As I blogged about a few days ago, I am reading the book “Still Stuck in Traffic: Coping with Peak-Hour Traffic Congestion” by Anthony Downs. This book is probably the most in depth look at the causes, and potential solutions to, traffic congestion that I have ever come across. It is also an extremely comprehensive look at the concept of “induced traffic demand“, which I think is generally ignored when making important transport related decisions in New Zealand. One of the main points of Downs’s book is that building additional transport capacity (whether it is extra roads or extra public transport) is unlikely to fix peak hour congestion – because of induced demand – but may still bring benefits.

Focusing first on the results of adding road capacity, here’s how Downs explains how that approach is unlikely to “fix” congestion:

“…once heavy peak-hour congestion has appeared in key parts of a region’s road network, building new roads or expanding existing ones there does not reduce the intensity of such congestion much in the long run. Once commuters realise the capacity of specific roads has been increased, they will quickly shift their routes, timing and modes of travel by moving to those roads during peak periods, thereby filling up the expanded capacity.

…the resulting triple convergence will soon bring congestion back to its maximum levels during peak periods. True, because of greater road capacity, peak periods of congestion may be shorter, and more drivers can use the roads during those periods – which are the most convenient times for their travel. Moreover, the overall mobility of the region is increased because more vehicles can use the expanded roads during peak hours and off-peak hours. So expanding existing road capacity does produce significant benefits, even if it does not end or greatly reduce peak-period commuting.

However, if the metropolitan area as a whole is growing rapidly, the traffic added by growth will soon overfill the newly built capacity, and periods of maximum congestion will go back to their prior length. Also, the added travel capacity may help persuade more people and firms to move into the region, or it may cause more residents already living there to buy and use automotive vehicles.”

So in the shorter run, there would appear to be transport benefits from widening roads, despite induced demand. If we take the widening of State Highway 16 as an example of a transport project that is very much likely to ‘suffer’ from triple-convergence, while the widening of that motorway will definitely not eliminate (or necessarily even reduce) the severity of congestion along it at peak hour, it is likely to shorten the duration of that peak hour – as more people who previously avoided the worst of the congestion by travelling should the “shoulder peak” periods will switch to travelling at peak time. This might be more convenient for them, and it might free up the road during those shoulder-peak periods. So there are some benefits in the shorter term.

However, in the longer term, widening SH16 is likely to encourage further development along the corridor served by the motorway (a process certainly likely to be helped by huge planned developments around the end of SH16). This development, plus a potential mode shift away from public transport and towards driving, appears likely to ‘eat up’ most of the gains from the SH16 widening project referred to in my previous paragraph. So in the longer-term, it really does seem exceedingly likely that the SH16 widening project will be a complete and utter waste of $800 million. That’s half a CBD Rail Tunnel.

So how about public transport? One of the things that slightly frustrates me about Downs’s book is how quickly he writes off public transport as a solution for most places because he says that urban densities are far too low for it to be viable. While it is true that many US cities do have extremely low urban densities (far lower than Auckland), my understanding of Paul Mees‘s books, particularly in relation to the network effect – as well as seeing how successful public transport has been over the past 10-20 years in very low density Australian cities like Perth and Brisbane  – is that public transport can be quite successful and popular in relatively low density suburbia, as long as you’re smart about how you provide it. Putting that issue aside though, what Downs says about the effects of expanding public transport capacity is quite similar to what he says about expanding road capacity – in terms of effects (or lack thereof) on congestion:

“…expanding transit capacity rarely reduces existing roadway traffic congestion that has reached high levels of intensity. This conclusion may seem counterintuitive. If all the expressways leading into a major downtown are jammed every day during peak periods, it seems reasonable to assume that building an extensive fixed rail system on separate rights-of-way also serving the downtown will divert thousands of commuters off the roads, thereby relieving congestion.

In fact, such extensive systems have been built since 1950 in San Francisco, Washington DC and Atlanta. But peak-hour congestion did not decline in any of these regions; in fact, it got worse. Because of the principle of triple convergence, any initial improvement in speed on the expressways caused by such diversions to the new transit system did not last. In the short run, all the auto-driving commuters who shifted from expressways to the new rail systems were replaced on those expressways by other auto-driving commuters who had formerly traveled on other routes or at other times or on other modes. In the long run, the expanded overall capacity of each region’s transportation network – including more highways built in the same time periods – helped encourage more people and firms to locate in those regions. The resulting “induced demand” for travel soaked up all the additional capacity of all types in each region. This outcome was certainly not caused primarily by transit expansion. But neither did that expansion succeed in reducing rising roadway congestion in any perceptible way.

