There has been quite a lot of discussion recently about the “cost” of congestion, as well as pretty dire future projections in relation to levels of congestion 20-30 years in the future. But, compared to other elements of transport costs, how does congestion actually stack up? Todd Littman takes a look at this in a really comprehensive study that has been summarised on the Pricetags blog:

Comparing Congestion With Other Costs

The UMR report claims that traffic congestion wastes “massive” amounts of time and money, estimated at 5.5 billion hours and 2.9 billion gallons of fuel, worth an estimated $121 billion.

Described this way the costs seem very large, but measured per capita they appear more modest: 17 hours, 9 gallons and $388 per year, or less than three minutes, 0.03 gallons and $1.06 per day. These represent less than 2% of total travel time and fuel costs, which is small compared with other factors that affect per capita travel time and fuel consumption costs.

This indicates that congestion is overall a modest cost, larger than some but smaller than others:


Because congestion is just one of many costs, it is inappropriate to evaluate congestion reduction strategies in isolation: a congestion reduction strategy may provide far less total benefit if it increases other costs, and is worth far more if it reduces other costs or provides other benefits.

For example, roadway expansions may seem cost effective considering just congestion impacts, but not if wider roads induce additional vehicle travel which increases other external costs. Conversely, improving alternative modes may not be cost effective based only on their congestion reductions, but are cost effective overall when co-benefits (parking cost savings, traffic safety, or improved mobility for non-drivers, etc.) are also considered.

Somehow we need to get past using congestion as the primary success/failure measurement for transport. After all, not only is it seemingly relatively minor in the scheme of things, but there’s a fairly good argument out there that congestion doesn’t hold back economic productivity:delay-vs-gdpUltimately it’s hard to work out what the real cost of congestion is, but I’m pretty sure it’s WAY less than we assume.

Share this


  1. It’s way less and also irrelevant, all the German cities I have lived in had high levels of ‘congestion’ but only of motor vehicles, however, it really was of no concern to most people as reliable public transit made the argument moot, as did the ease of cycling. It’s the lack of any feasible options for almost all journeys in NZ that make congestion such an obsession, and the problem that when roads are congested it impacts the buses which have no bus lanes and cyclists whom also have no lanes. The solution, separate out cars which will always be clogged from other modes to allow us to choose whether we want to sit in traffic or if we want to take a reliable congestion free alternative.

    1. Exactly, bbc, this is what is missed when the transport tsars only consider cars as a valid mode of transport for productive citizens.

      Road congestion actually doesn’t stop people getting around by other means, and left to itself, road congestion will increase to some level at which it annoys some drivers enough that they opt for other means.

  2. The standard idea of congestion is the road lovers best friend. They use it to justify more spending on more roads that create ever more congestion: perfect circle.

    At every opportunity the minister and all the figures form the highway/auto complex reinforce that congestion is the only measure that matters and that only by building more roads can it be addressed. This has been going on for 60 years in Auckland and congestion just gets worse, and now is projected to determinate further despite billions more worth of road projects.

    Neither point is true. Accessibility, regardless of mode is the real issue, and road congestion cannot be ‘solved’ by reinforcing auto-dependancy, quite the reverse.

  3. And here is economist Paul Krugman on the issue:


    the author argues that mass transit has a significant impact in reducing traffic congestion, even when it carries only a small fraction of commuters. Why? Because commuters who take mass transit are, very disproportionately, people who would otherwise be driving on the most congested routes. So even the small number of people taken off the roads has a surprisingly large effect in reducing travel delays.

  4. The Texas Transportation Institute report you’ve linked to is about ‘congestion costs’ defined as ‘costs [time, vehicle operating costs etc] experienced by motorists in excess of what they would experience on free flowing roads’.

    Mildly interesting information perhaps, but it’s unclear how it’s relevant to public policy. In a budget constrained world, surely the relevant concept is ‘the economically optimal level of congestion’ – the level where the cost of further mitigation exceeds the benefits.

