There’s a fascinating ongoing debate in planning circles about the question of how important density is to the likelihood of people using public transport (or a form of active transport, like walking and cycling). I’ve blogged about the issue a number of times before – particularly in terms of the argument put forward by Melbourne-based transport academic Paul Mees: that there are plenty of low density places in the world with good public transport, so we should focus on improving the PT and worry less about increasing density.

This argument seems to be supported by a basic look at the population density of various cities, and how that compares to the public transport usage in those same cities. The table below is an extract from data in Mees’s book: Transport for Suburbia (which, by in large, is excellent).

One limitation of the data above is how it looks only at the entire cities. Inevitably, one then gets into difficult situations working out where a particular city begins and where another one ends. Furthermore, the cities shown above have particular characteristics – aside from their density – that may play a particularly important role in explaining the relationship between their density and their PT use:

  • Los Angeles has very dispersed employment patterns, and spent very little on improving public transport post World War II until relatively recently.
  • New York’s employment is massively concentrated, plus benefits from a huge legacy rail network – almost exclusively constructed in the first half of the 20th century.
  • Las Vegas has mainly grown in the past few decades, with much of its high density residences being in ‘gated communities’ isolated from the rest of the city.
  • Vancouver made a deliberate policy decision many decades ago to not construct any freeways near its inner city.

Data that would perhaps be more useful than comparing the density and PT use between cities would be data that looks at different parts of the same city, and then analysed how the PT use in more dense parts of a city would compare with PT use in less dense areas. This work has, very helpfully, been done for a number of cities in NZ and in North America by Keith Hall – former CEO of the New Zealand Planning Institute. He has kindly emailed to me the presentation he gave at the NZPI conference the week before last (though it is a 21 MB document). A smaller version of the presentation, with all the words and numbers but without the pretty pictures, can be viewed here.

The first step in making such an analysis is splitting up the city into various ‘areas types’, depending on their level of employment/residential density. A table of how this division has been undertaken in an upgrade to New Zealand’s subdivision standards is shown below (modified further by Keith into sub-categories to provide a more fine-grained result):

When you start to overlay the level of PT use and active transport with the measure of land-use intensity, at least in Auckland there seems to be a fairly obvious connection between the two: It’s interesting to note that the above table is based on 2001 census data. With the big upswing in PT usage over the past few years – particularly in ‘Rapid Transit’ usage (the train system and the Northern Express), which often serves outer areas particularly well – it could be quite fascinating to see whether these trends have changed at all.

Looking at detailed data for Toronto, one can see the same trends emerging – although generally at a higher level of PT use all around. Interestingly, it is employment density that seems to be related to PT use to an even greater extent than population density: One particularly useful thing Keith does with this data is make a number of comparisons across cities – which not only shows that there seems a relationship between PT use/active transport and the percentage of the population living in relatively high densities, but also gives a really good indication of ‘over-performing’ and ‘under-performing’ cities in terms of their PT use and active-transport use. As can be seen in the graph below, Auckland is (or was at least) quite clearly an under-performing city. Clearly a result of the near complete lack of investment in public transport for nearly 60 years following World War II:

It’s promising to see Wellington performing so extremely well – for a given level of density it seems that Wellingtonians are much more likely – on average – to take public transport, walk or cycle to work than most other cities.

Ultimately, my feeling is that density does matter when it comes to PT use – particularly employment density. However, in the case of Auckland this doesn’t mean the “city is too spread out for PT to work” – as is often the claim. The graph above clearly shows that Auckland under-performs (or at least did when the data collection was done) in terms of PT and active transport modeshare for its density. In other words, we can achieve a lot without having to up our density – but if we really want to increase PT use significantly, one of the best ways to do that is to have more of the city (particularly more of the employment) in areas that have a density much higher than what most of Auckland has now.

Share this


  1. Some supporting points:

    It is useful to report PT use for journeys to work in a CBD against journeys to work outside one, as a way of explaining overall (regional) PT use. In Wellington’s case, as your chart showed, the regional number is held up (a) because half the workers in the CBD used public tranport to get to work and (b)the Wellington City CBD has about a third of the region’s jobs. Auckland has around 12 percent of the region’s jobs in the CBD, and about half of these use public transport – but if we want to lift overall PT use, we have to move the jobs back into the CBD as well. Hence the CBD Rail Tunnel.

    Three other points:

    * In Wellington’s case, very few people use public transport to get to work if a job is outside the CBD. The area has that in common with Auckland, anyway.

    * When I compared the use of public transport for the central city journey to work in Wellington, I found that the bus catchment (most of Wellington City proper) did about as well as the rail catchment (Tawa, Porirua, the Hutt).

    * If Auckland is to build a CBD rail tunnel, it will also need a strategic commitment to move jobs, and perhaps residences, back into the CBD, for the tunnel’s benefits to be realised. In the past the Byzantine nature of Auckland’s city politics would not have made that possible – everyone was too busy fighting for their local ‘interest’. Now, with a supercity, it should be.

  2. In my view density certainly takes a back seat to PT service quality as a measure of potential PT effectiveness – as you say there’s many less dense places with better PT than Akld.

