The media seem went gaga a few days ago about a new report that looks at the impacts of congestion in Auckland and other Australasian cities. It’s no surprise they reported on it either, transport in Auckland is a common gripe and so the story of Auckland being the slowest city, confirming what many people already believe, was too much to pass up.

But even reading the news articles, it struck me that this was likely another case of the wrong questions being asked but that by in large, the right answers were reached, much like ATAP. Perhaps even more so it wasn’t the main findings of the report that were important but some of the ancillary comments.

The report is by Austroads, a key organisation in the development of road networks in Australasia. Here’s some of what the Herald said about it.

Auckland commuters can groan in agreement with a report that reveals the city has the worst travel times and reliability in Australia and New Zealand.

This is despite having the fastest road in the two countries combined.

Austroads Congestion and Reliability Review measured the levels of congestion and traffic across major cities in Australia and New Zealand and released their findings in a report.

The cities were categorised into groups of similar population size. Auckland was in group two alongside Perth, Brisbane and Adelaide.

In its group Auckland performed worst across “most measures”. These included the highest travel time delay, morning and afternoon peak reliability.

Auckland came out worst in it’s group with an average speed of 77.6km/hr. The other cities were between 82km/hr to 86km/hr. American cities of a comparative size went up to 104km/hr.

The slowest roads were St Lukes Rd, Wairau Rd and Lake Rd. While the fastest were the Northern Gateway Toll Rd (which topped the Australasian list with an average speed of 98.8km/hr), Upper Harbour Motorway and SH16.

By focusing on the speed of commutes, my immediate feeling was that this sounds like a very similar approach taken by Tom Tom and other organisations for their own congestion reports. This proved to be correct as this is what the press release says about the method they used.

The findings are based on an analysis of Google Maps data for 600km of roads for each major Australian city, enabling analysis of travel time along different road segments. The analysis was based on two months of data, comprising of 1km long road segments, with data points taken every 15 minutes, to calculate the six key congestion measures outlined in the report. An econometric analysis was then undertaken to provide insight into the drivers of network performance.

We’ve written before about TomTom’s reports, one of the key problems with reports that use this kind of methodology are that they are based on measuring the variance in the speed of traffic compared to the maximum speed limit allowed on the road. This ignores that roads can be more efficient and flow better with lower speeds, for example a motorway will move more vehicles an hour/flow better if they’re all travelling at 80km/h than 100km/h.

Also importantly, these reports only ever reference vehicle congestion/speeds, not the people using that corridor. Infrastructure like the rail network and Northern Busway allow a lot more people to be moved within a corridor, often faster and free of congestion. We know that 40% of people crossing the Harbour Bridge every morning on a comparatively small number of buses don’t suffer from the congestion those in cars next to them have endured.

But it’s here were we reach one of those right answer to the wrong question points. That’s because while we know that counting the speed of PT and other road users is key if we want a more accurate report, we also know that Auckland is one of the worst performers when it comes to use of other modes, something noted by the authors when they say

Auckland has fewer public transport options, compared with other cities. Plentiful low-cost parking in Auckland encourages commuters to drive.

Looking at the articles also raised a new potential issue with how these kinds of congestion metrics are created, in particular the mention that the Northern Gateway Toll Road was the fastest in Australasia. In short, why would you count a rural tolled motorway. Other than on holiday weekends, which wouldn’t be relevant in this discussion, this road has almost no bearing on the congestion most Aucklanders might experience. But that got me thinking, if the toll road is included in Auckland’s numbers, what else is included, especially in other cities.

So here’s Auckland’s map of the most delayed roads based on their methodology.

And as one example, here’s Perth which comes out much better in the metrics.

And Brisbane

One thing worth noting is it appears these maps are not at the same scale, the furthermost point away from the Auckland City Centre is about Drury, 30km away. By comparison in Brisbane and Perth some of the locations shown are over 50km away from the city centre. This hints at one of the potential issues, there are a lot more uncongested rural highways included in the numbers of other cities compared to Auckland. To be fair, in the middle of the report they do say that the types of roads selected for this report will impact the results i.e. more motorways generally results in higher average speeds.

The report also looks at some international cities as comparisons but oddly they decided specifically to choose cites with similar PT modeshares to the Australasian cities which means cities of similar sizes and densities in places like Europe and North America can’t be compared to see if there’s something cities here could learn if there was a different transport/land use focus.

