How and why does driver’s licence ownership vary across Sydney?

Sat 27 February, 2021

In a recent post I confirmed the link between driver’s licence ownership and public transport use at the individual level in Melbourne:

Unfortunately, spatial data around driver’s licence ownership is quite scarce in Australia, so not a lot is known about the spatial variations of licence ownership, nor what might explain them.

However, Transport for New South Wales do publish quarterly licensing statistics at the postcode level, and so this post takes a closer look at the patterns and possible demographic explanations of driver licence ownership across Sydney. I’ll also touch on the relationship between licence ownership and journey to work mode shares.

I have measured rates of licence ownership at the postcode level, and then compared these with other demographic factors that have shown to be significant in explaining variations in public transport mode shares in Melbourne (see my series on “Why are young adults more likely to use public transport”, parts 1, 2, and 3). These factors include socio-economic advantage and disadvantage, workplace location, age, recency of immigration, educational attainment, parenting status, motor vehicle ownership, population weighted density, proximity to high quality public transport, English proficiency, and student status.

I’m sorry it’s not a short post, but I have put some less profound analysis in appendices.

About the data

To calculate licence ownership rates you need counts of licences and population for geographic areas for the same point in time (or very close). Estimates of postcode population are only available from census data, so for most of the following analysis, I’ve combined 2016 “quarter 2” driver’s licence numbers (which includes learner permits) with (August) 2016 ABS census population counts. This is of course pre-COVID19, and patterns may (or may not) have changed since then.

I’ve mostly used population counts for persons aged 16-84. Obviously there are people over the age of 84 with licences, but I am attempting to discount people who may lose their eligibility to hold a licence due to aging.

I’ve also mapped postcodes to the Greater Sydney Greater Capital City Statistical Area boundary, and filtered for postcodes with a significant region within the Greater Sydney boundary (note that the boundaries do not perfectly align).

How does driver’s licence ownership vary across Sydney?

Here’s a map showing 2016 licence ownership rates for Sydney postcodes, with red signifying very high ownership, and green very low.

Technical note: For this map I have filtered to only show postcodes averaging at least 3 persons per hectare to focus on urban Sydney, but some excluded postcodes will be a mix of urban and non-urban land use so this is imperfect. Postcodes are not a great spatial geography for analysis as they vary significantly in size, but unfortunately that’s how the data is published (much easier for TNSW to extract I am sure).

The lowest licence ownership rates can be seen in and around the Sydney CBD, around major university campuses (especially UNSW/Randwick, Macquarie Park, University of Sydney/Camperdown), and at Silverwater (which includes a large Correctional Complex – inmates probably don’t renew their licence and would have a hard time gaining one!). There are also relatively low rates in some inner southern suburbs, in and near Parramatta, and near Sydney Airport.

Most outer urban postcodes have very high levels of licence ownership. One exception is postcode 2559 in the outer south-west, which contains a large public housing estate in the suburb of Claymore. More on that shortly.

Is there a relationship between licence ownership and journey to work transport mode share?

It will probably surprise no one that there was a relationship between driver’s licence ownership and private transport mode share of journeys to work. The following chart shows the average postcode mode share for the commuter population within specified bands of driver’s licence ownership.

I should point out that this a relationship, but not necessarily direct causality (either way). People might be more likely to get a driver’s licence because that is the only practical way to get work from where they live, and other people who do not want to – or cannot – get a driver’s licence may be able to choose to live and work in places that don’t require private transport to get to work.

And then there are some postcodes with pretty much saturated driver’s licence ownership but less than 60% private transport journey to work mode shares (top right). I’ll have more to say on these postcodes shortly.

The rest of this post will consider potential explanations for the spatial patterns of licence ownership, using demographic data for postcodes.

Socio-economic advantage and disadvantage

The following chart compares licence ownership with ABS’s Index of Socio-economic relative advantage and disadvantage (ISRAD, part of SEIFA), at the postcode level:

Near-saturated licence ownership was more common in the more advantaged postcodes, but lower rates of licence ownership were seen in postcodes in deciles 1, 7, and 8. Decile 1 stands to reason as areas of disadvantage (probably including many people unable to get a driver’s licence, eg due to disability), and the postcodes with very low licence ownership rates in deciles 7 and 8 contain or are adjacent to major university campuses.

