Video by Oran.

It’s sort of obvious that congestion is bad and makes people’s lives worse. But what actually causes congestion? I promise you it isn’t bad drivers. I’m going to try to explain a little bit of the economic theory that shows just how it makes people’s lives worse – known in economics speak as “reducing welfare” – and why charging people to drive makes society better off.


Too many drivers trying to access a roadway at the same time – in economic speak, excess demand for a good in fixed supply – causes congestion. When there’s no traffic, adding new drivers to a roadway does not reduce travel times. However, once a highway becomes congested, adding more cars to the road does not increase the number of cars the road carries (evidence suggests it actually reduces the number of cars through).

Supply on a road, once a road reaches it's maximum capacity, it cannot carry more vehicles

The demand for highway transportation represents the value that consumers place on traveling at a particular time, in particular manner and to their particular places. Some trips are valued very highly – “getting to the airport”, some trips not as much – maybe going to the supermarket across town that has the bread you like. The lower the cost of driving, the more people drive. This means drivers taking longer trips or more trips as costs decrease.

Supply and demand curves for an uncongested road.

When roads are free, the only cost to drive is time. Once a roadway approaches congestion, each additional driver impacts all other drivers, slowing them down and costing them time. This is a negative externality: the cost to the driver (his time) is less than the cost to society at large (his time plus the delay he caused to everyone else).

Supply and demand curves for a congested road. The red area is consumer surplus.

This time sitting in traffic is lost for everyone driving, and is a large net social loss. Everybody places some positive value on their time, so keeping people in their vehicles unnecessarily and away from productive uses (working, playing, exercising, etc) is bad for them. All that wasted time adds up, 4.8 billion hours in the US last year.

Congestion pricing solves this problem. You put tolls in place and time is exchanged for money. People who value their trip less than the toll price won’t make the trip, which is an increase in welfare since the price they were paying before was less than the price to society. A nice side-effect is that the money is not lost to society (unlike the time spent in traffic), the money could then be put to whatever productive use people can think of.

Supply and demand curves for a road with a congestion charge. As the charge gets larger, fewer people drive and the road becomes uncongested. Note the red surplus area is larger than in the congested case.

Congestion pricing would also be good for transit. Bus service is priced in hours, not distance, so reducing the amount of time buses spend in traffic would reduce the cost of operating the same level of service. It’s also possible to spend some of the money on new transit service or construction, something transit advocates should be in favor of.

Unfortunately, it seems more likely we’ll be passing laws that make congestion pricing illegal than passing laws putting it into effect. Part of this reluctance is a sort of status quo bias – why should I pay for something that was previously free? Part of it is also just a lack of understanding of how congestion works: I’d bet a sizable minority of people think congestion is caused by bad drivers. It’s all  really a shame, because congestion pricing works, the much needed revenue could plug huge wholes in our transportation budgets and when people are stuck in traffic no one benefits.

70 Replies to “Congestion is Welfare Reducing”

  1. It is also human nature to avoid tolls, fees, strangers in your car to form a car pool or anything that smacks of extra effort even while wasting effort to do so.

    Thus, people will avoid a toll road, a hot lane, or drive around a humongous parking lot burning fuel just to get that spot 100 feet from the entrance. It is only when they are presented with a stark choice of being late for a date/game/something really important will they consider alternatives. Look how empty HOV lanes are on 405 when the rest of it is at a standstill.

    Sometimes choices are not rational. But they are the choices humans make.

    1. Sometimes choices are not rational. But they are the choices humans make.

      Even more reason to help them make better decisions.

      1. Better decisions based on what? Your opinion? The only thing this will do is guarantee roads are full more hours of the day.

      1. I thought my description of hunting for a parking space close by the door would suffice. Sorry if it didn’t.

    2. Not sure what part of 405 you drive, but I used to drive southbound every day from Bellevue to Renton and the HOV lanes were always backed up during afternoon rush hour. I always thought they should make 405 HOV lanes 3+.

      1. I thought my description of hunting for a parking space close by the door would suffice.”

        it just sounds like finding the most desirable produce in the produce section. i dont see a waste.

