If there’s one over-riding dogma for transport planners throughout the past 50 years it would be this: “congestion is bad”. Almost everything that transport planning, traffic engineering and anything else related to this field has been about over the past 50 years is based on the belief that congestion is a bad thing, and we should do everything we can to minimise, avoid or – at best – completely banish congestion. A few years ago someone (we’re not really sure who actually) came up with the statistic that congestion was costing Auckland a billion dollars a year, or close to it. That figure (wherever it came from) has been used to politically justify much of the money spent on ‘improving’ Auckland’s motorway network over the past few years, in particular the almost religious belief that the Western Ring Route is the solution to all our problems.

However, all of this is based on an assumption – that congestion is a bad thing. Now of course we all hate getting stuck in a traffic jam, as it wastes our time, causes frustration and (possibly) wastes our petrol as we sit there doing nothing. So perhaps it seems nonsensical that congestion would be anything but bad. However, taking a broad view of matters if you really think about it there are a number of positive effects of congestion, that could lead us to really questioning whether traffic congestion really is enemy number one for traffic planner and transport engineers around the world. This post explores that possibility.

A classic example of a congested route in Auckland is Onewa Road on the North Shore. I drive this road reasonably frequently, including at peak times (I know, bad me, but I don’t really have any practical alternatives), and it does get very congested indeed. During the morning peak the road operates as two lanes – a general traffic lane that edges incredibly slowly down the hill from Highbury towards the Northern Motorway (at off-peak times this length of road shouldn’t take much more than 3-4 minutes to drive, at peak times it can take 20 minutes) and a T3 bus and carpool lane. Because the general lane is so congested, a lot of people catch the bus, carpool or drive during off-peak hours. All of these steps that people take to avoid the congestion of Onewa Road result in a positive outcome: higher public transport use, fewer people driving their cars, a greater spread of users throughout the ‘shoulder-peak’ periods and so forth.

However, these ‘benefits of congestion’ are rarely considered in the analysis of transportation improvements – with “time savings benefits” reigning supreme as the ultimate justification for roading projects. These time savings are, of course, created as the result of reductions in congestion – as people are able to undertake their trip quicker than before, thereby creating time at the beginning or the end of their trip they previously did not have, therefore creating a ‘benefit’. Now I think time-savings benefits are a load of rubbish and quite possibly non-existent, but that’s not really a debate for this thread. The bigger issue is “should we be worried about this?” Is it really the end of the world if our roads are clogged if we have a viable alternative? Clearly, the last bit of that is key – that there is a viable alternative in the form of a bus or (even better as it has its own right of way) a train.

An early chapter in Paul Mees’s book “A Very Public Solution“, which I have just started reading, probes this fundamental issue: is congestion really a bad thing? Could it be a good thing? An important aspect of answering that question is to look at who is affected by traffic congestion. Is it really a transport externality? The book states:

While dire estimates of congestion costs are frequently used to justify investments in roads, costing congestion is a difficult – if not nonsensical – task. The first problem is that congestion is not strictly an ‘externality’, the term economists use to describe costs that are borne by people other than those creating them.. Pollution from motor vehicles is an externality since it affects non-motorists as well as motorists. But congestion primarily affects the same group of people who produce it, namely road-users themselves. It may actually improve the lot of some residents, since slow-moving traffic makes less noise and is less intimidating for pedestrians and cyclists. And although there may be more accidents on congested roads, they will be less severe, owing to lower speeds. Congestion is only an externality in the sense that it is an effect each motorist imposes on other motorists.

So congestion doesn’t really affect anyone else but those on the road in a negative manner, and in some respects actually provides benefits to those other than the very people creating the problem. However, clearly it does seem as though there remains a problem – all that time lost while people are sitting in traffic doing nothing. Surely that’s economically inefficient? Surely there’s a real cost there?

This brings us back to the existence of ‘time savings benefits’, and also issues like induced demand that traffic planners and engineers tend to ignore because they upset the simplistic world of “predict and provide” they live in. Fortunately, there are a few smart thinkers out there who have looked at this issue more closely – including a guy called J.M. Thompson in a book called “Great Cities and their Traffic” – who outlines the following (quoted from Mees’s book):

Where a road and a rail system compete for patrons, Thompson argues that there will be an equilibrium between the two travel modes which ensures that they are of roughly equal quality. An increase in traffic on the road would raise travel times, encouraging some motorists to shift to public transport, a reduction in traffic would attract passengers from public transport until congestion rises to re-establish equilibrium. This equilibrium can be upset by changes to the quality of either mode. Improving the road system will produce a decline in patraonge of the rail service. This will cause a reduction in service levels, leading to a further decline in patronage. If sufficient rail passengers shift to the road on account of the decline in service a new equilibrium will be reached in which, paradoxically, both road-users and public transport patrons experience a worse level of service than before. Investing in road improvements has actually made everyone worse off.

Now aside from extreme situations like horrifically crowded trains in Tokyo, generally public transport works better the more patrons it has (as increased services become viable) while roads perform better the quieter they are. Therefore, taking steps to encourage people to use public transport rather than roads – through shifting the equilibrium in that direction – is likely to result in benefits to everyone.

