It’s been pointed out many times before that ranking urban areas is a problematic undertaking e.g. see Is being “the world’s most liveable city” such a big deal?
It ranks 321 suburbs against 15 criteria, mostly related to access to services (see exhibit). The gold medal “winner” is the near-CBD suburb of East Melbourne, followed at an unknown interval by South Yarra and Toorak.
It makes intuitive sense that the most liveable suburbs would be the most sought after. So it’s no surprise the median price of a four bed room house in these suburbs ranges from $2.1 million to a whopping $4.25 million.
But there are some curious results that call into question the usefulness of The Age’s league table.
For example, the suburb of Ivanhoe ranks eighth (median sale price of $1.4 million for a 4 bed house), yet contiguous Ivanhoe East ($1.5 million) comes in way behind at 92 and near neighbour Eaglemont ($1.6 million) at a distant 148.
Another strange result is outer suburban Caroline Springs; despite being a master-planned community with a median price of $0.52 million for a four bedroom house, it’s ranked at 313, just eight places from rock bottom.
And Princes Hill, The Age tells us, plunged forty six places from 62 in the last survey back in 2011 to 16 this time around “thanks chiefly to its NBN connection”. So merely replacing cable and DSL with NBN increases liveability that much?
Something peculiar seems to be going on here, so it’s worth looking at the detail of how the The Age derived its rankings. That presents a problem though.
Although Fairfax frequently rails against public agencies for their lack of transparency, it doesn’t deign to explain in detail the methodology underlying the rankings and it chooses to deny its readers access to the consultant’s report.
We don’t have information on vital aspects like how the criteria are weighted; does living close to the CBD get the same weight as living among hills? Nor do we know the interval between rankings; is twentieth place a whisker away from first place or is it a yawning gap?
Even so, there are some evident problems with the exercise. A key one is the methodology appears to deal poorly with accessibility to shops, restaurants, culture, parks and schools; they only count if they’re located within the boundary of a suburb.
For example, Ivanhoe scores well because it has a number of schools within its boundary; but nearby suburbs like Eaglemont that enjoy high accessibility to these schools by both car and public transport score low on this criterion. (1)
That’s a serious failing. The sensible approach would be to rate each suburb according to the access it provides to key services; (say) something like the % of metropolitan jobs accessible within x minutes travelling time by private/public transport from each suburb.
But that highlights a related and arguably even more serious failing; the criteria don’t include accessibility to employment despite its obvious importance in residential location choice. There’s a criterion measuring closeness to the CBD but it’s not very useful for measuring job accessibility because only around 15% of all jobs in the metro area are in the CBD.
Another inexplicable failing is the absence of any measure of the amount of private indoor and outdoor space offered by the housing stock in each suburb. Yet for most Melburnians private space is a key part of the liveability of a location; indeed it’s one of the reasons Australian cities developed at low densities.
The inescapable problem with these sorts of rankings is that it’s necessary, in order to establish a rank order of suburbs, to assume a standard set of preferences defining an average or typical Melburnian.
But clearly a city centre suburb like Melbourne (it ranked fifth), where 99% of the housing stock is small apartments, appeals to a vastly different idea of liveability than a suburb like Ivanhoe (ranked eighth) where only 12% of dwellings are apartments.
Thus households with children comprise 23% of the population in Melbourne (suburb) compared to 64% in Ivanhoe. These two populations will have vastly different ideas of what constitutes liveability.
It didn’t stop it, but The Age seems to recognise that a league table of suburbs is a dubious exercise.
For those who’re prepared to click through, it also offers a “calculator” that enables readers to find their “perfect suburb” by weighting each of the offered criteria in accordance with their personal preferences. Readers can also set a price limit.
Although there are still serious deficiencies in the criteria it offers, it’s at least recognition that liveability isn’t defined in the same way by everybody; it depends on the preferences of individuals (as mediated by prices). (2)
If The Age is intent on a universal league table, it should start with the simplest metric of all; prices. Selling prices/rents provide a summary statistic of the complex forces that shape why one suburb is preferred over another.
A more useful exercise would’ve been to work backwards from property prices and figure out the key components explaining why prices differ between suburbs.
Ivanhoe improved from 32 last time to 8 this time, mainly due, The Age says, to schools. The Age notes Charles La Trobe college and Ivanhoe Grammar School are located within the boundary of Ivanhoe (although it doesn’t realise the latter is co-ed) but is unaware (unrelated) Ivanhoe Girls Grammar School is also located within the suburb; if it did, would Ivanhoe score so well?
There are some other oddities too. For example, there are three separate criteria relating to the importance of living near a train stop, a tram stop and a bus stop, respectively; do these “double-count” the other seven criteria on the importance of living close to specific services and facilities? Are suburbs that don’t have a tram service penalised, even though they might have high accessibility by other modes?