simplifying assumption about your freeway network – that there are no capacity mismatches. What does that mean? It’s probably easiest to show by way of a few examples. Note that traffic jams on freeways do not necessarily indicate there’s a problem on the road at that location; rather, they are often acting as a queue of cars, pointing towards a downstream bottleneck. There are also questions for long distance trips.

The Off-Ramp Strangler: The 10 at Cloverfield

On weekday mornings, the 10 westbound into Santa Monica backs up starting at the Cloverfield/26th off-ramp. There’s a lot of employment in the area around the future Olympic/26th Expo Line station, and the local streets can’t handle the traffic volumes at peak times. The off-ramp acts as storage for cars waiting to distribute themselves on the local street network, and when the off-ramp gets full, cars start queuing up on the mainline of the freeway.



If you’re managing an HOT lane, it’s pretty easy to keep that lane flowing at a reasonable speed. You’d just charge a higher toll for the lane up to Cloverfield, and then a lower toll beyond that. The general purpose lanes act as a spillway, soaking up whatever traffic comes out of the HOT lane.

What would happen in practice if the whole freeway was tolled? Some people will try to change their travel patterns by leaving earlier or later, which is the real intent of congestion pricing. However, some people will just hop out onto the free local street network. If you charge an arm and a leg to get from Bundy to Cloverfield, maybe I decide to get off at National, Overland, or Bundy. That moves the queue of cars trying to get to office parks in Santa Monica off of the freeway and onto the arterial grid.

Disastrous Lane Drop: The 5 at Norwalk Narrows

Everyone in LA has probably experienced this at some point: you’re cruising north on the 5 in Orange County, enjoying some of the world’s finest freeway engineering, and then boom! You pass the 91 and you slam (figuratively, we hope) into gridlock on the three-lane section of the 5 through Santa Fe Springs and Norwalk. This is one of the last unreconstructed 1950s-era freeways in LA. It’s being widened as we speak, but it’s a great example of a capacity mismatch between adjacent sections of a freeway mainline.



If you’ve got a managed HOT lane here (and the Orange County section is clearly designed for that possibility), you can keep it flowing by charging a punitive toll through the Norwalk Narrows. If the entire freeway is tolled, you’d have to charge very high tolls to keep things moving on the three-lane section – so high, that you might not be able to charge anything on the five-lane section to the south. That results in a very cheap section leading into a very expensive section.

Again, the incentive is going to be for people to use the cheap section of the freeway, and then bail out onto the free local arterial grid.

Alternatives with Issues: The 405 vs North-South Arterials

This one isn’t quite so much about a freeway capacity mismatch as it is about the amount of existing congestion on local arterials.

Northbound congestion on the 405 has several causes. For one, the prolonged steep grade approaching Sepulveda Pass degrades vehicle performance, resulting in some vehicles slowing down. At the top of the pass, you have an intense weaving section leading up to the busiest interchange in the country, the 405 and the 101. Further upstream, you simply have a lot of traffic from Westside employment centers entering the freeway between the 10 and Wilshire to head home to the Valley.


Contrary to popular conceptions of LA, the north-south arterials on the Westside are significantly underpowered. Sepulveda is the only true through arterial between Lincoln and Robertson; the rest – Bundy-Centinella, Sawtelle, Barrington-McLaughlin, Westwood-Overland, Beverly-Beverwil-Castle Heights – are Frankenroads, incomplete, cobbled together from various parts, and not even two lanes in each direction. This contributes to a major lack of north-south mobility on the Westside.

If the 405 were tolled to maintain higher speeds, some traffic would shift to this free ragtag network of north-south arterials. Again, this might be an undesired side effect of tolling all freeway capacity.

Long-Distance Trips

Existing HOT lanes, like the express lanes on the 110 and the 10, are managed dynamically: prices are adjusted to respond to real-time traffic conditions. If the lane starts to get congested, prices are increased to reduce the number of drivers that decide to enter. Pricing information is conveyed to drivers using variable message signs. If you’re already in the lane, the price you saw when you entered is honored for your destination.