As mentioned earlier though, despite issues with some of the ways this report has been approached, I do think it comes up with some useful points.

More important that the speed of a particular journey is the reliability of it. If you know it’s always going to be slow then at least you can accept that, or adjust when or how you travel to avoid it. But when that varies considerably every day it can be infuriating. Of course, Auckland fares poorly in the reliability rankings

They also make some very salient points about why Auckland performs poorly. These include:

  • Auckland’s geography, particularly its harbours and waterways, impose constraints on the transport system, meaning the main transport links are confined to narrow corridors
  • Compared with other cities there is a lower level of public transport provision for commuters
  • There exists highly available and low cost parking in Auckland which encourages commuters to drive

Auckland’s geographic constraints forces a lot of people to travel through narrow areas, if only there was a way to move a lot of people without needing a lot of space to do so. Despite what some urban myths would have you believe, that makes Auckland ideally suited to public transport, if was provided it well. This is something backed up to some degree by ATAP, which noted that there are only limited options for expanding road capacity within most of the urban area.

There is no one solution to ‘solve’ congestion but definitely a “more of the same but bigger” approach will not work. If Auckland wants a chance of improving congestion it will need provide better alternatives so the best option isn’t to sit in a car and hope for the best.

Share this


  1. Ignoring PT (which I know is a big ignore), what would be the correct methodology to determine (purely) road/driver congestion?

    I’d think average speed would be correct (as others like average time would favour smaller cities)? Does this survey use the wrong way to calculate average speed i.e. is that 78kmh actually 78 or is it a composite?

    1. methinks average speed would also be endogenously determined along with city size.

      The key indicator I’d want to understand are transport costs per capita. And I suspect that data sources like the MoT’s household travel survey provides a much better indication of travel costs (travel-times, distances, and speeds) across New Zealand cities and towns.

      The other good thing about the HTS is that it includes data on all modes ocer time, and captures data for a number of regions in NZ. I also understand there are comparable surveys in some Australian states.

      Personally, these AustRoads / TomTom studies are a waste of time. They go to a lot of effort trying to communicate information that is essentially useless.

      1. But that has problems too

        Imagine 2 cities, each of which have only 1 driver and 1 road between Home and Work

        City 1: Driver averages 100kmh to work
        City 2: Driver averages 80kmh to work

        In that case, city 2 would rank higher despite being clearly slower

        1. Personally, I think the whole analysis is largely unhelpful is not bogus. It’s unhelpful because the performance of the transport system in Perth doesn’t really tell people in Auckland what they should do, either when they are planning or using the transport system.

          It’s bogus because from an economic perspecive a good road is a well-used road, and a well-used road will (from time to time) be a congested road. In some ways this kind of analysis tries to turn socio-economic measurs of value on its heads.

          Ultimately, the main variable I’d look towards to understand if a city is succeeding is population growth. If people are moving to one city over another, then it probably has something going for it. In the case of Auckland, I think we are doing pretty well on that front.

        2. World’s largest cities include
          Sao Paolo
          Mexico City

          Any analysis that says these are good cities because they are big is flawed.

        3. Alumnus, no in that situation they would rank equally as both cities essentially have no delay compared to optimal use.

    1. They were comparing similar sized cities, Sydney and Melbourne are in the next category up as they are much bigger. It is a relief to see a report comparing Auckland with cities of similar sizes, rather than the usual irrelevant attempts to compare us with much larger cities.

      1. Similar sized population that is, not area, in which Auckland is far denser and more compact. One thing to note is on the diagrams above, the Auckland linear scale is half that of the other two, so in area terms Auckland is shown four times larger than Brisbane and Perth.

  2. It seems Austroads suffers from the same silo-ing single mode focus that bedevils much of NZTA. It is well understood that the quality of the alternatives to any system of movement ultimately govern how much that system suffers from destructive over use. No viable alternative; then whatever system you have will be prone to breakdown through overuse at times and places; both efficiency and effectiveness will decline suddenly and rapidly. Furthermore spending to expand that one system instead of completing alternatives will become increasingly expensive and wasteful as offers ever diminishing returns. Any truly objective look at urban movement structures will show this.