True, the expanded transit systems surely increased travel choices, thereby producing benefits for those who used them. And they enabled more people to travel during peak hours on transit and roads combined, thereby benefiting many auto-driving commuters too. But they did not reduce the intensity of peak-hour traffic congestion.”

This is a particularly interesting point actually, that just as triple convergence ‘eats up’ many of the benefits gained from adding roading capacity, it also eats up many of the benefits to road users of adding transit capacity. So perhaps just as it’s really debatable to say that the Waterview Connection will produce $2.6 billion of time savings benefits, it’s also debatable that 85% of the benefits of the Onehunga Line are really benefits that will be enjoyed by road users through reduced congestion at peak times. Of course, that’s not to say there are no benefits of expanding transit capacity – in fact I would hugely argue the opposite (and so does Downs, to be fair) – just that it’s unlikely that adding transit capacity will reduce the intensity of peak hour traffic congestion.

So if all these things don’t reduce the intensity of peak-hour congestion, what would? Well quite a lot of Downs’s book is dedicated to answering that question – and in short it is all about measures that can ‘spread the load’, ways in which more people can be encouraged to shift from travelling at peak times to travelling at off-peak times. It would appear as though the most effective way of doing that is through road-pricing: putting a price on road capacity that relates to the demand for it. Effectively, if you charge a high enough rate on congested roads at peak times a certain percentage of people will choose to travel somewhere else, at some other time or via some other mode: effectively a “triple divergence”.

Of course road pricing has its problems too, particularly in terms of social equity (to generalise, it will be the poor who are priced off the roads to create enough room for the rich to drive). Another issue is that unless you can come up with some sort of state of the art pricing system that covers every single road, pricing motorways is likely to result in traffic spreading to local roads – not exactly a desired outcome. Ultimately, it does seem as though there are two ways of dealing with the typical “big city” situation of too many cars and not enough peak hour transport capacity: you either price enough people off the road to free up that capacity, or you live with a certain level of congestion.

Which brings me to the question of whether we should actually even try to “fix” peak hour congestion. Or more widely, whether reducing congestion is actually what transport policy should be about. Road pricing, whilst potentially quite effective, does appear to be inequitable and works in three ways (diverging times, routes and modes) – only one of which (diverging modes) may necessarily be a gain at an overall level. By that I mean the disadvantages of forcing people to travel outside the peak period may actually outweigh the disadvantages of congestion – clearly that seems to be the case at the moment as people choose to travel at peak time even though they know they’re going to suffer from congestion (perhaps they have to work set hours, perhaps the benefits to the economy as a whole of having most people work the same hours is worth the congestion issue). Furthermore, as I already noted, spreading traffic away from major roads and onto local roads is likely to degrade the urban environment quite significantly, hardly the kind of outcome we would want. Putting people off driving altogether, by encouraging modeshift, or eliminating ‘unnecessary’ trips (and in the longer run encouraging more intensive urban forms) appears to me as the only real, long-lasting, benefit of road pricing.

So perhaps we’re asking the wrong question here. Perhaps the question shouldn’t be “how do we fix congestion?” Perhaps it should be “how can our transport system improve accessibility, making it easier for people to get around?” In a wider sense, I also think that we should also be concerned about questions like “how can we make our transport system more sustainable” and “how can our transport system help create a better Auckland”, but in a simple sense, it really does seem as though improving accessibility, rather than ‘fixing congestion’ needs to be the primary aim.

And there is a significant difference between the two. While building a new railway line might not fix congestion on the surrounding roads, it does provide potential capacity for 6 lanes worth of traffic in each direction to get where they want to go, at peak time, and be relatively unaffected by unexpected delays. In my mind, that is a huge transport improvement, even if all the people it “takes off the road” are quickly replaced through triple convergence. By contrast, although there may be benefits gained by doubling the width of a motorway – in that more people are able to travel at the time they want to – if all those people are stuck in congestion all you’ve really achieved is a situation where more people are sitting in traffic, not really a transport improvement in my mind.

Ultimately, I don’t know if it’s really possible, or even desirable, to try to “fix congestion”. The main tool for achieving such an outcome – road pricing – is potentially more trouble than it’s worth. But that doesn’t mean there’s no point in trying to improve out transport system, we just need to understand the reason we’re doing it: to improve accessibility, not to fix congestion.

Share this


  1. “So in the longer-term, it really does seem exceedingly likely that the SH16 widening project will be a complete and utter waste of $800 million.”