This works well for a managed HOT lane in isolation; no one knows what the toll will be when they enter the freeway, so the general purpose lanes just soak up whatever traffic doesn’t want to use the HOT lane. With a network of HOT lanes, this will still work pretty well. The number of destinations you can reasonably indicate on a VMS sign is limited, but you’d always have the option to leave when you reach the next tolling section. Let’s say you’re in the HOT lane on the 10 east and you hop on the 5 south to go visit the mouse, and you don’t like the prices. No problem, you just take the free lanes.

If the entire freeway is dynamically tolled, this starts to fall apart. What do I do if I get on a freeway and I’m not willing to pay the going price? For short trips, you could check before you leave, but for long trips, it would be an issue. If you get on the 101 in Woodland Hills and you’re going to Anaheim, what happens if you get on the 5 and the toll is more than you’re willing to pay? Do you take arterials? Do you just get off and park somewhere, waiting for prices to go down?

Private Parts

Now, you may have been chomping at the bit as you read this post, thinking that there are technological solutions to these problems: use congestion pricing on the arterials as well as the freeways, and quote people a price for their entire trip before they start it.

Those ideas are certainly theoretically possible. However, they may prove politically impossible, for some very good reasons.

Tolling arterial capacity, using existing electronic tolling methods, would prove unreasonably costly. It would more or less require turning every traffic light into a tolling location. It would require trying to communicate toll rates on a block by block basis. Both of these would be impractical. You could do it without any roadside equipment by requiring every vehicle to be equipped with GPS, and having the vehicle’s on-board equipment report the GPS data to a central facility for calculation of tolls.

Getting a price quote for a trip before you take it is something we’re all familiar with for things like flying, ferries, tours, and so on. In the case of flying, the details of your travel are reported to the government in advance. However, flying is something most people do rarely. Requiring advance requests for auto travel fees would bring that level of oversight into people’s everyday lives.

To be blunt, I don’t think many people would be comfortable with having to tell the government where they’re going before they leave, and I don’t think many people want their movements being tracked by GPS. If you don’t like the NSA recording your phone calls and reading your emails, you should be worried about the prospect of having the government follow your whereabouts. While this would obviously still leave walking, biking, and transit as options for anonymous travel, it would be an imposition on people’s right to freedom of movement.


This isn’t to say we should give up on the idea of tolling highway capacity. I would be curious to see research on detailed modeling of a real road network (freeways and arterials) under these scenarios. For example, what would happen on the Westside if the 405 and the 10 were dynamically tolled but the arterials were still free? Regarding privacy, would people be more comfortable if the advance price was obtained through a third-party intermediary (such a car-sharing service) that could make the reservation with the system in the corporation’s name?

In the meantime, a more realistic option than real-time dynamic pricing might be managing freeway capacity the way that street parking is managed in downtown LA. In that model, utilization of street parking is monitored, and then prices at different times of day are adjusted up or down to try to optimize utilization. For freeways, a schedule of prices could be published and updated every month, so that users would be able to determine prices before they leave.  For example, say that in August 2014 it costs $0.25 to go from La Cienega to Robertson on the 10 on weekdays at 12:30pm, and the level of congestion is still too high. The rate for September would be increased to $0.30 or $0.35.

In the case of capacity mismatches, it might be desirable to deliberately underprice freeway capacity so that the amount of traffic diverted to arterials isn’t too large. Many people would rather have a queue of cars on the freeway, leaving arterials a little less congested and available for things like local trips and emergency vehicles.

Congestion pricing has great potential to improve mobility in urban regions. But the devil’s in the details, and we don’t have them worked out just yet.