    The problem comes when we have built up whole institutions around one system that are then either unable or unwilling to consider the city the as a whole and all its potential systems holistically. Then there is a danger of forever doubling down on the full mode at great cost, which only doubles down on the problem. Traffic congestion is a kind of paradox; it is a success that fails. And the solution to it is a kind of paradox too: The best way to reduce the problem of traffic congestion in cities is to invest away from the system, to stop feeding it with more tarmac, and provide attractive, connected, well run, frequent Rapid Transit alternatives. And by reducing the need for mechanised short trips at all, through cycling infrastructure, and creating great intense walkable centres.

    For Auckland to succeed this century it must have a clear policy for becoming a thriving ‘post-congestion’ city. With urgency. The bones are there; let’s build whole creature, well and quickly.

    1. The mode myopia is hardly surprising when two of Australia’s State Transport Agencies are called “Main Roads Western Australia” and “the [Queensland] Department of Transport and Main Roads”.

      1. The thing with Main Roads WA is they only deal with the main roads in the state (as well as directing state and federal money for local governments for their networks throughout the state). WA still has a Department of Transport, and the Public Transport Authority as well, for everything else. How they interact and dominate the discussion is a whole other can of worms.

  3. Of course this also makes the mistake of averaging, without any measure of variance or spread. If your average drive is 30 minutes, but a couple of times a fortnight it takes 60 minutes, then really you as an individual have to allocate 60 minutes every time to get places when you need to be there…. so your real world trip is actually 60 minutes each time.

    Anyway, surely simply journey time is the correct measure? That takes into account both intensity of traffic and exposure to traffic. In other words I’d rather live in a city where you can get where you need to go in fifteen minutes at 20km/h that spend an hour at 50km/h to achieve the same purpose. That’s the fallacy with looking at the northern toll motorway through orewa. It may be fast and lightly used, but if you are using it you’re basically driving 30km or more between Puhoi or further north and Albany or further south, even with zero traffic that will take about 20 minutes.

  4. Benpaul, surprisingly no. Sydney motorways on the whole flow much better than Auckland’s and negotiating the city streets is comparatively easier too. This is down to multi lanes each way with no stopping on main arterials, which we should be doing here. This is from experience I have for a week every month that I am based there.

      1. It will improve traffic flow *on the particular route that it follows* (Strathfield to Camperdown and the adjacent Parramatta Road). BUT because it will induce more traffic, and because no-one’s trip starts and ends exactly on a motorway ramp, it will increase congestion on the existing arterial road network over a wide area around it – including congestion suffered by many people who never go near the new motorway.
        Even if the increased congestion from this particular project is relatively minor, you’ve still blown $10 billion on infrastructure that hasn’t solved any long-term problems. Always think opportunity cost: that’s money that you could have spent on improving congestion-free transit ….

      2. … This is the essence of the problem of cars in big cities: If you plan the transport system of a large, growing city in a way that means that over 90% of all trips will be by car, you’re planning for serious, permanent, insoluble traffic congestion. There’s no way you can avoid that by building more roads, because –
        – the amount of new roads needed would be wildly unaffordable; and
        – they would take up so much space, and make your city so ugly and unpleasant for every activity apart from driving, that you wouldn’t want it anyway.
        If you’re willing to suffer the congestion in order to have the other conveniences of a low-density, car-dependent lifestyle, you’re entitled to your opinion. As a politician or planner, you’re not entitled to bulls**t people by pretending that you can give them a car-dependent lifestyle AND build your way out of congestion.
        All large prosperous cities have traffic congestion. They should be judged by how well they provide congestion-free transport alternatives and appropriate landuse planning so that more people can avoid being in the congestion.

        1. “As a politician or planner, you’re not entitled to bulls**t people by pretending that you can give them a car-dependent lifestyle AND build your way out of congestion.”

          Ideally, no. In practice, yes. I expect any day now the announcement that the additional harbour crossing is going to be built, and how it will finally “fix this bottleneck in our system” and “eliminate the traffic jams on Onewa Road, Esmonde Road and the northern motorway”.

  5. Wouldn’t a more useful measure be vehicles per unit of time passing a given point or even better, number of people per unit of time past a given point.

    The actual speed of the vehicles or the average speed of the vehicles doesn’t tell us how many people are getting to where they want to go.

    1. We’re conflating two things:
      1. The ability of a system to move the greatest amount of people a given distance in the smallest time, and
      2. The time it takes for any individual driver to get from A to B

      For example, an incredibly crowded city might have an incredibly well-designed and efficient system that moves thousands of people from home to work

      A very sparsely populated city might just have a couple of roads, but because of a lack of traffic, the commute is really short

      From an overall design perspective the first city is “better” but for individual drivers it isn’t.