However there are postcodes with licence ownership rates below 80 in all deciles – the relationship here is not super-strong and there are many exceptions to the pattern.

For people less familiar with the demographics of Sydney, here is a map showing 2016 ISRAD deciles for Sydney postcodes. Note that these deciles are calculated relative to the entire New South Wales population, and Sydney overall is more advantaged than the rest of the state, hence more green areas than red.

Workplace location

Workplace location is a known major driver of commuter mode share, with people working in the CBD much more likely to commute by public or active transport (see Why are younger adults more likely to use public transport? (an exploration of mode shares by age – part 2, plus analysis below). So how does it compare with licence ownership?

Here’s a scatter plot that shows that relationship. I’ve added socio-economic advantage and disadvantage colouring for further context, and labelled selected outlier and cloud-edge postcodes (unfortunately there is a slight bias against labelling postcodes containing many suburbs).

There is perhaps a weak relationship between work in Sydney CBD percentage and licence ownership, with postcodes containing larger shares of commuters going to the CBD (30%+) having lower licence ownership.

The chart also shows that disadvantaged postcodes generally had both fewer CBD commuters (as a proportion) and lower rates of licence ownership.

Commuter mode shares were much more strongly related to workplace location than licence ownership, as the following chart shows. Note that for this chart colour indicates licence ownership rate.

Within the main cloud, postcodes with lower rates of licence ownership (shades of orange) had slightly lower private transport mode shares and/or slightly lower percentage of commuters heading to the CBD. The upper outliers from the cloud include many wealthy postcodes that were not well connected to the CBD by the train network, while postcodes in the bottom-left of the cloud are on the train network.

To explore that further, here’s a similar chart, but with the data marks coloured by a relatively blunt measure: whether or not the postcode contained a train or busway station (based on point locations for stations, which is not perfect as some postcodes are very large and only part of the area might be within reach of a station, while other postcodes might have a station just outside the area):

Generally the postcodes with a train or busway station are towards the bottom-left of the cloud, and those without towards the top-right. I’ve labelled a few exceptions, which include university suburbs such as Macquarie Park, Kensington, Camperdown, and some larger postcodes where a station only serves a minority of the postcode area (eg 2027 and 2069).

The next chart plots commuter mode shares, licence ownership, and socio-economic advantage/disadvantage:

You can see a significant – but not tight – relationship between licence ownership and commuter mode share. Within the main cloud, disadvantaged postcodes are to the top-left, and the more advantaged postcodes to the bottom-right. That is, many disadvantaged postcodes had high private transport mode share despite lower licence ownership, and many more advantaged areas had lower private mode share despite higher licence ownership.

This suggests licence ownership was not the strongest driver of commuter mode choice, at least at the postcode level. Workplace location seems far more influential.

Many advantaged areas are closer to CBD(s) and often have higher quality public transport, walking, and cycling options. People in more advantaged areas are also more likely to work in well-paying jobs in the central city, where public transport is a more convenient and affordable mode. These people also probably face fewer barriers in obtaining a driver’s licence for when they do want to drive (eg access to a car).

While disadvantaged postcodes generally had lower rates of licence ownership, fewer people in these postcodes worked in the Sydney CBD, and they also tended to have high private transport commuter mode shares. I suspect this may be related to many lower income workplace locations being generally less accessible by public transport (particularly jobs in industrial areas). Any cost advantage of public transport is less likely to offset the relatively high convenience of private transport (not to suggest the design quality of public transport services is not important, and not to go into the issues of capital v operating cost of private transport).

However, I suspect public transport could be more competitive for travel from these disadvantaged low-licence-ownership areas to local schools and activity centres. I am aware of some disadvantaged areas of Melbourne that have highly productive bus routes, but not necessarily high public transport mode shares of journeys to work (particularly parts of Brimbank). These areas may be worth targeting for all-day public transport service upgrades, to contribute to both patronage growth and social inclusion objectives.