    Not a waste. You already pointed out that the widening would allow extra development in the SH16 corridor. That development provides homes for people who’d like to work in Auckland and jobs for them to work at.

    Motorways and bridges have been necessary for Auckland to grow from its pre-motorway population of 300,000 to its current population of 1.4 million. Take 1.1 million additional people and a conservative average development value (domestic real estate, commercial real estate, community real estate, and infrastructure) of (say) $200k per person and you’re looking at around $220billion of development. Not a bad payback for a few billion bucks of road.

    The question often comes back to whether you want a growing Auckland or not. If you want a growing city then you need to provide quality roads AND quality rail. The proposition that the Green MP posed in yesterday’s video clip, that someone is claiming that a particular road will “complete” the network is absurd. Provision of roads, railways, and other infrastructure will be an ongoing process unless you want to arbitrarily freeze Auckland as a museum of the year 2010.

    1. But Vancouver has managed to become a city of 2.5 million with only one motorway, and that motorway doesn’t go anywhere near the city centre. So I certainly disagree that you need motorways to allow urban development.

  2. I agree with Jarbury. Anyone saying “we need to continue building motorways because we need to have a BALANCE between road and rail” is living in fairyland.

    That balance was lost ago – it lies tipped over like an overweight truck, and we will not get it back on track by building five new motorways while adding one little spur line to the rail network.

  3. How much of peak congestion is education-related?
    Last month, I was sitting on the side of the road waiting for a bus and wondering why there were so few cars on the road- then I realised it was school holidays.

    How would you go about working out how much of this education related transport could be shifted to PT or otherwise by different means e.g. free public transport for school age students?

    Yes- it would mean a whole lot of other things need to happen, and it would have a whole range of other positive & negative impacts, but focussed on economics (and not ideology)- Surely this would save more in congestion, than it would cost?

    What are the figures? How would you work it out?

  4. The trouble with using Vancouver as an example is that they don’t have much of a rail system, either. They’re mostly flat, have a grid of some fairly substantial roads, and (as we discovered from a resident last time we discussed this) these grid roads are pretty conjested for a large proportion of the day. The grid wouldn’t work in Auckland which developed its local roads around the existing hills and, hey!, no one wants to flatten the city and every hill and building and start from scratch anyway.

    But if we ignore Vancouver, are there any other examples of cities that have quadrupled their population over 50 years and done so without building a hierarchy of roads?

  5. So what if Vancouver’s roads are congested? It still seems like an easy city to get around, has many transport options and so forth.

    Grids are helpful, although Auckland could achieve similar results with a hub and spoke system in the outer areas.

  6. Carl, a lot of travel is directly education related but we also need to realise that many non-education commuters change their travel patterns during school holidays too, i.e. many parents stay home for some or all of the holidays, others will start later or come home earlier, parents obviously stop doing the twice daily school run, teachers and admin support don’t drive to work and whole families go on holiday for a week or two etc. So even moving all students onto public transport wouldn’t create the same effect as the school holidays.

    “Motorways and bridges have been necessary for Auckland to grow from its pre-motorway population of 300,000 to its current population of 1.4 million.”
    Motorways and bridges allowed that growth, sure, but they were not necessary for it. A motorway system is only one way to facilitate growth in urban areas. Bear in mind that Auckland also grew from a walking town of about 25,000 in the 1880s (when the railway was established) to a transit city of 300,000 in the early 50s, mostly on the strength of the tram system and to a lesser extent the railways and harbour ferries.

    “The question often comes back to whether you want a growing Auckland or not. If you want a growing city then you need to provide quality roads AND quality rail. ”
    I disagree, either one alone or some alternative would support futher growth, I guess the real question is what is the best combination of transport options to allow growth in an economic and environmentally and socially sustainable way. I would argue that Auckland actually has an extensive network of quality roads and motorways for a city of its size, the real problem being the lack of much else.
    Sure a modern city needs some elements of both, but to me that means a lot of catch up with public transport before we think about expanding the road network again.

  7. “The trouble with using Vancouver as an example is that they don’t have much of a rail system, either.”

    I disagree, obi – no, we don’t need to talk ignore this one and instead hunt for other example cities. What the example of Vancouver shows is that you can have larger, and even very large cities and not dice and slice them into isolated pockets by covering them with motorways.

    HOW that is done is the secondary importance. Integrated work/life areas AND/OR great bus, rail or tram public transport systems AND/OR grid networks that disperse motor vehicle traffic better AND/OR cycling networks that get you 5, 10, 20% of all trips on bikes… all are possible options that need no motorways (or at least a lot fewer of them).