  2. Some congestion IS caused by bad driver, Obviously not all traffic, but enough of it to make both over-capacity, and bad-driving an issue.

    Look for instance at the difference between Chicago and Washington DC. They both ranked for worst traffic congestion in the nation with 70 hours of average delay in both cities, however, Chicago has 5 times as many people as DC (city proper)(Metro to Metro, Chicago is twice as big as DC), and less highways per capita, so somehow Chicago is making FAR better use of those roads than DC is. Admittedly, about 1 million more people per day ride transit in Chicago than in DC
    , but I suspect there is a whole host of other things that contribute to this, including bad driving. Most people would call chicago drivers horrible, but I would call them extremely efficient.

    I understand that yes, plenty of congestion is due to over-capacity roads, but at the same time, that is not the ONLY cause of congestion as you make it sound above. Bad driving causes a sizeable enough amount of congestion to be an issue as well.

    1. First, I don’ think this is right:

      (Metro to Metro, Chicago is twice as big as DC)

      http://en.wikipedia.org/wiki/Table_of_United_States_Combined_Statistical_Areas

      Chicago Metro: 9,804,845
      Washington DC Metro: 8,440,617
      That’s a 16% difference. I’m also not sure where you get the less highways per capita thing, but I’m not sure it matters.

      I don’t know about transit in general, but the DC Metro gets about 50% more riders than the Chicago ‘L’
      http://en.wikipedia.org/wiki/List_of_North_American_rapid_transit_systems_by_ridership
      DC Metro: 919,300
      Chicago L: 638,100

      Anyway, if it were true, that Chicago makes better use of its roads (which could be lots of things including commute patterns, etc.), I don’t see how it follows that this means the drivers are necessarily at fault. Maybe it’s planners? Land Use?

      It’s not drivers, except in the case of accidents.

      1. http://en.wikipedia.org/wiki/Washington_Metro
        590,000 daily rail ridership
        http://en.wikipedia.org/wiki/Washington_Metropolitan_Area_Transit_Authority#Metrobus
        400,000 bus
        http://en.wikipedia.org/wiki/MARC_Train
        33,000

        http://www.transitchicago.com/assets/1/ridership_reports/2010-Annual.pdf
        Rail 671,000
        Bus: 973,000
        http://en.wikipedia.org/wiki/Metra
        Metra:322,000

        Chicago: ~2,000,000
        DC: ~1,000,000

        And I read somewhere yesterday that the DC metro was 5.4 million people, though I don’t have a link to that so it may be wrong.

        AS to less per capita roads, this is what my professor told me in one of the classes I took last semester, that Chicago has the least highway miles per capita of any major city in the nation (Kansas City had the most, with minneapolis up near it)

        based on this logically roads in Chicago should experience much more congestion due to being over capacity. yet they don’t. this leads me to believe that among a host of other things that affect congestion, that DC drivers don’t utilize the space as well.

        I know from experience that drivers in chicago will drive with about a second or two of following distance at 80 mph, which while it may be “un-safe” (which driving is in general…) increases capacity.

        I also know that Chicago’s congestion stems from what I consider to be very poor highway interchange design. Dumping two 3-lane freeways together into a 4-lane freeway is stupid. (heading towards downtown)

      2. There are like ten different ridership numbers is that post, 590K is the lowest, 727,684 is the middle and 919K is the highest.

      3. Chicago has extremely wide roads, even the minor streets. *Miles* of roadway doesn’t measure it at *all*. Just so you know.

      4. Further regarding Chicago freeways, having an entrance and an exit every two blocks is generally considered terrible design nowadays, but that’s what they do downtown….

        The lack of congestion is largely due to people taking the ordinary (non-freeway) roads, which rarely end up completely congested (unlike NYC). The freeways are frequently completely blocked.

    2. The only bad driving that causes congestion is someone tailgating and then slamming their brakes. This causes a quicker brake wave or shockwave or traffic wave or whatever you want to call it. The “phantom traffic jams” that seem to happen for no reason. Japanese scientists recreated a traffic jam, you can search youtube for ‘shockwave traffic jam’.