Just to round this out, here’s a very similar chart, but with Sydney CBD commuter percentage used for colour:

For most rates of licence ownership, there was a wide range of private transport mode shares and a wide range of Sydney CBD commuter percentages. There is a relationship between licence ownership and mode share, but it is not nearly as tight as the relationship between Sydney CBD commuter percentage and mode share.

Age

There’s obviously a relationship between age and licence ownership and NSW thankfully publishes detailed data on licence ownership by individual age. The following chart shows licence ownership by age, animated over time from 2005 to 2020.

Licence ownership peaks for ages around 35-70, and is lower for younger adults and tails off for the elderly as people become less capable of driving.

But there is a very curious dip in licence ownership around age 23-24, which became more pronounced after around 2008. Why might this be?

One hypothesis: People getting learner’s permits around age 18 but not progressing to a full licence and having their learner’s permit expire after 5 years – i.e. around age 22 or 23. I wonder whether people are getting a learner’s permit largely for proof of age purposes. NSW does have a specific Photo Card you can get for that, but the fee is $55 (or $5 at the time you get your driver’s licence), whereas a learner’s permit costs just $25 (and an Australia Post Keypass proof of age card costs $40). As of September 2020, there were 185,329 people aged 18-25 with a Photo Card, and 211,004 people aged 16-25 with a learner’s permit (unfortunately data isn’t available for perfectly aligning age ranges). Did something change about proof of age in 2008? I don’t live in Sydney but maybe locals could comment further on this?

However, I think I have uncovered a more likely explanation which I’ll discuss in the next section.

It would stand to reason that postcodes with more people in age ranges with lower licence ownership might have lower rates of licence ownership overall. I’ve calculated the ratio of the population aged 35-69 (roughly the peak licence-owning age range for 2016) to the population aged 15-84 (roughly the age range of most licence holders) for all postcodes to create the following chart:

You can see a very strong relationship between age make-up and licence ownership rates for postcodes (a linear regression gives an R-squared of 0.75). That is, the more the population skews to people aged 35-69, generally the higher the licence ownership rate.

Recent immigrants

My previous analysis found a strong relationship between public transport use and recency of immigration to Australia (see: Why were recent immigrants to Melbourne more likely to use public transport to get to work?). So does a similar relationship apply for licence ownership?

While I cannot directly match licence ownership and immigrant status at the individual level, I can compare these measures at the postcode level.

For the following chart I have classified postcodes by the percentage of residents who arrived between 2006 and 2016 – as at the 2016 census (my arbitrary definition of “recent immigrants” based on available data for this analysis), and compared that with licence ownership levels.

This chart shows a fairly strong relationship, and suggests more recent immigrants were less likely to have a driver’s licence – although the relationships is weaker for more disadvantaged postcodes (red/orange postcodes).

So why might recent immigrants be less likely to have a licence?

  • As we’ve already seen, some of these postcodes with low licence ownership are adjacent to universities, and no doubt included many international students who did not have a need for licence to get to study or work.
  • Many other skilled immigrants would work in the CBD(s), for which high quality public transport connections are generally available. In Melbourne, I found many recent immigrants live closer to the city where public transport is more plentiful, and many also live near train stations. Sydney is likely to be similar (more on that in a moment).
  • For some it might be because they cannot (yet) afford private transport (particularly immigrants on humanitarian visas) and/or that they don’t have sufficient English to get a learner’s permit (more on that later).
  • For some it might be that they are happy and attuned to using public transport, walking and/or cycling to get around, like they did in their country of origin. However when I analysed Melbourne commuter PT mode shares by immigrant country of origin, I didn’t find relationships I expected.
  • The age profile of immigrants skew towards younger adults, who for various reasons are less likely to own a driver’s licence.
  • I had wondered if some immigrants were driving using international licences instead, but NSW rules state that you can only drive on an international licence for up to three months, so that’s unlikely to explain the pattern.