8 thoughts on “Congestion Pricing Questions

  1. D. Malcolm Carson

    Great article. I’d like to suggest another alternative: no tolls on the freeways themselves, but dynamically toll the on-ramps. There’s a network of 1400 freeway ramp metering lights in Caltrans District 7, nearly 1000 of which are remotely controllable in real time. At present, District 7 doesn’t do anything with that network to reduce congestion on the freeways because of political concerns, except in response to emergencies. But as you demonstrated in this post, anyone with Google maps on their phone can see where congestion is building up in the freeway network, in real time. LADOT has already put into a place a real time dynamic system of controlling stoplights on arterial streets, and it actually does a pretty good job of regulating the massive crush of traffic that they handle. It would be significantly less complicated for Caltrans to do the same thing with the ramp metering lights such that the freeways are treated as a coherent whole system that should always be kept humming at the optimal level (+-50 mph). Having done that, it would be a simple next step to then toll the places where demand to access the system outstrips the supply of ramp space to accommodate it. In many places there is a metering light and a carpool lane also. Converting those to ExpressLanes (again still available to carpools and transit) would open up still more opportunities to use pricing as an incentive to reduce congestion.

    Yes, in some cases, there will be more traffic on the streets, but logically, the freeway is not the place to store excess volume in the system. The freeway at 50 mph is capable of moving far more cars and people per hour than an arterial street no matter how free of congestion. The best place to store excess volume is obviously on the ramp itself, to the extent that there’s space there. Second best, on the streets. Worst on the freeway. Also, slow traffic on city streets makes them safer for bikes and pedestrians, and better for local businesses. But again, if you introduce tolling and HOT lanes in those places where the ramp capacity is insufficient, then you cut down on the effect on the immediately adjacent streets.

    So the system would look like this: (1) at the first sign of congestion (Google map yellow) in any particular segment, all ramp meters upstream would be slowed down until free flowing traffic (green) was restored. If the congestion reached “red” levels, then upstream slow down would become even more aggressive; (2) as long as there is sufficient ramp capacity at a given location, people could wait there to enter the freeway for free; (3) once the limit of ramp capacity was reached (all ramps already have loop detectors installed that indicate when the ramp is full), then tolling would kick in and the price would be set high enough such that the ramp would be near full but not past capacity; (4) depending on the capacity and congestion levels at each ramp, lanes might shift back and forth between 3+ carpools and transit only, HOT, and general traffic; start with the most congested points in the system and then work backwards from there.

    1. anonymouse

      Traffic on the streets gets in the way of local bus service, pedestrians, and general street life/local transportation. As a non-driver, I pretty much don’t care at all what’s going on on the freeway, especially if the few freeway buses have their own congestion-free lanes. Really, the best place to store excess cars during periods of congestion is people’s driveways and garages, or better yet, car dealer lots or car factories. But failing that, using car-only space for excess car storage is better than using surface streets.

  2. Pingback: Today’s Headlines | Streetsblog Los Angeles

  3. RandomSchmoe

    These are interesting ideas, but i reject toll lanes entirely. I would rather have traffic stay as bad as it is than contribute further to making the US even more unequal and unfair. And, we can stop building new housing, too.
    As it is, I don’t see myself ever voting for another MTA bond in my life, I am so offended by what they have done on the 110. And now Caltrans too. And I never miss an election.
    But it’s nice that you come up with all these fun little theories.

  4. Zmapper

    For sake of argument I’m assuming a paradigm that Interstates should be paid from user-fees due to their nature as a competing club (uncongested) or private (congested) good, while local streets should be treated as a public good funded out of general tax revenue due to their local access nature.

    It is likely that some traffic would be diverted in urban areas if the freeways were tolled and the local streets provided as a public good, but as more traffic is inherently local, and as many conflicting variables exist, rural areas provide a functional means to examine tolling.

    Using Nebraska as an example, if the state Department of Roads (yes, it is still named that) were to find a way to toll I-80, many time-insensitive travellers would divert to parallel two-lane roads US-30 and NE-2. At first glance, giving travellers the option of driving a slower-but-free route (assuming that user-fees wouldn’t be feasible due to its dual nature as a local road and cross-country highway) is reasonable. While some time-insensitive travellers would save on the I-80 toll and would be willing to trade time for money, everyone else loses. Residents of the towns now have more congestion and road noise, especially from tractor-trailers, without necessarily gaining any additional customers. As two-lane rural roads and city streets, which now have hurried tractor-trailers mixing with young schoolchildren crossing the street, are noticeably less safe than four-lane Interstates, more collisions and road deaths are likely.