  3. Good overview. You get some really weird answers when you ask people what causes congestion. One guy told me it was curves in the road.

    1. The guy was partly right obstacles, like sharp curves in the road do cause drivers to reduce their speed which causes bunching which can escalate into stopped traffic and congestion.

      1. obstacles … which causes bunching which can escalate into stopped traffic and congestion.

        Obstacles are reducing the capacity, which makes congestion easier, but it’s not the cause of congestion.

  4. Joshua – congestion is caused (or at least exacerbated) by curves in the road. A good example of this is the curves on 520 at Portage Bay leading down to the floating bridge (traveling east this effect can be observed quite handily.) Even if traffic appears to be flowing at something close to the speed limit, it often slows to a crawl and sometimes a halt just short of those curves, because many drivers brake when they don’t have to, causing a ripple effect backward as the whole herd brakes. Thus even if there’s nothing holding up traffic ahead of the curves, everyone behind the curves suffers increased congestion which can take some time to untangle.

    Only roads well under capacity do not suffer increased congestion as a result of total stop incidents forward of any given point. While it’s maybe not a good idea to make a value judgement as to whether people are good drivers or not, there is certainly an argument to be made that driving in congestion is a different exercise than driving on a clear road, and that road capacity would probably increase overall if everyone on congested roadways drove with an eye toward eliminating the number of times anyone feels a need to slam on their brakes.

    1. It is true that bad roads have slower speeds. A curvy mountain road is going to have a slower overall speed than a straight highway – or at least it should if you want to live. But that’s not what we’re talking about when we say “congestion”.

      1. Just to be clear in this post when I talk about congestion I’m talking about this: congestion starts at the point when introducing more cars to the road does not increase the flow at the either end.

    2. I’m not talking about speed when I talk about congestion. I’m talking about smoothness of traffic flow adjusted for the safe speed vis-a-vis the conditions present. Drivers currently do not adjust their expectations of the speed they are able to drive on the roadway based on the congestion conditions. This causes them to try to go faster than it is possible to go and that causes them to slam on their brakes, causing the people behind them to slam on their brakes. This effect ripples backward and also causes other problems to crop up for people trying to merge. Some work has been done in the UK with variable speed limit signs and lane clearing that has helped in this regard, though it is not a panacea.

    3. Only roads well under capacity do not suffer increased congestion as a result of total stop incidents forward of any given point.

      are you describing a congested road or poor driving habits?? they seem quite different to me.

      i though tcongestion was more cars on a road than the road was designed to handle at speed limit or even near speed limit?? a physical space issue, iow.

      1. I am describing both a congested road and poor driving habits. I am not arguing against pricing people off congested roadways, that’s for politicians to figure out. The issue is the assumption that the speed limit is the important factor in a roadway. It doesn’t matter what speed the roadway was designed for at what capacity. If today there are more people on the roadway than it was designed to handle at the speed on the sign, the speed on the sign needs to change to reflect the conditions. A roadway designed for a given volume of cars moving at 60 mph which has say 6x the cars can’t flow smoothly at 40mph, but it can flow rather smoothly at 20mph. I know I’m not that crazy about driving 20 mph on the freeway but after having had hellish commutes caused by people effectively taking the operating speed of the roadway down to 5-7 mph, mostly as a result of poor driving habits, I’ll take what I can get.

  5. Yes, congestion pricing should also have applications in transit. One that comes to mind is that using a method of fare payment that causes congestion in a bus route should cost a premium charge (e.g. no free transfer).

    This is also a social justice issue. The revenue from our regressive tax system is being used to subsidize individuals’ lack of effort to get an ORCA card. If they want one, they can get it. The best marketing Metro could do is to simply cut off the free transfers, and make it worth people’s while to get the card. Sans “congestion pricing”, many won’t bother, and will continue to take advantage of the de facto longer transfer time.

    Metro, show us you can at least pick the low-hanging fruit!

      1. No, just the ones who don’t have an ORCA card. I meant to say “free paper transfers.” Sorry.

        As it is, ORCA transfers aren’t worth as much as paper transfers (except for inter-system multi-seat rides), so the system discriminates against those who have made the effort to help make the system faster and cheaper.