Here’s a chart showing that immigrants skew towards young adults. The chart shows the New South Wales 2011 population for each calculated approximate age of immigrants when they arrived in Australia (= age + arrival year – 2011) (the best data I have available at present):

The most common ages at arrival were around 23-25 years. Sound familiar? It is also the age where driver’s licence ownership rates dip in New South Wales. I reckon there’s a good chance the influx of immigrants of this age may explain the dip in licence ownership rates for people in their early 20s.

My recent Melbourne research found recent immigrants were also less likely to own a motor vehicle. This evidence suggests low rates of driver’s licence ownership is also strongly related to the relatively high use of public transport by recent immigrants.

For reference, here’s a map showing the percentage of residents in 2016 who had moved to Australia between 2006 and 2016. If you know a little about the urban geography of Sydney, you’ll see higher concentrations around the CBDs, university campuses, and along some major train lines.

Parenting status

We know parents are less likely to use public transport (at least in Melbourne, but probably in all Australian cities), so are they also more likely to own a driver’s licence? The following data compares licencing and parenting rates (defined as proportion of adults doing unpaid caring work for their own children aged under 15) for postcodes:

There is a significant relationship, with postcodes with higher rates of parenting generally have higher rates of driver’s licence ownership. This may well be related to licence ownership rates also peaking for people of the most common parenting ages, and also the fact many young families live in the outer suburbs (where private transport is often more competitive than public transport). The postcodes with the lowest licence ownership rates also have very low proportions of parents (and probably contain many young adults who are studying).

For reference here is a map of parenting percentages for Sydney postcodes:

Motor vehicle ownership

It stands to reason that areas with higher driver’s licence ownership rates might also have higher motor vehicle ownership rates. I’ve calculated the ratio of persons aged 18-84 to household motor vehicles for each postcode, to create the following chart:

You can see the relationship is very strong, with more advantaged (and often near-CBD) postcodes towards the top of the cloud, and more disadvantaged postcodes mostly at the bottom and middle of the cloud.

Silverwater is an outlier – but I should point out that my calculation of motor vehicle ownership only counts people living in private dwellings while licence ownership is for all residents (including the many who resided in Silverwater’s correctional facilities).

There are also a small curious bunch of outliers with around 100 motor vehicles per 100 persons aged 18-84 but only 70-90 licences per 100 persons aged 16-84. These include urban fringe suburbs such as Marsden Park, Riverstone, Oakville, Rossmore, Gregory Hills, Leppington, Voyager Point, Kemps Creek, and Horsley Park. Perhaps these areas may contain farm vehicles that might skew the motor vehicle ownership rates.

While spatial data about licence ownership is unfortunately not readily available for most states of Australia, this chart suggested that motor vehicle ownership (something thankfully still captured by the census, despite ABS trying to drop the question) is a reasonably strong proxy for licence ownership.

Population weighted density

Given postcodes can be quite large (one has a population of over 100,000!), I prefer to use population-weighted density as a metric of urban density (as opposed to raw density). Here’s how that related to licence ownership (note a log scale on the X-axis):

That’s a pretty strong relationship, and of course not unexpected. Areas with higher population density generally have great public transport services, and more services and jobs would likely be accessible by walking, reducing the need for a car or driver’s licence.

Proximity to high quality public transport

I’ve previously confirmed a relationship between public transport mode share and proximity to high quality public transport, so does the presence of high quality public transport also relate to driver’s licence ownership?

As mentioned above, I’ve classified postcodes as to whether or not there was a train or busway station contained within the postcode boundary in 2016. It’s a blunt measure because stations may only serve a small part of large postcodes, or there may be a station just outside a postcode’s boundary that still provides good rail access to that postcode. Some postcodes were also served by light rail and/or very high frequency bus services, just not a train or busway station. I’d love to be able to look at licence ownership by distance from stations, but licensing data is unfortunately only available for postcodes, which does not provide enough resolution.

You can see postcodes with a station generally have lower rates of licence ownership than those without, but there is still plenty of variance across postcodes.