      1. Why is the first better, the assumption ‘given distance’ is a confounding variable, i.e. better transport/land use systems have shorter distances between A and B, all else being equal.

        1. You’re now conflating land use and transport networks

          If we take as an assumption that in both cities the distance from home to work is the same, the latter is still better for any given individual driver (though the advantage will disappear as n increases) but the former is a much more efficient system.

        2. “You’re now conflating land use and transport networks”

          I’m not conflating them, I’m actually exploring the interplay between the two.

          “If we take as an assumption that in both cities the distance from home to work is the same”

          But that is an insane assumption, the driver in a city twice as dense has twice as many options within the same distance, so they don’t have to drive as far to reach any one of them

          The relevant comparison isn’t “How fast can you travel x km’s”; that’s meaningless without some relation to how far you need to drive. The appropriate comparison is “on average, how long does a person take to get to work, compared to driving at speed limit”.

          Your measure is a measure of how much congestion there is, mine is a measure of how much congestion people experience. As an example, London has really intense congestion but less than half of all people drive to work, Houston is the other way around, Londoners experience far less congestion than Houstonites though because they travel less far and on congestion free trains, on bicycles, or on foot.

  6. Not sure about their data. Google Maps can underestimate time delays due to congestion, sometimes by a large margin. Especially with queues at traffic lights and ramp meters.

    More importantly, will it tell you how many people pass along a given segment? If you have two segments, (1) with 100 people passing by at 100 km/h, and (2) with 1500 people passing by at 20 km/h, naïvely averaging the speed gives you 60 km/h, while most people actually experience 20 km/h. The average weighted by amount of people passing by gives you a more meaningful value of 25 km/h.

    1. I imagine the reason Google miss the mark on predictions would be because they are using past data to make a future prediction. When you leave work to head home at 5pm it might be quite accurate to say it is currently taking 45 minutes to do that trip, but as the traffic builds after 5pm it might end up taking you 1 hour.

  7. We could introduce a similar metric for walking.

    Walking in Auckland CBD roughly works as follows: you walk for a block, press a pedestrian button and then wait for a few minutes for a green light. Repeat at every single block. If you don’t jaywalk, it may take 20 minutes to walk a kilometre.

    Now, let’s think about this flash new blob of tract houses in Millwater: There’s still intersections, and while having to give way to traffic coming from your full 360° field of view, you’re unlikely to wait for minutes at every crossing. Your average speed will be a much higher than in the CBD.

    In conclusion, Millwater is a better place to walk to work. Aren’t those metrics just marvellous?

  8. The best metric is presumably minutes of delay per person / tonne of cargo (monetised and added together). Then work out congestion costs per capita.

  9. one more reason why this analysis may not be that helpful: Tolls.

    In the case of Brisbane, it looks like a number of the new toll highways (Clem 7, Legacy Way, AirportLink) are coloured green. That may be way Brisbane achieves far faster speeds on its highway network: because it has built a lot of highways that aren’t very well used.


  10. Another reason is that travel times between modes reach an equilibrium. Perth and Brisbane have fast trains on some routes that can offer end-to-end travel times competitive with driving. For CBD trips, this would entice enough commuters from road to rail until freeway traffic speeds achieve a new equilibrium. Perth has some of the fastest urban rail speeds in Australasia on its northern and southern lines, while Brisbane now has fast services from the Gold Coast and northern suburbs. Following this model, addressing the train dwell time issue in Auckland would increase peak hour motorway speeds.

  11. How about this as a depiction of transport success/failure? A kind of topographic map/image where instead of height it is -at every point there is a measure of how many people can reach that point in a given time (30min, 60min) by whatever mode, at a particular time of day?

    I have no idea how such a map would be created -it might take a lot of computer power….

      1. That map makes it look like the only area with any significant amount of employment is the CBD. Secondary centres like Albany or Manukau barely register.

        Now does anyone have a similar map for driving?

        1. Its not an map of employment in those areas, its a map of access to employment available for residents in those areas.. Not surprisingly the centre of the region is the point with the best access to regional employment… because its in the centre. Living in the CBD gives you the best number of jobs accessible within 30 mins.

          The secondary centres in the outer suburbs barely register because if you standing in one, you don’t have many jobs accessible within 30 minutes, in most cases only the jobs in the secondary centre and the suburban area around it. That makes them no better than the suburbs 20 minutes away that also only access that centre and the suburban area around it.