  6. The problem is that your congestion plan is worst for the poorest people. Society isn’t a person! Just because it makes society better off doesn’t mean it makes each person better off, and you basically want to charge poor people off the road.

    1. You can solve the social inequity problem by providing a working tax refund where poor people get a lump sum, ie a re-distribution.

      And you could offer HOT lanes priced on income levels, such that a rich person pays more for the privilege. But then you have to watch for scams where the car is registered in the hired driver’s name/income. And you have to deal with the people who don’t believe in graduated tax schemes.

      1. It might be easier to simply dictate that everyone makes the same income regardless of their economic output. That’s simpler than adjusting prices according to income.

      2. Har har.

        Seriously, though, taxes/fees generally have two goals. The first is to raise money for the government with the minimum distortionary effects. In other words, no person should be relatively better or worse off after the tax than before, and no activity should be more or less desirable. And the second is to encourage or discourage certain behaviors which are socially desirable or undesirable, by providing incentives to engage in the good behavior, and disincentives for engaging in the bad.

        The first goal is why most taxes are percentage-based, rather than head taxes. The federal budget is something like $4 trillion, which works out to about $13,000 per person. It goes almost without saying that it would be ridiculous to charge every person $13,000 a year in taxes. For some people, this would be devastating. For others, it would be a drop in the bucket. Thus, such a policy would heavily favor rich people at the expense of poor people, and that kind of distortionary effect is something we try hard to avoid.

        Percentage-based taxes are better, but a flat percentage still isn’t optimal. If someone earning $20,000/year had to pay 20% in taxes, that would hurt a lot more than if someone who earned $2,000,000/year had to pay the same rate. $20,000 is hard to live on already, and $16,000 is even worse, but $1,600,000 would be just fine for almost anyone. Thus, many taxes have graduated rates, so that as the marginal utility of extra income goes down, the extra income is taxed more highly.

        The exact same reasoning applies for congestion pricing. In the absence of government intervention, some number of people would drive. Some percentage would be rich, and some percentage would be poor. If the government does intervene, they have the potential to create a large distortionary effect. A flat rate, or even a flat percentage, will mean that a higher proportion of drivers will be rich. A more graduated tax will have less of a distortionary effect, such that the ratio of rich to poor drivers will be the same after the tax as it was before. That seems like a worthy goal — at least, I can’t think of any legitimate reason why the government would want to more strongly discourage poor people from driving than rich people.

      3. Milton Friedman proposed a Guaranteed Minimum Income, like the Alaska Permanent Fund dividend, where everyone simply gets sent a good-sized check by the government every year, enough to maintain a bare-minimum standard of living. This keeps everyone out of poverty to the point where they can’t legitimately complain about little things like the gas price (so that can be used to discourage dangerous pollution), while still insuring that people who work more and do more valuable things are the ones who can buy more luxuries (like nice clothes, good food, televisions, etc….)

        Of course that would make too much sense — there is a large faction of this country which seems to feel that poor people need to be punished, or *forced* to work even if they are crippled, and that they should work as hard as possible even if the work doesn’t pay enough to live on. I don’t know what to attribute this attitude to — Calvinism? Residual support for slavery? Simple meanness?

      1. If someone earning $20,000/year had to pay 20% in taxes, that would hurt a lot more than if someone who earned $2,000,000/year had to pay the same rate. ………….

        would it be more likely that they wouldnt have an income at all??

    2. Not really, Norman. Many of those of us who are the “poorest” are those who are transit-dependent and don’t own a car. Often, we live in neglected parts of the city or unincorporated areas near the city; we are thus much more affected by congestion and lack of transit service than those who can afford to drive.

  7. I agree strongly with the economic argument listed. However, I also understand why there is such strong opposition to this kind of pricing system.

    Consider that for someone who commutes to and from work each day, Monday-Friday, even a $1 cost each-way adds up to $500/year (that not even including non-work-related trips). Something like 520 where the cost is $2-3 each way, this can add up to over $1500/year. Consider that for families that earn less than 30k/year, and pay most of that just to cover their living costs, $500-$1500 can signifigantly cut into their liquid income. While one might argue that they are then gaining some value from reduced commute times, it still doesn’t improve a family’s bottom-line budget unless those individuals pick-up a 2nd job or have the flexibility to work more hours.