The green postcodes in the top of the left column include Camperdown (University of Sydney, close to the CBD with very high frequency on-road buses), Ultimo (just next to Central Station and the CBD), Kensington (includes UNSW campus, with strong bus (and now light rail) connections), Chippendale / Darlington (wedged between Central and Redfern Stations), and Waterloo / Zetland (very close to Green Square Station and also served by high frequency on-road buses).

Many of the postcodes with stations but high licence ownership (bottom of right hand column) are in the outer suburbs, where train frequencies may be lower, and public transport services in non-radial directions may have lower quality.

So the exceptions to the relationship are quite explainable, and I’d suggest there is a strong relationship. Again, it may be people without a licence choosing to live near public transport, and/or people not near high quality public transport deciding they must have a licence to get around.

Educational qualifications

I have also found a relationship between educational qualifications and commuter mode shares in Melbourne, so are licencing rates related to levels of educational attainment in Sydney?

There’s not much of a relationship happening here between licence ownership and education, other than some inner city postcodes with a high proportion of educated residents and lower rates of licence ownership. There is of course an (expected) relationship between advantage and education.

But just on that, one curious outlier postcode on the chart is Lakemba / Wiley Park (2195), with 29% of the population having a Bachelor’s degree or higher, but it being in the most disadvantaged decile. This postcode has a large proportion of people not born in Australia, with significant numbers born in Lebanon and Bangladesh. Perhaps this reasonably well-educated but highly disadvantaged population is a product of lack of recognition of overseas qualifications, and/or maybe issues with discrimination.

Distance from Sydney CBD

In Melbourne, distance from the CBD has a strong relationship with mode choice, and I would not be surprised if there was similarly a relationship with licence ownership. However Melbourne only has one large dense employment cluster (the central city), while Sydney has multiple large dense employment clusters which is likely to lead to different patterns (see Suburban employment clusters and the journey to work in Australian cities).

From the first map in this post you cannot see a strong relationship between licence ownership and distance from the Sydney CBD – it is clear that many other factors are influencing licence ownership rates across Sydney (such as proximity to university campuses and employment clusters). Having said that, it seems clear that most “outer” suburban postcodes have high levels of licence ownership, but distance from the CBD is probably not a good proxy for “outer”.

Also some postcodes are quite large, and are a little problematic to assign to a distance value or range from the CBD, and the presence of two large harbours means crow-flies distance to the Sydney CBD is not necessarily reflective of ease/speed of travel to the Sydney CBD.

For these reasons I’ve not crunched data on home distance from the Sydney CBD. With a lot more effort, perhaps a metric could be created that considers travel time to Sydney’s major centres (although these centres vary in size).

Which factors have the strongest relationship with licence ownership?

The factors shown above had the strongest relationships with licence ownership (I tested three other factors which had weaker relationships, covered in the appendices below).

I put all the factors for Greater Sydney postcodes into a simple linear multiple regression model, and without labouring the details, I found that the following factors were significant at explaining postcode licence ownership rates (each with p-values less than 0.05 and overall an R-squared of 0.83), listed with the most significant first:

  • Ratio of population aged 35-69 : population aged 15-84. For every 1% this ratio is higher, licence ownership per 100 persons aged 16-84 is generally 1.0 higher (all other things being equal)
  • Rate of motor vehicle ownership: every extra motor vehicle per 100 persons aged 18-84, there are generally 0.35 more licences per 100 persons aged 16-84 (all other things being equal)
  • People who have a bachelors degree or higher: For every 1% this is higher, licence ownership per 100 persons aged 16-84 is generally 0.18 higher (all other things being equal)
  • Postcodes containing or adjacent to a major university campus or correctional centre. These postcodes generally had 14 fewer licences per 100 persons aged 18-64 (all other things being equal)

Factors that fell out of the regression as not significant were Sydney CBD commuter percentage, presence of a train or busway station, socio-economic advantage/disadvantage, population weighted density, parenting percentage, student status, and percent of population speaking English very well. Of course many of these metrics would correlate with the four significant factors above.