    The report’s ‘strategies for relieving congestion don’t discuss the cost of the solutions, which means it’s pretty useless in addressing the question of how much should be spent to reach the economically optimal congestion level. It does mention densification to bring destinations closer (although it regrettably fails to distinguish mobility and access**).

    Interestingly the report estimates that the congestion reduction benefit, from the use of public transport, in 498 US urban areas, is worth about 40c per passenger mile on PT. I hazard a guess that that could be quite a significant amount in context of the per passenger mile PT operating cost (though the report itself doesn’t discuss that detail)

    ** The point here is that it would be courageous to expect densification to reduce traffic congestion: the hoped for extra walking/ cycling/ public transport use is unlikely fully to compensate for having less road space per person. But if it brings people closer to the places they want to go, to a degree that outweighs the slower travel, it is still a benefit.

  5. Congestion reduction planning is an exercise doomed to fail as its an exercise in planning for removing of local peaks not averages.
    And you can never remove all the peaks, only smooth them out.

    The congestion measurements used now all seek to remove congestion at the peak due to its perceived damaging effect on journey times .
    This makes for a very expensive fix as you have to over-build to cater for the peak.
    When you instead look at the average journey times and design to minimise that you find a lower level of engineering is needed to give the majority the same or slightly longer journey time at peak. But at the same time you are maximising the throughput overall.

    This ties in with Patricks comments and those from Krugman about the “marginal” effect moving a few car users onto PT at peak can have, as this improves the average for everyone, which is a much bigger impact than reducing the peak for a few.

    And also ties in what I said before in the post about AA increasing speeds to 110kph – about treating the system like a pipe and looking at total flow of the liquid (or gas) in the pipe, not how fast one molecule of gas/liquid moves through the pipe.
    By doing this you will find the overall flow can be quite “fast” – even when the speed of the molecules in the pipe is not.

    So what we need is more haste, less speed.

    1. Where do you get that from? I work in the highway design and planing industry and we never aim to totally remove congestion during peak time.

      Also of note is that for any given arterial road in Auckland. Some 50% of the users are on it during the peak 4 hours.

      1. “for any given arterial road in Auckland. Some 50% of the users are on it during the peak 4 hours.”

        By “users” you pretty much mean cars right? (1 user = 1 car) or are you counting say a bus load of 40 PT users as 1 “user” (i.e. vehicle) or 40 actual people/users?

        Regardless, by that metric we spend at least half the money on roads simply catering for a 2 hour AM and PM “peak”.
        And in fact if you were to add in the shoulder hour to the AM PM peaks (e.g. 3 hours AM and 3 hours PM peak+shoulder hour) you’d probably get up to 60+% of the usage in a day for a given road occurs in those 6 hours?

        But lets be generous and say only 50% (not 60%) of the total road cost is the amount required to cater for the peak periods.
        But even so that 50% is needed for having to over build the roads to cope with the peak.

        If thats not over-designing a road to “minimise” congestion at peak times what is?

        Sure, even with that, individuals will face congestion on their journeys at peak over non-peak in any case, but 50%+ of the road spend for the peak is quite a penalty.

        And yet you simply can’t build your way out of congestion with more and more roads.

        So what you need to do is use the roads you have more efficiently, one of the most obvious ways is (a) use another mode instead and/or (b) if you have to use the road, maximise the people using the road space – a full bus being the current best way to do that.

        I was listening to a recent BBC “What if” series, radio documentary called “What if… Everyone had a car?”.
        [the solution they explored was to make cars physically smaller and automate the driving process with driverless cars, not remove the need to have one in the first place].

        But during that documentary a traffic researcher said that on average about 25% of the road surface is actually taken up by the vehicles travelling on it at any point in time. i.e. 75% of the road space is “free” of vehicles – this is to allow for braking/slowing down, speeding up, lane changing and the general fact that humans are not very good at driving = so we need a lot of room around the vehicle to manage the job.

