“Sprawl” and population density: should we care about the average or the distribution?
I’m quite interested in the debate going on in Auckland at the moment about whether population growth should be accommodated by more “sprawl” or increased density. In particular, people claim Auckland currently has quite a lot of sprawl and the vision set out in the Auckland Plan (and implemented in the unitary plan) is that we will move towards a “compact city” (i.e. increased density). A related discussion is what type of transport investments we should prioritize to accommodate that growth (i.e. roading vs public transport/cycling infrastructure).
I was therefore interested to see the New Zealand Institute attempting to debunk the myth that Auckland has a lot of sprawl by posting the following graph, in their very useful “graph of the week” series
Source: New Zealand Institute
On it’s face, this is a bit of a home run against the argument that Auckland is a sprawling city. Though having been to a few of these cities, the relative rankings started to worry me a bit. For example, Auckland has the same density as Los Angeles and both are apparently more dense than New York.
This illustrates why you always need to look at how statistics are defined, so that you know whether the number is actually answering the question you are asking (which brings us back to the transport/cost of living debate). Which is why I found this follow up post by the NZI very interesting. They note that for New York, the US census figures give a density five times higher than the Demographia figures (used for the graph), but the discrepancy can be explained as follows
Indeed, the 2010 US census put New York’s population density at 26,953 people per square mile. That’s five times denser that the statistics published by Demographia.
The difference is easily explained. The US census measures population density inside New York’s metropolitan urban limit – the formal boundaries of the bustling and chaotic City that Never Sleeps. This is an area of 303 square miles (488 square kilometres).
Demographia measures New York’s population density over a much larger area – an area of 4,406 square miles (7,091 square kilometres). It uses techniques like satellite photos of suburban lights on a clear night to assess the extent of urban sprawl beyond the metropolitan limit.
They then go on to argue that the Demogrpahia measure is more appropriate for assessing “sprawl”
If the purpose is to assess the extent of urban sprawl, it does not make sense to stop at the city limit if this is not where the sprawl ends. Using photos of the lit-up areas at night to gauge where the built-up areas end makes sense.
That is why the Graph of the Week focused on Demographia’s measure of density. It shows that Auckland’s population density must be greater in the outskirts of Auckland than in the outskirts of New York’s metropolitan limit [emphasis added]
Which brings me back to the point of what question we are trying to answer…what is “sprawl” and why do we care about it? What the two numbers illustrate to me is that New York has a very uneven distribution of population density (i.e. a very concentrated “core”). Whereas Auckland might have a more uniform distribution of population density (I don’t think I can make any firm conclusions on this without more data).
So the question is, when considering “sprawl” in the context of “an ever-growing population, increasing fuel costs and environmental concerns” as the NZI note in the opening paragraph of the linked article, what is the right measure? Both probably matter, but my hunch is that the distribution of density may be more important than the aggregate average, since it tell us what the “core” of the city looks like. A big city (geographic and population wise) with a concentrated core and very few people living on the periphery will have very different problems to a city of the same size that distributes its population evenly over the same area.
Update: The very knowledgeable folks over at Auckland Transport Blog have pointed out the concept of “weighted density” via twitter. This measures captures what I was talking about above. If you’re interested, check out a post they have done on the topic here as well as some Aussie stuff here and here. I also found this article in my follow googling quite useful.
Article updated to discus “population weighted density”, thanks to some helpful pointers by the people at @AkTransportBlog
Peter Hall, Alain Bertaud, Patrick Troy, Peter Gordon, and probably others, have pointed out that urban growth containment policies that force up the price of land, cause a distortion in the distribution of density.
The lowest cost land is of course always at the urban fringe. Land cost is part of the cost of housing. Households have to make a trade off between the cost of land and the cost of transport. The higher the cost of land is, the MORE people tend to be “priced out” of the more central areas, and the denser “fringe” housing gets.
You can literally see on Google Earth, the difference between US cities without urban growth boundaries, and cities with them. Those without boundaries, have the density at and nearer the core and the density tapers off gradually and almost imperceptibly into the surrounding countryside. But those with boundaries, have “10 to the acre” developments, and even denser, right at the urban fringe and nearby.
This unintended consequence actually increases the average distance of the population, from the centre of the city. “Intensification” at efficient locations is slowed down because of the sheer cost of the land. Land becomes a speculative commodity held by people who have little or no interest in its efficient use, just as people hold gold. The prospect of capital gains is everything.