    Economically speaking, a “perfect” system would reimburse everyone up-front for their expected commute cost. For example, suppose a congestion pricing system were put in place, and suppose that with my commute it would cost me $2000/year. If the government put the system in-place, then cut me a check for $2000 and let me do whatever I wanted with it, I could continue to commute as before an would be no worse off (and thus would not oppose the plan). However, I now also have the option of keeping the $2000 to spend on other things if I find an alternative way of commuting; making me more likely to do this.

    A congestion pricing system WILL improve transportation, independed of where the revenue generated is allocated. In my oppinion, the only which a congestion pricing system would be politically feasible is if the revenue is then spent on reimbursing the “losers” of the system; decreasing the financial impact it will have on them. Admitedly, doing this can be very complex and is not an easy thing to do. However, I believe that this is the key problem which must be solved in order to make congestion procing feasible.

    1. Couple things.
      1) For families earning $30K or less per year, a car is a luxury they shouldn’t need to have. We need to do a better job of providing low-cost transit options for them.

      2) There are lots of things you could do with the money, but I disagree that we have to reimburse the “losers”. In fact, progressive taxation is not especially popular right now.

      1. For families earning $30K or less per year, a car is a luxury they shouldn’t need to have……

        does anyone say that it is a luxury they need to have?? and mean it??

      2. “does anyone say that it is a luxury they need to have?? and mean it??”

        They don’t say it but our de facto lack of transit implies it. My roommate works part time at a warehouse in Kent with start times at 6pm, 9pm, and 4am depending on the week. He wants to get a Seattle job where he can ride transit to work but he’s been applying and there haven’t been a lot of offers in this economy. He’s also a firefighter but those jobs are even farther out: Black Diamond, Pierce County, etc. And he’s a reservist so he goes to Ft Lewis or Everett once a month. There are no transit options for these except ridiculous ones like walking 30 or 45 minutes to the 150 or 280 (Renton night owl). And there’s no reason to move to Kent because he doesn’t want to live there and doesn’t want to be working there long term, and there aren’t any houses near the office parks anyway.

      3. “what percentage of families earn less than 3ok per year??”

        Given that the national median income is $35K, a little less than half of all families in the US earn less than 20K per year.

    2. Those who are more affected by the cost would have a lot more incentive to carpool. That would hep them save money on transportation and help ease the congestion at the same time.

  8. When roads are free, the only cost to drive is time. Once a roadway approaches congestion, each additional driver impacts all other drivers, slowing them down and costing them time. This is a negative externality:

    are you calling something a negative externality when comparing it to a non-existent positive??

      1. i guess. have there been long held economic beliefs that tuned out to be wrong??

        it seems to me that 7000 people per square mile with most having cars isnt a mystery and people still flock to the cities….with cars. the known ‘congestion’ must not be too much of a negative because more and more cars end up on the roads??

        i dont think you can fold up the roads during slow times and bring more out when needed like tables at a banquet.

        a train is designed to stay on tracks. if it wrecked on property it was allowed to pass thru, that would seem like a negative externality…a city is ‘congested’ by nature. its a large city.

      2. the known ‘congestion’ must not be too much of a negative because more and more cars end up on the roads??

        Scott, the whole point of the post is that people don’t take into consideration their effects on other people when driving (why should they?) so they make congestion worse.

        especially if you have no other option than to drive.

      1. Free to the user/ decision-maker. Not free to the producer. That’s a decent alternative definition of a negative externality, actually.

  9. : I’d bet a sizable minority of people think congestion is caused by bad drivers. ”

    some risk. a sizable minority??

  10. when people are stuck in traffic no one benefits.”

    stuck isnt permanent is it?? i guess there is a benefit to some to creep along in congested roads twice a day and to zip along pretty good for several other hours of the day.

    to me city driving is tedious at times.