        2. Yes, and it appears most colored areas (other than dark blue) are those areas which have easy access to the CBD. The Southern Line is easily recognisable. On the North Shore you can recognise Takapuna and Birkenhead, which are the only town centres with easy access to the CBD. Somewhat surprisingly, the northern busway is almost invisible.

          OTOH there’s not much evidence that there’s any area around Albany where there’s a lot of jobs easily accessible. I was expecting the CBD to stand out more than the other centres, but not by a margin that wide. Perhaps the color scale is too coarse to display the lower numbers around those centres. Or maybe public transport doesn’t quite work as well over there.

          Also curious what’s so special about the Ellerslie – Panmure highway.

          This definitely shows that if you’re used to the isthmus, you really have to be careful with assuming a lot of people in the rest of Auckland could use PT to go to work.

        3. Roeland — just to clarify: The map shows a locations accessibility to employment when traveling by a combination of PT *and* walking.

          So those red areas on the southern line are a function of localised employment centres that can be reached by walking (e.g. around Greenlane), rather than anything to do with the PT network. The fact the NB and Western line don’t have any similar red areas is not a function of transport outcomes, but the absence of employment within the walking catchment.

  12. One of the things that we discovered in Wellington at the Basin Bridge Board of Inquiry, was that congestion is only calculated by the amount of cars during working hours Monday-Friday. Because that is when people are earning money, and therefore they need to be sped up. If congestion occurs later in the evening or – god forbid – in the weekend, then it is of no concern to the traffic planners, as it is just non-essential traffic, like people going to watch their children play Saturday sport, or going to catch a movie or theatre, or even going on non-productive events like a holiday out of town.

    A curious way of looking at the world, and one which may explain why they really don’t give a shit about traffic jams full of Sunday drivers. Tough!

  13. The metric they have used is valid. They can’t report each person’s travel time so they have to present an average of some kind. Speed is something most people can understand and compare. Given that 82.7% of people in Auckland who travelled to work on census day in 2013 drove or went as a passenger in a private or company vehicle then this metric is going to have meaning to the vast majority of Auckland workers. It is just that the way social media works that most of the people who read or post on this website probably come from the 17.3% minority of workers in Auckland.

    1. Sort of. It fails to distinguish between private vehicles and busses in that I can get in a private vehicle and go, whereas I would have to wait for a bus. Truly capturing journey times also includes capturing the timeframe that you have to allow to make the journey, not just the time you spend actually making it and in a vehicle or on a bus/train.

    2. yeah, nah.

      Let’s consider some of your statements in more detail:

      – Mwfic statement #1 –> Need to aggregate data across many individuals. I agree, provided the distribution is not too skewed.
      – Mwfic statement #2 –> Average speed is s a valid metric because people can understand it. I disagree because AustRoads reports such as this are not targeted at the average Joe Plumber, but rather transport professionals and (of more concern) policy-makers and politicians. Thus the information presented should be critically evaluated in that light.
      – Mwfic statement #3 –> Many people drive therefore average speed is relevant. I agree, but would suggest that what people actually care about is the time spent travelling, and to an extent reliability. Time spent travelling is a product of average speed and distance traveled. This begs the question of whether we see more variability between Australian cities in terms of 1) average speed or 2) average distance traveled? If we see more variability in terms of the latter, then that would be the metric I’d present first. Of course that’s a function of both exogenous and endogenous factors, such as geography and land use density, but no more so than average speed.
      – Mwfic statement #4 –> People who comment here are biased based on their own personal choices and/or preferences. Playing the man not the ball = classic cop-out, especially when you haven’t even put forward a strong case for your own point of view.

    3. I find the whole thing a little meaningless. Average speed is no more useful than average wage. It’s correct for no one and unhelpful for me as an individual to compare to. I find median values are more valuable. Median speed, median delay, median wage etc. The problem is that most people don’t even know the difference between mean/median unless it is explained to them. If 85th percentile speeds are well above the speed limit it would skew the data if you are comparing the average against the posted speed limit.

  14. The things about roads is that the desired situation is that they are almost empty. But wait,surely the more packed they are the more worth the huge expense of building them. Likewise buses and train packed doesn’t cause delays. Reward public transport users with low fares. Charge more for car parking

Leave a Reply

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