        What this also says though, is that even with the physically smaller cars and the best automation in the world, you could only get a 4x improvement in road use by reducing vehicle size, fully automating driving completely etc so that there was no need for that 75% currently “wasted” road space.
        And of course, thats a theoretical maximum and not possible right now.

        To look at it another way – you could in theory make a 1 lane road perform like a 4 lane road – would that reduction of space allocated to roads ever happen? I doubt it.

        A single rail line can replace 10 lanes of cars right now – and thats do-able with currently available tech e.g. EMUs
        – not some 20+ years pie in the sky, for a full computer controlled driverless system.

        Yes we do overbuild the railway system to cater for the AM/PM peaks (by having more trains and designing for closer train spacing with the signal system).
        But thats more to do with a current side-effect of the crappy system we have now and the vast difference in frequency between peak and off-peak train services. In the near future with train services more frequent all day every day, the difference in utilization between peak and off-peak trains will be much reduced than it is now.

        Roads – not so much, if at all.

        So do we continue to throw 50% of our road spend at dealing with peaks or do we try a different approach?

        “we never aim to totally remove congestion during peak”

        You and your colleagues might not – but your political masters can and do.
        And they make a big thing about saying to all and sundry thats what they’re doing/going to do – and they put the money up to allow you to make it so.
        [Witness Brownlee and the RoNS as a classic example of this from the Governent, but its not the only example].

        And would you say you’ve never worked on a roading project to “reduce congestion” – for which a better solution than more roads was needed and was available – yet the bosses demanded a road based solution?

        1. Your arguments for peak usage could be made for any transport mode. For instance why should we double track rail lines when they are empty most of the time? and why have so many buses trains and ferries when again they are almost empty most of the time?

  6. And what is compelling about the chart above is that it shows that high movement cost burden is really due to auto dependency itself, not just the one issue of congestion. Clear as a bell.

  7. I see congestion as more a measure of ease of travel. So although on average we may suffer 3mins of congestion a day, that’s 21mins a week that may very well be the one and only trip we made by car.

    After suffering those 21mins, or whatever, we may decide that we won’t make that trip at all. We pretty much see this in every big city with people deciding not to make trips or changing their plans solely based on the difficulty of travel.

    1. And is this really such a big deal? What is important is that there are other options available. Unfortunately the myth that bufing roading capacity, unblocking those congestion points has meant that we have put pretty much all our eggs in one basket (it is Easter after all)… And in such a way that cycling a d walking, the cheap modes are also extremely unwelcoming and feel unsafe. If Auckland started to decouple “congestion & e onomic growth” from consideration then perhaps there might be a greater chance of getting a framework that prioritises the movement of people as opposed to cars

      1. The big deal is that if people decide not to make a trip is that some economic activity has just been lost.

        We can certainly give people options as there are a number of trip types where people don’t want to take a car, but there are also a number of trip types where people do want to take their car.

        1. While I don’t disagree, the flipside is that arguably with increasing congestion more people will find more ways to make economic use of their time: ordering goods online, using a phone, video conferencing, working from home, changing jobs or work hours. etc. doesn’t work for everyone but we don’t need everyone to work from home.

        2. The same argument could be said to promote downgrading the PT system, which I don’t think is a good idea.

        3. Don’t agree that every trip not made is a loss of economic activity (unless your perspective is that of an oil company).
          Part of the problem is the cars feel cheaper than they really are, so we drive across town to save $10 on that new toaster when it would in fact have been better to buy it locally or on line and pay the delivery fee.

        4. The role of a transport system is not to force people to stay at home and buy toasters online, its meant to be their to provide people with a realistic level of freedom to go about their lives.

          Roads being the most general provision also provide for municipal services and the movement of goods.

        5. Patrick. How else do you suggest coming up with a measure that represents how effective a transport system is and the impacts of various proposals, be they good or bad?

Leave a Reply

Your email address will not be published. Required fields are marked *