Mees concludes that perhaps the best approach to congestion is for everyone to just relax a bit. There will always be congestion in large cities, but perhaps the focus should be on providing alternatives – as in cities with well developed transport alternatives  people will be able to choose whether or not to endure it. The 1993 Vancouver Long Range Transportation Plan takes a similar viewpoint:

Congestion is usually considered an evil; however, allowing congestion to deteriorate for single occupant vehicles is a practical method of promoting transit and carpools. More congestion for single-occupant vehicles would magnify the impact of some travel demand management. For instance, buses/carpools in high occupancy vehicle lanes will gain an edge since the relative time saved by escaping lineups will be gone.

In fact, I think some level of peak hour congestion is actually probably a desirable outcome. If we had a complete absence of congestion at all times it would surely be a sign of huge over-investment in the roading network (if it was even possible, remembering the effects of induced demand).  Places like Paris have more congested streets than a city like Los Angeles, but I’m doubtful that Parisians have poorer access around their city than their Californian counterparts. In fact, it’s likely the opposite is true. Slower traffic encourages alternatives means of transport that are often more sustainable, it encourages shorter trips and thereby encourages higher development densities and more mixed-use development. Which are all good things.

In the end, I think it’s stupid, and probably even counter-productive, to attempt to eliminate congestion. Instead, perhaps it’s more prudent to plan for an optimum level of congestion – keeping in mind other environmental, economic and social goals. If we set the equilibrium at the right level, in the end we will all benefit from it.

Share this


  1. Very interesting… The mind instantly thinks congestion is something to be eliminated at all costs but you seem to propose providing a BALANCED transport network with CHOICE… No bad thing…

  2. Ha ha! Your post is basically a summary of one of my responses in Mees’ exam that I sat last night.

    The externality issue is interesting, it is often assumed that becuase everyone drives congestion affects everybody via the economy etc… but that doesn’t mean that you can apply that reasoning to some particular road project that has only a small and specific set of users. Very intersting.

  3. Eliminating congestion is never economically efficient, but the optimal throughput of a road is just below the point when flow breaks down and is irregular, which will be below the speed limit.

    Bear in mind congestion increases emissions, as it is the suboptimal use of engines, but it IS largely about wasting time. Most regard waiting for a phone line before making a call, or substantial queues for most commodities, or power cuts to ration usage of electricity, or water rationing as being grossly suboptimal. Indeed the same is when public transport is grossly overcrowded and people have to wait till the second or even third vehicle. In all of the cases I describe, pricing is used to some extent to manage that demand, and spread it.

    The public transport effect of high patronage being a good, is the Mohring effect, in that more people allows more frequencies and capacity to be better utilised – up to a point, which is when additional capacity for spikes of peak usage involves far too much capital. Bear in mind most capacity built into urban road and rail networks is for short spikes of usage in single directions, which is a lot of capital tied up in servicing that sort of demand.

    The key point is that roads are a scarce resource, and roads are at the most efficient when the most users can travel at an optimal speed with delays minimised. Given most roads have large peaks and troughs of demand, the most optimal way to manage congestion in the first instance is to spread demand across time. A second option if available is by route.

    Congestion is a result of demand exceeding supply. It either indicates underinvestment (particularly if congestion lasts for considerable periods of the day, or is concentrated around a junction which may simply be inadequate), or poor pricing (which is pretty much universal). There are those who value time and for whom congestion is a significant cost to them, particularly the freight sector and businesses which are transport dependent (e.g. service sectors such as tradespeople, salespeople). They will pay to use roads at peak times that are relatively uncongested. There are those who don’t value time much for whom congestion isn’t a big deal, typically those on leisure trips. However all users impose the same marginal cost on each other. Bear in mind congestion can cost lives, probably rarely an issue in New Zealand, but gridlocked narrow streets can make emergency services responses particularly fraught with difficulty – which is one reason why priority lanes are positive, but they only have limited application.

    I had an argument with a Green Party advisor once who said congestion was good, and I asked whether vehicles sitting in traffic burning fuel going nowhere fast was good, or whether it was better to price some of them off the road so only those who really valued the road at that time and place used it – efficiently. He appreciated the point.

    It’s worth noting that something like 20-50% of urban congestion can be purely due to incidents blocking roads, like accidents or breakdowns – effort in addressing those and informing users to avoid routes or driving because of them are usually extremely good value for money.

    Congestion is simply wasting time and in this case, fuel, and increasing pollution. Eliminating it is a nonsense, but treating it as a positive thing is quite Soviet in outlook, for obvious reasons. Pricing to manage demand is the norm in other sectors, it is about time it was for roads.