I was a little surprised to see educational qualifications show up as significant, given the weak direct relationship seen in the scatter plot, however the impact was small (0.18) and it may be acting as a proxy for other factors such as proportion of commuters working in the Sydney CBD (which was the “strongest” factor that fell out – having a p-value of 0.11).

This analysis was done using postcode level which has issues in terms of blending populations. It is possible to look at individuals using household travel survey data, and I’ve had a quick look using VISTA data from Melbourne. Without going into full detail in this post, I’ve found stronger relationships with age, sex, household income, parenting status, main activity, distance from train stations, and a weaker relationship with distance from CBD. Maybe that could be the focus of a future post.

I hope you’ve found this interesting.

Appendix 1: English proficiency

Probably related to recent immigrant figures, postcodes with a larger proportion of residents speaking English very well generally had slightly higher levels of licence ownership, although the relationship is not tight:

Curiously though, the relationship seems to be stronger for more advantaged postcodes. Disadvantaged postcodes with lower levels of English proficiency still had licence ownership rates of around 80 per 100 persons aged 16-84 (top-left of the cloud).

As an aside: is English proficiency lower in postcodes with many recent immigrants?

The answer is yes, but lower levels of English proficiency are not always explained by recent immigration. Of course some of the recent immigrants will speak English very well (many settling in places like Manly, Darlinghurst, Waterloo, Pyrmont), while others will not, depending on their country of origin. The large red dot to the bottom-left is postcode 2166, which includes the migrant area of Cabramatta (sorry about the label that overlaps other data points). It would appear that this postcode has many longer term residents who don’t speak English very well (although they might rank themselves as speaking English “well” rather than “very well”, which is below my arbitrary threshold of “very well” plus native English speakers).

Appendix 2: Student status

I have recently found a relationship between student-status and and journey to work mode shares in Melbourne (although yet to be published at the time of writing). So does the proportion of residents (over 15) who are studying have a relationship with driver licence ownership rates?

Here’s a scatter plot, with socio-economic advantage and disadvantage overlaid:

Apart from some exceptional postcodes with larger proportions of students, there appears to be little to no relationship between studying and licence ownership.

1 Comment | Car ownership, drivers license, Migration, Mode share, Sydney, Urban density | Tagged: , , | Permalink
Posted by chrisloader

Why were recent immigrants to Melbourne more likely to use public transport to get to work?

Mon 7 December, 2020

I’ve recently been analysing how public transport mode share varies with age and associated demographic factors. In part 3 of that series, I found that immigrants – and particularly recent immigrants – were much more likely to use public transport (PT) in their journey to work. This post explores why that might be, using data for Melbourne from the ABS Census (mostly 2016).

About immigrant data

The census covers both temporary and permanent residents. I’ve counted all people who were born overseas and came to Australia intending to stay for at least one year as “immigrants”, regardless of whether they were temporary or permanent residents.

It’s worth looking at the number of immigrants living in Greater Melbourne by age and arrival year, as at 2016:

Except for the first and last columns, each column represents 10 arrival years. You can see a significantly larger population of immigrants who arrived between 2006 and 2015, and they skewed significantly to ages 20-39. We know from previous analysis that younger adults are more likely to use public transport, so age is likely to play a role.

But how many immigrants are temporary residents? The census doesn’t include a question about permanent residency, but it is possible to track arrival year range cohorts over time.

The following chart tracks the number of immigrants for arrival year ranges between the 2006, 2011 and 2016 censuses (using Significant Urban Area geography).

If there were a significant number of temporary residents (although still intending to stay at least one year), then you’d see a large drop in the population of people who arrived 1996 to 2005 over time between 2006 and 2011/2016. There certainly was a drop off, but it was a small proportion.

This suggests most migrants end up being long-term residents (including many who enter on temporary visas but then gain permanent residency).

Numbers in all arrival year ranges dropped slowly over time through people leaving Melbourne (and possibly Australia) and deaths (particularly for immigrants from earlier years many of whom would be in their senior years).

