Victoria’s Twenty Largest Clusters of Disadvantage (2024)

(Disclaimer: all views/comments are my own)

In this post, I’ve mapped out Victorian disadvantage in a different way to the usual postcode-based approaches.

Postcode-based reports like Dropping Off The Edge are commonly used by policy-makers to analyse socioeconomic disadvantage. However, they can be misleading due to the arbitrary nature of postcode boundaries. Below is an example. The first panel shows the levels of disadvantage across Werribee. The red patches on the left are the low-income neighbourhoods around the former Heathdale Housing Commission estate. They are home to about 7,000 people. The second panel shows the levels of disadvantage when averaged by postcode. The disadvantaged parts of Werribee disappear within the large ‘3030’ postcode, which has a population of 100,000 and includes the high-income areas of Point Cook. Meanwhile, the tiny ‘3026’ postcode (Laverton North) emerges as the area’s apparent priority for disadvantage, even though it has virtually no residents — it’s an industrial estate with a caravan park in one corner.

In the list below, I’ve addressed this issue by mapping disadvantage at the most detailed level available: the ABS’s ‘SA1’ level. It divides Victoria into 14,000 neighbourhoods of roughly 150 households each. I’ve classified these areas as ‘disadvantaged’ if they’re in the bottom 10% of the ‘SEIFA’* Index of Relative Socioeconomic Disadvantage, and grouped them together if they’re adjacent or within a kilometre of each other. Because of the concentrated nature of disadvantage, the twenty largest such clusters account for most of the state’s disadvantaged neighbourhoods — approximately 70% of the total. The remainder are mostly found in small country towns. For more details, see the notes at the end.

*Socio-Economic Indexes for Areas (ABS 2033.0.55.001)

See Also

Toorak - The most expensive suburb

As a general rule, we don’t have a lot of good, up-to-date data on socioeconomic disadvantage. One of the silver linings of the COVID-19 crisis is that it has prompted the development of some more detailed and timely datasets (see for example the ABS list here). Below are some notes on the far-from-perfect data used in this post…

• Most of the figures are from the 2016 census, except the unemployment rates (see below).
• The areas of disadvantage only include SA1s that have a population of more than 100 and are located within an Urban Centre and Locality (UCL). This is because large, low-density SA1s can be misleading when mapped. This made very little difference in the end: 99% of SA1s in the bottom 10% of the index were high density and within a UCL.
• The variables for the Index of Relative Socioeconomic Disadvantage are set out in the table below (taken from the technical paper).

• Note that the variables are proportions. Obviously not everyone who lives in a disadvantaged area is disadvantaged. For example, roughly 25% of the households in the clusters in this list earn more than the Australian median household income of $85,000 a year.
• There is a known issue with the census data in Robinvale, which has a large seasonal workforce. Based on non-census estimates, Robinvale could have approximately 3,500 people living in the bottom 10% of the index, which would put it in a similar position to Maryborough.
• The estimated 2019 unemployment figures were calculated using a two-stage process. First, I looked at the change in unemployment between 2016 and 2019 for each SA4, using the SA4 results from the census and the twelve-month averages from the December 2019 release of the ABS Labour Force Survey. Second, I looked at the change in unemployment for each SA2, using the 2016 and 2019 averages from the Small Area Labour Markets (SALM) product. For each SA1 in a disadvantaged area, I assumed the change in unemployment since the census was the average of the SA4 change and the SA2 change. The reason for averaging the figures, instead of just using the SALM data, is that the SALM data isn’t complete and there are debates about its reliability (because it uses a mix of administrative and survey data). The table below shows the intermediate calculations.

• The ‘bottom 10%’ threshold for classifying disadvantaged SA1s was obviously arbitrary. Roughly speaking, it captured the parts of Victoria that have experienced persistent double-digit unemployment in recent decades. If you bring the threshold down significantly, to 5% or 2%, you tend to get quite a different list that contains only the most highly concentrated pockets of disadvantage: the inner-city towers, Norlane, Broadmeadows, West Wendouree, Moe, Morwell, etc.
• The one-kilometre threshold for grouping SA1s was similarly arbitrary. It was designed to group together areas that were essentially contiguous but might be separated by a park or transport corridor. But I had to cheat a little, insofar as I excluded two areas that technically met the threshold: ‘Carlton-Richmond’ and ‘Flemington-Footscray’. Both these areas are home to large numbers of people who live in the public housing towers, and those towers are within a kilometre of each other. However, they are separated by areas with high incomes, so it’s hard to describe them as ‘clusters of disadvantage’ in their entirety. If they had been included, they would have been 6th and 7th on the list, with populations of 12,000 and 10,500 respectively. See the map below.

• The graph below shows the total populations of the twenty clusters of disadvantage (noting again that not everyone who lives in a disadvantaged area is disadvantaged).

If you’ve read this far, here are some other pieces I found interesting while putting the data together:

• ‘Loss in the Time of Cholera: Long-Run Impact of a Disease Epidemic on the Urban Landscape’ by Ambrus, Field and Gonzalez — a good example of something that shows up regularly in the data: the degree to which place-based disadvantage persists after an exogenous shock
• ‘We no longer share a common lived experience’ by Larry Summers — the same general issue from a US perspective
• ‘New houses for old: fifty years of public housing in Victoria 1938–1988’ edited by Renate Howe
• ‘An Overview of Brown Coal Mining and Electricity Generation in the Latrobe Valley, from the Establishment of the State Electricity Commission of Victoria to Privatisation’ by Meredith Fletcher.

The key data sources for this research were the 2016 census, Monash University’s Victorian Places site, and the annual reports of the Housing Commission of Victoria.