  11. Congestion is Welfare Reducing”

    i dont see where congestion has ant relation to welfare?

    It’s sort of obvious that congestion is bad and makes people’s lives worse.”

    why do people flock to cities?

  12. Sorry, Andrew.

    Failure to pull into the intersection while awaiting the chance to turn left or right — thus blocking the bus behind you — is bad driving.

    Fairure to anticipate and smoothly negotiate merger geometry — such that it can take 10 minutes to clear the Ballard Bridge’s right lane (left lane is totally clear, but that doesn’t help the bus) — is bad driving.

    Thoughless driving may not be the root cause of congestion, but it most certainly makes it worse!

    1. “Failure to pull into the intersection while awaiting the chance to turn left or right — thus blocking the bus behind you — is bad driving.”

      If you cannot clear the intersection before the light turns red, you should not enter the intersection. If you block the intersection, that is bad driving.

      I agree with your point on merging (in general, not restricted to helping out the bus). To me, it seems like a lot of traffic delays are exacerbated by selfish folks who don’t like to queue up.

      1. Wrong, wrong, wrong, wrong, wrong.

        If cars pull out into the middle of the intersection, the chance of making the turn that cycle increases to 100%. Often well before the light changes (less remaining distance to negotiate during a gap in the counterflow). Otherwise, when the light changes to red, and perhaps even for one more car behind you, which would be impossible had you dawdled behind the line.

        Sorry, AW. This one is non-negotiable. East Coasters make this work millions of times a day. Seattle drivers have this wrong, and contribute significantly to back-ups (especially of buses behind them) as a result.

      2. Also, my point about the Ballard Bridge merger was that people coming down 15th are ostensibly too dumb to get in the left lane when they see the bridge going up. So then the right lane, while merges from literally four different directions, takes forever to clear up, while the left lane is smooth sailing.

        It’s not that Seattle drivers have a slow learning curve. They have no learning curve at all!

      3. One car could pull into the intersection; it doesn’t help a following bus unless they can clear the intersection before the light turns red.

        I’ve driven in Boston often enough that I know that east coast drivers don’t drive better than we do, they just drive badly in different ways.

      4. I’m pretty sure that (less asphalt per lane) + (still getting places faster) + (fewer accidents) + (lower rates of pedestrian fatalities) => Boston driving is better than Seattle driving.

        “It doesn’t help a following bus unless they can clear the intersection before the light turns red.”

        I’ve been on buses that spend entire light cycles unable to squeeze between a parked car and a reticent turn-attempter with its blinker on. If that car were out in the intersection, the bus could have sailed right around it. Trip time saved: 60+ seconds (often multiple times per trip).

      5. “If you cannot clear the intersection before the light turns red, you should not enter the intersection. If you block the intersection, that is bad driving.”

        Wrong. You are not supposed to pull out unless you can clear the intersection *if you are going straight ahead*.

        When turning left at an intersection without a left-turn cycle, you are *required* to pull out into the intersection and wait for a gap in the oncoming traffic. If one does not appear before the light turns red, you turn left on the red cycle before the cars who now have a green light go.

        This guarantees that one left turn in each direction will be possible with each light cycle. This is not possible without left-turners pulling out into the intersection. Right-of-way law specifies that the vehicle within the intersection has the right-of-way over any vehicle outside the intersection, and makes leaving the intersection legal when the light is red. The traffic laws were designed carefully. :-)

  13. I have to agree with you on congestion pricing. Look at it this way — from about 9pm to 5am the roads are empty! While it’s extreme to think that people would shift to working as night owls, one can imagine a worker, with no kids, who might want to leave at 5pm but then think, well, if I wait, I can drive home for 75 cents.

    The important thing is that the carrying capacity be there if someone is willing to pay for it. I would say we have to extirpate the “War on Cars” culture from Olympia and substitute a market driven and resource based system that recognizes that along with automobile costs, there are also conveniences and yes, efficiences, in their use!

  14. Sorry, I didn’t get to read this till today (I was stuck in traffic!).

    It occurs to me that the author’s definition of congestion is about the same as the definition of inflation, but measured in terms of time as well as money.

Comments are closed.