Immigrants and public transport mode share of journeys work

To recap my previous analysis, the relationship between immigration year and PT mode share has held for the last three censuses (2006, 2011, and 2016), regardless of parenting status, birth year, or whether the someone worked inside or outside the City of Melbourne (local government area):

So why might recent immigrants be more likely to use public transport? From looking at the data, I think there are several plausible explanations.

To start with, they were more likely to work in the City of Melbourne, and we know journeys to work in the City of Melbourne have much higher public transport mode shares:

They were also more likely to live in areas with lower levels of motor vehicle ownership. Each column in the following chart represents the population of immigrants for a range of arrival years, and that population is coloured based on the motor vehicle ownership rate of all residents in the (SA1) areas in which they live (including non-immigrants). Note: immigrants themselves may have had different rates of motor vehicle ownership to the average of people in the areas in which they lived.

As I’ve mentioned previously, I do not have access to data to calculate the ratio of household motor vehicles to driving-aged adults within immigrant households, but I can calculate the ratio of household vehicles to all household residents (not all of whom may be of driving age).

The following chart shows that more recent immigrants were likely to have much lower levels of motor vehicle ownership that those who have been living in Australia longer.

Aside: Immigrants who arrived in Australia 1900-1945 had much higher rates of motor vehicle ownership than people born in Australia, but they were also all aged over 70 in 2016.

BUT if you look at PT mode shares for each vehicle : person ratio, there is still a relationship with year of arrival (see next chart), so car ownership doesn’t fully explain why recent immigrants were more likely to use public transport.

Looking at other factors, recent immigrants were slightly more likely to live closer to the city centre:

And they were more likely to live near a train station:

However not all recent immigrants to Melbourne lived near the city or a train station. Here’s a map showing the density of persons who arrived in Australia between 2006 and 2016 as at the August 2016 census.

There were significant concentrations in outer growth areas such Point Cook, Tarneit, and Craigieburn. These suburbs also happen to have very well patronised rail feeder bus routes, and unusually higher concentrations of central city commuters for their distance from the CBD.

Recent immigrants were more likely to live in areas of higher residential density:

And they were more likely to work near the city centre:

More-recent immigrants were also more likely to have a higher level of educational attainment than less-recent immigrants, and generally much higher than those born in Australia:

This probably reflects skilled immigration programs favouring people with higher educational qualifications. Indeed 60% of workers who arrived between January 2016 and the August 2016 census had a Bachelor or higher qualification. And we know from a previous post that highly qualified workers were more likely to work in central Melbourne, and were more likely to have used public transport in their journey to work.

Not only were more recent immigrants generally highly educated, many came to Melbourne to study to raise their educational attainment. Here is a chart showing the proportion of immigrants who were full-time or part-time students, by arrival year groups:

I will explore the relationship between student status and journey to work mode shares in an upcoming post.

How did immigrants shift around Melbourne over time?

Could internal migration explain why immigrants shifted away from public transport over time? Using census data across 2006, 2011, and 2016, it is possible to roughly track the population distribution of particular immigrant cohorts (although it’s not perfect because these immigrants may have moved in/out of Melbourne or left Australia between censuses, including temporary residents).

The following map shows the density of immigrants who arrived in Australia between 1996 and 2005 across census years 2006, 2011, and 2016:

In 2006 there were concentrations around the central city and many rail stations. But these concentrations reduced over time, with many of these people moving into other suburbs by 2011 or 2016 (or leaving Melbourne). In particular, many moved to outer suburbs such as Tarneit, Truganina, Point Cook, Derrimut, Craigieburn, Roxburgh Park, and Narre Warren South.

To help summarise these shifts, the following chart shows the distribution of this cohort across census years by distance from train stations, distance from the Melbourne CBD, and the motor vehicle ownership rate of the areas in which they lived:

You can see that they generally moved further away from train stations, further away from the CBD, and into areas that had higher levels of motor vehicle ownership. All these shifts are associated with reduced public transport mode share, and I suspect this pattern would not be unique to those who arrived 1996-2005.

Is there a relationship between PT mode shares and where people were born?