This is why planners in the UK insist there is “not a shortage of land for housing”; there are empty properties all over the place….! The fact that the average age of a first home buyer is now 39 years, or that the price of land is about 300 times higher per square foot than most US cities, does not matter to them.
Matt, seeing you are interested, research the densities of cities on the Demographia “World Urban Areas” data base; and the traffic congestion delays by city in the TomTom Annual International Congestion Index. The correlation is obviously between density and congestion delays. It is also between density and unaffordability.
Auckland’s peak congestion delay is 43 minutes per hour of driving, which is worse than every city in the USA. The worst in the USA is LA, which figures – LA being their densest city. WELLINGTON at 39 minutes, follows only LA at 41 minutes, and is worse than every other city in the USA.
For Pete’s sake, US cities of 1 million population and less, have delays of around ONE THIRD of what Auckland and Wellington do. They also have house prices around one third. Their density is typically half or less, of our cities.
Kiwis need to wake up to the fact that growth containment urban planning is based on a pack of lies that are totally the opposite of reality. The underlying problem is “garbage in, garbage out” computer models that have never been validated against real world outcomes.
The other nonsense is that infrastructure costs are more sustainable when growth is contained. There is no evidence whatsoever that less dense cities in the USA, with continued freedom to grow, have any correlation with fiscal difficulties. There are fiscal difficulties being encountered by local governments all over the world, but it is absurd to suggest that low urban density has any correlation. In fact, cities with “growth containment” are probably MORE certain to be in fiscal difficulties.
All that actually matters for efficient urban function; is dispersion of employment and amenities; a low, flat land rent curve minimising the “pricing out” effect on any particular participant in the urban economy; and good intensity of road lane miles sufficient for the dispersed travel patterns. Radial highways are a mistake. What is needed is lots more 4-laned suburban connectors.
Increase one lane to two; and capacity is doubled. Increase two lanes to three, and 50% is added. And so on. Dispersed jobs and residences use the network in both directions at all times, which is like a de facto capacity increase.
The sheer cost of land acquisition for roads and other infrastructure is prohibitive when it has been left too late. Planners should be starting from scratch with greenfields developments that do not repeat the mistakes of the past, and allow plenty of rights of way for future expansion, not build out every square inch of land. Congestion delays from intensification without road capacity expansion always more than negate the alleged benefits of “mode shift”, and mass public transport’s efficiency advantage over modern cars is grossly over-rated anyway.
Most of an urban area has to be something other than housing anyway, so doubling the intensity at which people live does not halve average travel distances. Doubling the intensity at which people live, requires twice as much space locally for schools and parks and public facilities and retailing and so on. The actual real life “saving of distance” is about 7 to 10 percent for each doubling of the intensity of housing.
The more geographic features that are unable to be built on, the worse this effect gets. Auckland and Wellington’s geographic difficulties are a REASON to run with dispersion, not concentration. It is absurd to start with “commuter rail mode share” as your planning objective in itself and expect the entire economy and society to order itself around that. The unintended consequences have many times greater costs than the alleged benefits are worth, and I say “alleged” because they will never be realised in practice.
Hi Phil, A couple of quick reactions:
1) While Matt is the main
author here at TVHE, he didn’t write this, there are a couple other
authors lurking in the background:)
2)
The post was not arguing for a binding urban limit and increased
density…I have an open mind on whether urban limits are a bad thing. As an economist I’m naturally suspicious of arbitrary caps on anything! The point of this post was simply to discuss the problems with measuring density at
an aggregate level, and how arbitrary changes in the boundary can
dramatically change the number, as demonstrated by the difference in the
demographia and US census stats. In that context, the evidence you have
provided that a binding urban limit causes a build up in density at the
boundary is quite interesting.
3) People more knowledgeable than
me would argue that using LA as an example actually contradicts the
point you are trying to make? I don’t have any data, but isn’t LA
usually cited as one of the worst examples of sprawl and low PT usage? Isn’t Auckland trying to do the opposite? I just did some
very brief googling and found this article
(http://www.uctc.net/access/37/access37_sprawl.shtml)
which points out
that LA has a very flat distribution of density, despite having a very
high average density. This reinforces my point above about the distribution,
not the average mattering.
Appreciate you taking an interest. I just couldn’t resist raising the broader topic in which these misinterpretations of density are taking place.
Longer comment here:
http://www.newgeography.com/content/003856-the-evolving-urban-form-portland#comment-23924