  4. Liberty-
    In pure economic terms then- shouldn’t we be charging vehicles for the use of roads on a user pays basis. More popular routes could be tolled higher, therefore reducing use and less travelled routes tolled for lesser amounts based upon traffic flows at different times of the day. This would allow for the optimally efficient use of roads.
    You could then upgrade the routes that are making the most profit (the ones that motorists most want to use).
    If you wanted to benefit the economy you could reduce toll costs for more economically valuable traffic like trucks, ambulances etc.
    The other big advantage of this as I see it is that it eliminates low value use of some roads. If they had to pay, local road users who could otherwise walk, cycle or use public transport would elect not to use high priced roads, which would speed up travel times.
    The other use of tolls is that you could charge for the loss of amenity a road causes. if a road for example was noisy you could use the profit derived to double insulate nearby houses or compensate locals and local businesses who have had their quality of life or streetscapes diminished or parking reduced.
    Might even be a way of recovering carbon credits derived as a result of the CO2 produced- so we can compensate our children for the effects of 500+ ppm CO2.

  5. I have to agree with liberty on this point, most public transport spokemen will avocate about minimising time, polution, health etc, yet when it comes to roading totally contridict themselves by supporting such things, albeit indirectly. Congestion in the end is only a bad thing when its related to all forms of transport, including public transport.

  6. I actually think we are all saying pretty much the same thing here: that it is pointless to try to eliminate congestion, that congestion has some benefits but also some negative effects, that it’s very wise to offer a variety of transport options, and that during peak times congestion happens because there is an ‘under-valuing’ or under-pricing of road space.

  7. Topcat: Yes, I believe in theory all vehicles should pay a form of RUC, which is calculated as thus:

    Vehicle type (based on weight/axles with optional factor for emissions categories) x distance x location/road type x time of day = charge

    Vehicle is static, distance is easy to follow, location may be based on 3-5 types of road (arterial, access and unsealed at least) and gradations of urban/rural (central, suburban, provincial, rural) with time of day having a peak, shoulder and off peak split. There can be variations, but you get the point.

    Given that there is a long list of countries in Europe adopting technology (albeit for trucks so far) for pricing roads on the sort of basis I am talking about, or at least with the capacity to do so, it wont be long before commercial vehicles, at least, are already pre-equipped with the means to do this. The Netherlands is talking about extending it to cars after it has a truck electronic RUC system in place, not surprising since it has the worst road congestion in Europe. That will be revolutionary, for all road users, but also rail.

    At that point, we will get efficient improvements to roads, capacity increases when demand exceeds supply over periods longer than short daily peaks, far more clever traffic management to maximise use of hard shoulders or even tidal flow lanes and priority lanes.

  8. So you support an agency/authority or *gasp* private company with the ability to track your vehicle movements at all times Liberty..?

  9. That’s what I’m thinking, working in law enforcement, I’m pretty proud of the Privact Act and if anything think it should be strengthened… The laws the government are currently passing are very troubling, kind of like our own Patriot Act…

  10. The laws that the government are now passing are beneficial, if it insures our own safety and fair protection from scamming idiots, then why not, the only reason you could be against what the government is proposing is if you got something to hide. Glad I got that out of the way 🙂

    However I cant see this system being implemented although it could have good safety features attached. To me it would be to complicated, people modify their cars all the time and it wouldn’t be hard for most of them to put a kill switch in to say take half their actual use. Although that would probably happen at the moment with diesel cars and their odometer.

  11. The “something to hide” argument is very naive, these laws are being passed without any increase in oversight, they will be abused, people will have their computers searched illegally more often, innocent people will have their property seized…

    I worry about that very deeply…

  12. Guys, the govt can do that already. Haven’t you seen the episodes of Spooks where they track vehicles with all the thousands of traffic cameras in London. How do you think tollways catch up with non payers for the London congestion charge is administered. If you don’t want anyone to know where you go (why would anyone care?), don’t carry a cell phone, don’t drive, don’t go into any shopping centres, don’t have a twitter account, don’t use your credit card, avoid using your computer. The extra information that could be derived from congestion pricing would be minimal.
    The success of the London congestion charge shows what congestion pricing can achieve, I suspect the only people against this move would be the oil companies as it would make people think carefully about how they use the roads.

  13. Jeremy Harris; Movements do not need to be tracked at all times. All that is needed is an aggregate of distance travelled on roads classified by different types. The system architecture I am familiar with as an option has the unit on the vehicle measuring all these factors, spitting out a figure which is either the charge or can be translated into the charge at a back office. The information of where exactly you are isn’t necessary. In fact it is a ridiculous amount of data to transmit.

    As long as it is 100km on road type A-1 (sealed, urban arterial) at time P (peak so high charge) in vehicle type P (passenger car) = $x
    A-1 could be the southern motorway at Gillies Avenue or the northern motorway at Esmonde Rd or Ngauranga Gorge Wellington.

    I do give a damn about this sort of thing, but I am convinced that it is easy to manage privacy with it. After all, smartcard ticketing systems can also monitor movements in theory if people wanted them to.

    Topcat: You make a good point that there is vast amounts of data collected now about what people do, but most of the time there is little central interest in using it. The London congestion charge does only keep number plate information until someone pays up or the plate is matched to an exempt category. It’s not worth it to keep vast amounts of data of images daily for any other reason.

Leave a Reply