It’s often said that for public transport to work, you need high residential densities. I don’t remember the exact figures, but I know that there are a few books I have read over the years that have gone to the level of calculating the densities required for different levels of public transport service – ie. what you’d need for a half-hourly bus service, a 15 minute frequency bus service, a railway line and so forth. It’s certainly true that the cities around the world which have the best public transport systems tend to also be those with some of the highest densities. Hong Kong is a classic example of this – with some of the highest densities found anywhere in the world and a public transport system that caters for well over half the number of trips taken in the city.

However, over the past year or so I have examined the relationship between public transport and population density in a bit more detail, and while the “very high density, very high public transport use” rule still holds, when you start comparing cities in New Zealand, Australia, Canada and the USA with each other the whole matter gets rather more complicated. For example, Brisbane and Perth have significantly higher levels of public transport use than Auckland; yet at the same time they are also significantly lower density cities than Auckland is. In fact, when one simply looks at the overall population density of Auckland, we’re up there with Sydney and Melbourne – and well above places like Calgary. Yet our public transport usage just doesn’t stack up with those other cities. So what’s going on? Is this a sign that urban density doesn’t really matter that much after all, or is the way that ‘density’ is being measured a bit too simplistic?

I would answer the above question by saying both. Firstly, the somehow ‘magic’ belief that just because you have higher densities public transport will somehow become popular is misguided. Auckland’s population densities since the 1970s have been increasing, particularly on the isthmus which was largely built out by that point, but yet we’ve seen public transport use decline during that time, at least until the last 10 years when we’ve seen that trend reverse. Transportation policy is critically important, and – unfortunately – for the past 50-60 years Auckland’s transportation policies have been almost staggeringly roads-centric. Simply banging on an urban limit and encouraging the subdivision and infill of large sections has not been able to overcome such strongly pro-roads transport policies.

Secondly, and perhaps more importantly, measuring density in its simplest way – the whole population of Auckland divided by the area of the city, is misleading in terms of really getting some idea of the suitability of the city for public transport. If one were to, for example, compare Auckland with Vancouver, you would find that the densities in Vancouver are far more ‘lumpy’, with high concentrations around railway stations and the CBD, but with lower densities in other areas. By contrast, in Auckland we generally see (with some exceptions) a relatively uniform density across the whole city. Particularly in newer built areas, there’s a staggering amount of uniformly sized 400-500 square metre sections – generally with a large standalone house in the middle of them. This uniform density tends not to lend itself well to public transport, which works best when there are significant development nodes around train stations, or intensive corridors around high-frequency bus routes. Many of the plans and strategies that Auckland has come up with over the past decade look to remedy this issue, but they are turning out to be extremely slow in actually changing things.

Much is made of how land-use planning influences transport use and options, but perhaps what is a bit ignored is how this process also works in reverse – that transport policy, projects and so forth can enormously influence development patterns. I do have to wonder whether it was planning rules that created Auckland’s relatively uniform density (and thus reinforced its auto-dependency), or whether it was the transportation policies that created the auto-dependency first – and that auto-dependency created our relatively uniform density. This would be largely because there was no real motivation to cluster development around the train stations, because the service was so poor, or bus routes – because there were no bus lanes and the trams had been ripped out in the 1950s.

The relevance of exploring how density, land-use patterns and transportation policies interact is when we look at which transport projects we may be choosing to prioritise. As I have explained in a number of previous posts [link to post about how to prioritise transport projects] accurately working out which project is needed the most, or which will have the best outcomes, is very complicated and difficult. Perhaps a tool that we’re ignoring in that process is looking at which projects will create the land-use outcomes that we’re actually seeking. For example, if we were to build the CBD rail tunnel it is most likely that development in the CBD and in suburbs serviced by the rail network (such as New Lynn) would be more likely to encourage development and intensification. On the other hand, if we are to prioritise projects such as the Puhoi-Wellsford motorway then the land-use outcomes most likely to occur would be further development to the north of Puhoi – taking advantage of the shorter travel times between places like Warkworth, and Auckland.

When one thinks about transport projects in this manner, a lot of the haze about which project makes more sense seems to clear. A lot of work has gone into figuring out where it makes most sense for Auckland to grow, so perhaps when looking at what transport projects we should be focusing on delivering, closer inspection needs to be made into which projects deliver the land-use outcomes that we’re most after. This would also enable transport projects to be more ‘forward-looking’, as the current ‘cost benefit analysis’ approach is very much fixated upon a ‘what’s the biggest problem in need of fixing’ way of thinking, rather than a ‘what are we really going to need in ten, twenty or thirty years?’ approach.

Share this


  1. From what I understand a lot of the whole Density = Public Transport argument comes out of an ongoing piece of research that has been conducted by Newman and Kenworthy. This suggests across world cities there is a relationship between population density and petrol consumption (and by extension, public transport usage presumably).

    However, the research has been criticised on the basis of the density figures used, most of them use gross residential density, i.e. density of the political region of a city’s administrative boundaries. Good old Paul Mees has published a report saying if you use the actual urbanised area, there is no significant relationship. To their credit the authors have continued to improve and revise their work over the years, and as the do the effect becomes less strong.

    It kinda makes sense, if you look at their graphs the first thing you see is that there is a lot of variation in petrol consumtion between cities of the same density (and vice versa). Basically the plot suggests that cities of low and high density can have low petrol consumption, and secondly that increasing density in acity like Auckland need not necessarily decrease petrol consumption (as there are some cities with two or three times the density with similar levels of per capita fuel consumption).

