Population density of inner city SA2s relative to their area, Sydney – City and Inner South SA4, 2016 (source data: ABS)

The Sydney Morning Herald informed its readers yesterday that the Homebush Bay-Silverwater precinct in Sydney’s west has more in common with gritty, bustling New York than meets the eye.

With 1773 people a square kilometre, Homebush Bay-Silverwater has the closest matching population density to the New York urban area.

Sydney’s parallels with the world’s great and glamorous cities don’t end there:

On the other side of the city, surfside Maroubra, with a density of 5591 people a sq km, most closely matches London’s density of 5600 people a sq km. A few suburbs away, Paddington-Moore Park (4394 people a sq km) most closely resembles the Tokyo-Yokohama urban area (4400 people a sq km), while the Concord-Mortlake-Cabarita precinct (3706 people a sq km) in the inner west is the closest match to Paris (3700 people a sq km).

What are Herald readers to make of these comparisons? Do they confirm the fears of those who complain high levels of immigration are turning Sydney into an ultra-dense city with all the associated problems of reduced amenity, inadequate infrastructure, and poor housing affordability? Or do they indicate Sydney is turning into an exciting, walkable city up there with the world’s leading destinations?

None of that; the claims are rubbish. This is a misleading comparison. Consider that ABS statistics show the Homebush-Silverwater Statistical Area (SA2) cited by the paper covers an area of just 11.3 sq km and has a population of only 19,965 persons. But the New York urbanised area the Herald compares it with is in another universe; it covers a whopping 11,642 sq km and has a population of over 20 million persons.

The same sprite melon versus watermelon problem applies to the other comparisons:

  • The Maroubra SA2’s 33,120 residents live in an area of 5.9 sq km; London’s 10,350,000 residents occupy 1,738 sq km.
  • The Paddington-Moore Park SA2’s 16,366 residents take up 3.7 sq km; Tokyo-Yokohama accommodates 37.75 million people on 8,574 sq km.
  • The Concord-Mortlake-Cabarita SA2’s 23,440 residents occupy 6.3 sq km; Paris’s 10.87 million live on 2,845 sq km.

Homebush-Silverwater’s density of 1,773 persons per sq km looks much less impressive when compared to the smaller City of New York, which has a population density of 10,947 per sq km. NYC is a lot smaller than the urbanised area but it’s still big; it covers 781 sq km and has 8.55 million residents. Manhattan is smaller again – 1.64 million residents occupying 59 sq km, so it’s still much bigger than Homebush-Silverwater – but its density is a staggering 27,826 persons per sq km.

A fairer comparison might be between Manhattan and the Sydney-City and Inner South SA4, made up of 19 inner city SA2s (see exhibit). At 64 sq km it’s much the same area as Manhattan but it has a population of 322,556, giving a density of 4,923 persons per hectare. That’s about one fifth of Manhattan’s density.

Or compare Australia’s densest SA2, Ultimo-Pyrmont, with the neighbourhood of Yorkville on Manhattan’s Upper East Side. With 22,881 residents occupying 1.5 sq km, Ultimo-Pyrmont’s density is 15,346 persons per sq km. Yorkville is similar in area – 1.3 sq km – but has around 80,000 inhabitants, giving it a density of just over 60,000 persons per sq km.

Comparing small populations/areas with larger ones is fraught, but surprisingly common (see More overreach on the problem of high-rise towers?). The problem is small units show a lot of variability; this tends to be averaged out as the population size/area increases. For example, tiny Surry Hills houses 18,021 people within 1.3 sq km at an average density of 13,685 per sq km, whereas the five million residents of Greater Sydney cover 12,368 sq km (incidentally, similar to urban New York) at a density of just 405 persons per sq km.

The silliness of extrapolating from small areas to make a point about large areas is shown by a simple example; a single tower with 500 residents on a one hectare site is equivalent to a density of 50,000 persons per sq km. Australia’s capital cities have many individual residential towers that far exceed the urbanised area population densities of New York, London, Tokyo, and Paris. Quite a few exceed the density of Manhattan or central Paris. But these can’t be sensibly compared to larger areas containing many buildings of various sizes as well as supporting infrastructure and amenities such as parks and streams.

The Herald’s report does eventually acknowledge “there are limits to the comparisons that can be drawn between small neighbourhoods and large urban areas”, but that’s not until after 223 words and two exhibits. Anyway, there’s no indication for the layperson of what those limits are, or how important they are. In characteristic fashion, Fairfax’s sub-editor runs the click-catching but misleading headline and then puts the arse-covering but meaningless qualifier well into the article.

This isn’t just another media botch; it goes to having a well-informed public understanding of key urban issues like density.