    The state couldn’t ban trucks, as local deliveries and farm movements still would use the two-lane roads, and there is no practical way to determine who is a local and who isn’t without a traffic stop; additionally, the roads in question are federal highways, roads that are by definition designed for interstate commerce. Tolling the parallel highways would be impractical, as the cost of the gantries would be prohibitive, and some traffic would continue to find lower and lower hierarchy roads to use; additionally, some locations can only be accessed by the two-lane roads in question, which conflicts with the stated assumption that local access is a public good.

    A compromise may be to use gantry-based tolling for cars, while requiring tractor-trailers to be tolled by GPS. Many tractor-trailers are already GPS logged for the purpose of ensuring hours-of-service compliance, and privacy objections aren’t an issue due to the vehicle being used for work as interstate commerce. Due to the power-of-four rule of vehicle weight and damage, tolls based in part off of roadway damage would mean that cars would pay comparatively low rates, while tractor-trailers pay steeper tolls. As cars would pay lower rates and thus be less financially tempted to divert off of the Interstate. Many cars are travelling for leisure (road trip) or for non-work purposes, which means that to a reasonable extent they aren’t as utilitarian in their choices as a driver to whom frequent tolls directly impact their profitability. Any cars diverted wouldn’t be as large, noisy, and dangerous as trucks, any may be more likely to make impulse stops at in-town businesses.

    In cities, more traffic is local and regularly occurring, which means they are more likely to use local streets. While freeway traffic would be free-flowing, this comes at the cost of more congestion on local streets, which poses direct impacts to neighborhood road safety, local-stop bus delays, etc.

    1. letsgola Post author

      An interesting case study in this is what actually happened in New Jersey when the 95 freeway was cancelled between Trenton and New Brunswick, leaving the tolled New Jersey Turnpike as the only option. The freeway was cancelled partly due to local opposition and partly because the Turnpike Authority didn’t want its toll revenue jeopardized by a free road nearby (like happens in south Jersey with the 295). However, many trucks started taking the 206, a two-lane rural road through Princeton, rather than paying the toll. The state solved that problem by designating certain routes for interstate trucking, and prohibiting 102″ wide trucks (the common width of tractor/trailer combos) on the 206. That restriction was ruled unconstitutional in 2006, because it impacted interstate commerce, so they went back to the drawing board and came up with rules requiring all trucks to stay on the national network unless making a trip to a terminal, and then, requiring those trips to be by the shortest route. See for details.

      In a place like Nebraska, it’s certainly an issue too, because the time savings from using freeways in rural areas are often pretty small. You can go fast on rural two-laners, which as you say are more dangerous. The penalty for slowing down at rural crossroads isn’t that much.

      Bottom line, freeway tolling proposals will have to find a way to deal with trucks, either through NJ-type regulations or GPS tolling as you suggest.

  5. Devin

    What about tolling (for instance) the 405 through Sepulveda Pass (and probably Sepulveda, for good measure)? It seems like it has more in common with the Golden Gate Bridge (or should that just be “with Golden Gate Bridge”, given how funny SFers talk about freeways?) than it does with the 10 through the westside.

    It seems to me that this is sort of like making parking more expensive to deter more trips and relieve congestion: you have to go through Sepulveda Pass, the alternatives are already jammed and will take ~0 more cars; it’s a natural chokepoint. Toll it, raise some revenue, relieve some congestion, and fund your supersteep light rail or a subway underneath.

    I think similar arguments could be made for the 1, the 101, maybe the 5 btwn Glendale and Burbank, the 210 if enough traffic moves over there. Unfortunately for the valley, that is where most of the nice chokepoints are–but perhaps this can be assuaged by devoting most of the revenue to improving transit/etc there.


Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s