    From what I understand is that the density of Melbourne and Toronto is very similar, and they both have developed PT infrastructure. However the PT mode share in Toronto is double that of Melbourne. According to Mees the difference simply comes down to the way buses are integrated directly into the Toronto rail network, where as in Melbourne they are largely separate and people only walk or drive to the station.

    I guess the take home message is that while increased population density *can* support high PT usage, it doesn’t necessarily do so, and likewise it is possible to have high PT usage in a low density city. It seems to have been flavour of the decade in planning theory that increasing (gros) population density will lead to improved public transport and sustainability outcomes, but now it seems like things are coming around with the realisation there are much more significant factors at work.

  2. Good points Nick. Newman and Kenworthy’s research has been fantastic in terms of providing the grunt work to back up reasons for not encouraging sprawl. However, clearly the question is ‘how does one measure density?’, and that’s much more complex than people per hectare.

    Interesting comparison between Melbourne and Toronto actually. I tend to think of Melbourne as a city that ‘has it sussed’, although it’s interesting to see it in a different light.

  3. I see Melbourne as a city that has the latent potential to have it sussed, but it actually has some of the very same issues as Auckland, not least the fact that it still has pretty much the same network it did in the 1930s. At first glance with some 16 radial rail lines feeding a four track city loop, what is arguably the worlds largest tram network and a model ticketing and fare system to integrate it all it should be perfect. Indeed the recent patronage growth has been phenomenal, but currently the system carries a fraction of what it did half a century ago and indeed much less that some systems with half the route length.

    The secret shame here is the abysmal state of the bus service, many routes have pathetic headways (45 mins, one hour) very early closing times and the majority do not run on a Sunday whatsoever. Bus to train integration, the backbone of the Toronto system, is nonexistent barring a few special cases (the shuttle to Monash university campus I use each day for example). While many stations have a bus stop adjacent (one step ahead of Auckland), the services are by no means co-ordinated. Bus to tram integration doesn’t exist except by the occasional coincidence of routes.

    There is also a systematic lack of integration between tram routes and train lines in the outer suburbs, a legacy of ancient commercial competition. See here in my neighbourhood for example:,174.769135&sspn=0.378274,0.617294&ie=UTF8&ll=-37.883389,145.063992&spn=0.046675,0.077162&z=14

    See how the ‘Carnegie’ tram along Glenhuntly Rd stops on a residential suburban corner about a kilometer short of Carnegie town centre and train station. Scroll up, it happens again with the tram along Waverley Rd stopping about 1km short of East Malvern Station, and just north again with the Wattletree Rd tram and Darling Station. Ludicrous that they have never seen the need to make the systems interconnect.

    Basically, anyone who is not within about 500m of a tram stop or train station can’t really use public transport. Commuters park and riding at train stations overcomes that to a certain extent, but there are huge swathes of Melbourne built since WWII that have very limited access to any public transport except a bus service (if they are lucky) that runs occasionally on weekdays.

    There are two cities here, the star shaped transit oriented one that I live in, and the car and freeway based one filling in the gaps in and around it that most of the populace live in. The problem is the failure to reconcile the two, despite the fact they effectively cover the same area.

    One last scary statistic for you, according to the latest census 80% of Australians never use public transport for any reason, let alone regularly.

  4. Yeah I have heard that about Melbourne quite a few times before (probably from you actually) that while the train & tram systems are pretty good, the city is let down by a very very poor bus system. While trains and trams are the high-profile bits of a public transport system, in many ways buses are the “bread and butter”, and in many cities seem to make up the majority of public transport users.

    Fortunately (for Melbourne), I think a bad bus system is probably easier and cheaper to fix than a poor rail system. Co-ordination doesn’t really require much extra money, it just requires someone to sit down and think things out. However, I guess the worry is that Melbourne’s bus system is so highly stigmatised by now that perhaps it’s almost beyond help?

  5. Maybe it is implicit in what you both argue but I think that PT usage has a significant level of “history” or “inertia”. For example, if you already own a car and have driven to work everyday for the past 5 years then you are much more likely to drive a car tomorrow than take the bus. Likewise, if you already ride the bus then tomorrow you will probably ride the bus… even if it was a bit late today. Indeed, a “hardcore” PT user will (just like a bank customer) will put up with quite a lot before they switch back to their car (or switch bank), but over time if the PT service is bad then people will indeed switch.

    Without knowing anything… I would argue that the different levels of PT usage achieved by a city are heavily dependent on their history (but the difference between today and tomorrow depends quite weakly on the quality of the PT service and congestion on the roads etc.). I would argue that density feeds into the difference between today and tomorrow albeit weakly (high density = close to bus stop = more likely to switch from car to bus). Running with this idea… over a long period of time you might find that higher density cities tend to develop higher levels of PT usage but obviously other things influence this (eg. investment PT infrastructure).

  6. I think the urgument is very complex… For example Perth has used very low density to their advantage by having widely spaced train stations with good feeder buses and park and ride it means their trains spend most of their time flying down the middle of their motorways past gridlocked cars at 130km/p/hr (what better way to get people to shift)…

    Having just got back from Welly I’m quite convinced that electrified PT leads to much better transport planning, ergo better land use planning…

  7. Yup it sure is complex Jezza. I think the thing is with Perth is that the southern line does not run through continuous urban development, but instead after a while of sprawl basically services a bunch of satellite towns.

    Auckland could do something similar if we go the Southern Line up to speed, with Pokeno, Te Kauwhata & Huntly being great satellite towns of potentially around 15,000 people. You would need excellent rail service – and probably no more than an hour’s trip length (I think Perth’s southern line takes 48 minutes, average speed: 90 kph). We would also need to triple track the whole southern line so trains didn’t get caught behind all-stopping suburban services.

Leave a Reply

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