    I think another good point is the argument (put forward by Chris Harris I think??) that refutes the myth that Akld is “too spread out” and geographically unsuited to PT – Auckland’s natural constraints/choke-points – the harbours, the Tamaki/Whau Rivers etc; make it a place geographically more suited to PT than motorways given the narrowness of the land in places.

  3. I think density is ONE ingredient in the overall pie. There are other ingredients such as frequency, connectivity and policies on both cars and things like how and who organises transport in the city.
    The way I see it is that density is being used as a proxy measure for ease of access. The unspoken assumption within this is that access must be by walking to the (train/bus) station. Of course this limits the catchment area to 800m, which is not very useful. Access could be by walking but also by car to the station, bicycle or feeder bus, by doing this you increase the catchment area and collect more people for a given density. The price and availability of car (and bicycle- no one has looked into parking for bicycles!) parking might also have an impact there as well.

    One main objection I have to these types of graphs is that they solely focus on the journey to work.
    Is there a reason for the planning profession to be only doing this, or is this like the proverbial story of a person trying to find their lost car keys under a streetlamp because that’s where the light is and they actually lost their keys elsewhere?

    A good public transport system is used well in both the off-peak and peak periods. I’ve noticed that Brisbane has relatively high (for Australian Cities) journey to work mode share, however I strongly suspect that this isn’t the full picture. Perhaps I am biased, but apart from the BUZ routes, public transport in Brisbane is shocking. A very significant portion of the bus network consists of peak-hour only express and rocket bus services, and it is the same with the train system. Outside of peak hours, service falls off a cliff, and it becomes fiendishly difficult to get anywhere without leaving home at least one hour before you need to go anywhere (because trains come every 30 minutes in the off-peak, and buses generally run every 30 minutes to one hour off-peak if they run at all) and your total journey time can easily be plus or minus 30 minutes for this reason.

    It would also be interesting to see a similar analysis performed on Adelaide and Perth.
    Perth’s train system carries 5 times more trips than the Adelaide system, despite Adelaide being higher density, going to more places, having more stations and a similar population as Perth.

  4. I know I have posted this before, but maybe worth doing again:
    Application of a commuter railway to low density settlement, Peter Martinovich, Public Transport Authority, WA.

    Some quotes:

    “Perth’s low density settlement, limits numbers who walk to a train station, to a max of 500 daily.”
    “Average Dwelling Density – 10/ha”

    “To make a transit system productive when homes are not intensely concentrated , facilities like park-and-ride lots should be provided to encourage people to get themselves concentrated ,
    but on their time, not the transit system’s. In this way, a transit system can start out well loaded, offering reasonably fast, high quality service that will be productive even in low density areas”

    “Traditional mass transit achieves its “mass” through penetration of high urban density.
    In low densities the “masses” must be brought, or come to the railway.”

    None of this is to say that density doesn’t matter. But low density isn’t a “get out of jail for free card” that it seems to be used for sometimes.
    Low density is more of an argument to adapt transport systems to this land use, rather than an argument against public transport full stop.

    1. the type of system described there works very well for suburb – CBD work trips but useless for anything else. Very similar to Wgtn system. However in Auckland the rail system is envisaged for regular off peak use, and for journeys that don’t include the CBD. The rail system serves many town centres/nodes; and strengthening these nodes will further aid the aim.

  5. I’m not sure where you get your information stating that LA is more densely populated than NYC (I’ve read this on previous posts). New York City is BY FAR the most densely populated city in North America, especially Manhattan.

    1. It depends on what your definition of New York is. If you just take the five boroughs, then of course New York is by far the most densely populated city in the USA. However, if you take the entire metropolitan area (a better measure in my opinion) then New York’s density falls away dramatically because much of its outer suburbia (in New Jersey, Long Island and upstate) is particularly low. By area, New York is by far the world’s largest city.

      1. Really? I always thought the LA metropolitain area was far larger than New York’s (in area, not population) as it’s very sprawling, but I guess it does depend on how far away you include satellite towns in the metro area. Even the New Yorks suburbs (those in New Jersey and surrounding counties of NY) are still quite dense, especially in NJ. I know there are places included in NY’s metro area which are 2hrs drive away. I was also surprised that the transit share of NY was only 24%, as all the information I have read has put it at 50% or more (75% for Manhattan), but I guess that must just be NYC itself and not the suburbs.
        Very interesting article anyway.

        1. Here is a quote from Wikipedia (yes, I know) that cites the US Census bureau:

          “Los Angeles has a long-standing reputation for sprawl. However, according to the 2000 Census, the “Los Angeles-Long Beach-Santa Ana” Urbanized Area had a population density of 7,068 inhabitants per square mile (2,729 /km2), covering 1,668 square miles (4,320 km2) of land area, making it the most densely-populated Urbanized Area (as defined by the United States Census Bureau) in the United States.[4] For comparison, the “New York–Newark” Urbanized Area as a whole had a population density of 5,309 per square mile (2,050 /km2), covering 3,353 square miles (8,684 km2) of land area.”

Leave a Reply