  8. I wonder if you are being too hard on road pricing. If a comprehensive system was put in place that replaced petrol tax with peak and off peak pricing then travel at low demand times could be potentially be cheaper for those not commuting in the peak (such as beneficiaries, seniors, etc).

    For those that need to travel in the peak this is precisely the time when public transport is at it’s strongest and able to provide an alternative to driving. So yes some would not be able to afford to drive in the peak but if the revenue from those that do was invested into a decent rapid transit network there would be an affordable alternative than shouldn’t be a second class option to those still driving (but probably not getting to their destination any faster than the PT user).

    Someone commuting from say a rural area where PT doesn’t operate could maybe still avoid peak road pricing by using a park and ride on the edge of the city at a point where there is some spare road capacity.

    There are many issues and also many possible ways to deal with them. For example maybe there would be free local travel zones so people that live, shop and work locally benefit while people that live on a lifestyle section and commute great distances across the city might want to reconsider their lifestyle choice.

  9. LX, yes fair point, although my post does say that the biggest potential benefit from road pricing is modal divergence. Putting the money raised into improving public transport – so that people did have a high quality alternative – would be the key.

  10. Totally agree Jarbury. Good to see the point made that PT improvements don’t on their own generally result in reduced traffic congestion but can contribute to improved urban mobility. I am always deeply uncomfortable when I see PT promoted as a cure for traffic congestion.

    The only cases I am aware of where actual traffic levels have been reduced are in cities such as Zurich where road space for private vehicles was also reduced at the same time as PT improvements. I recall there was also a case study with a German city where a citywide 30km speed limit resulted in reduced traffic congestion. Makes sense as reduced car travel speed should result in shorter average journey lengths and improve the relative travel time of PT, cycling and walking. The opposite of building a new motorway.

    Otherwise there is also the text book example of managing traffic levels through pricing that is Singapore. Again coupled with a high quality affordable rapid transit system.

  11. However the flipside of that is once you’ve paid your ten grand rego in Singapore you drive as much as you possibly can to get value out of it. No car owner would ever go back to public transport once they had invested that much in driving.

  12. Very true Nick. I’m generally not a fan of making charges ‘up front’ like that, for the very reason you mention.

    In fact I’d prefer car registration was a lot lower, with the difference in ACC levies made up through higher levies on petrol. In that case the more you drive, the more you pay.

  13. I agree with your conclusions. I think the goal shouldn’t be to “fix” congestion, but to generally improve transport choice and accessibility.
    In fact I think there is a strong degree of self -regulation that can play a part here. Once congestion gets to a certain level, many people will get so fed up that they will either change where they live, or change where they work.
    Although that theory hasn’t necessarily been borne out in parts of America, where many folks seem happy to put up with two hour per day commutes in pursuit of the McManion in the suburbs

  14. There is an economically efficient level of congestion, but what it comes down to is pricing.

    No network utility should ever function with such appalling inefficiency as roads do when severely congested.

    I’ve written enough about this before. LX makes the key point.

    If roads were priced reasonably efficiently, then peak time commuting would be very different indeed as it would mean:
    – Those who wished to use the busiest roads at those times would pay a lot and get relatively good traffic flows (not totally free flow)
    – Non-urgent users (tourists, freight) would travel at other times.
    – Interpeak users in cities would probably pay the same as now, evening use and those outside urban areas would pay less. Given public transport outside urban areas will always be sparse, that will seem fairer as well.
    – Public transport pricing could be reformed, with peak pricing being much higher than now (reflecting the very high fixed costs) and off peak pricing somewhat cheaper. This would also encourage peak spreading and more usage of underused networks during the day.
    – Cycling and walking at peak times would take off.
    – Telecommuting, working from home and the like would also take off, as would what was once “glide time” commuting.

    My biggest frustration of the two traditional “sides” of urban transport debates is that both focus on supply far too much.
    One side believes in building public transport, the other believes in roads.

    The road side has effectively had to concede that cost rules out building ones way out of congestion without pricing. Unfortunately the public transport side has yet to concede that without efficient pricing, it is still a massive transfer from all to a minority who use it (many of whom would have used it anyway).

    The last government’s investigation into Auckland transport was that the rail network would be seriously underutilised without road pricing. Sadly, it seems off the agenda for now. By contrast, replacing fuel taxes with road pricing is very much on the agenda in the US now.

Leave a Reply

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