Firstly, here’s a chart showing the birth regions of Melbourne workers who were born outside Australia, by year of immigration (mostly 5 year bands). I’ve used ABS’s country of birth groups, except that I’ve separated North America from the other Americas.

The early half of the 20th century saw significant immigration from Europe, whereas in more recent times this has shifted to Asia, with southern and central Asia now the biggest source of immigrants. (Southern and central Asia includes India, Sri Lanka, Bangladesh, many former Soviet republics south of Russia and all “-stan” countries.)

So do journey to work public transport mode shares vary by immigrants’ region of birth?

There certainly is some variance between birth regions, but not quite what I was expecting. Immigrants from seemingly car-dominated north America had much higher PT mode shares than those born in European countries with reputations for higher quality public transport.

Of course people born in different parts of the world may be more or less likely to work in the City of Melbourne, and might be more or less likely to be parents. These factors strongly influence PT mode shares. So the next chart disaggregates the data by parenting status and work location (note a different X-axis scale used for each work location division).

This birth regions in this chart have the same ordering as the previous chart, but in most quadrants the mode shares are no longer in order (the top-right quadrant being the exception: non-parenting, working outside the City of Melbourne). Southern and central Asia tops PT mode shares for the other three quadrants, and by quite a large margin for City of Melbourne workers.

We know year of arrival into Australia is a significant factor in PT mode shares, and relative composition of immigrants has certainly changed over time. Also, age itself is likely to be a factor. The next chart adds these two dimensions. However, I have had to remove people working in the City of Melbourne, those under 20 and those over 60 – because the population for these categories became too small, introducing meaningless noise.

You can see there was a relationship between year of arrival and PT mode share within each age band, for both parenting and non-parenting workers. Central and Southern America generated the highest average PT mode shares while North Africa and the Middle East often had the lowest PT mode shares.

Here’s another look at that data, but comparing mode shares primarily by age rather than year of arrival. For this chart I’ve (also) removed parenting workers, and those who arrived before 1982, because they are mostly spread across just two 10 year age bands which isn’t really enough to show an age-based trend:

This chart shows that there was certainly a relationship between age and PT mode share for most birth regions (as well as year of arrival), at least for non-parents working outside the City of Melbourne.

I cannot be certain that this pattern also existed for all birth-regions for parenting workers and people who worked within the City of Melbourne, but I have previously shown a relationship between age and PT mode share for these categories (when ignoring birth region), so a relationship is likely.

So even with a changing mix of immigrant sources over time, age (or some other age-related factor) remains a significant factor when it comes to explaining public transport mode shares.

I hope you’ve found this at least half as interesting as I did.

Leave a Comment » | Car ownership, Melbourne, Migration, Mode share, Mode shift, Urban density | Tagged: | Permalink
Posted by chrisloader

Why are younger adults more likely to use public transport? (an exploration of mode shares by age – part 2)

Sun 27 September, 2020

This is the second post in a series that explores why younger adults are more likely to use public transport (PT) than older adults, with a focus on the types of places where people live and work, including proximity to train stations, population density, job density, motor vehicle ownership and driver’s licence ownership.

In the first post, we found younger adults in Melbourne were more likely to live and work close to the CBD, but this didn’t fully explain why they were more likely to use public transport.

This analysis uses 2016 ABS census data for Melbourne, and data for the years 2012-18 from Melbourne’s household travel survey (VISTA) – all being pre-COVID19. See the first post for more background on the data.

Proximity to train stations

Melbourne’s train network is the core mass rapid transit network of the city offering relatively car-competitive travel times, particularly for radial travel. It’s not Melbourne’s only high quality public transport, but for the want of a better metric, I’m going to use distance from train stations as a proxy for public transport modal competitiveness, as it is simple and easy to calculate.

In 2016 younger adults (and curiously the elderly) were more likely to live near train stations:

Almost 40% of people in their 20s lived within one km of a station. Could this partly explain why they were more likely to use public transport?

Well, maybe partly, but public transport mode shares of journeys to work were quite different between younger and older adults at all distances from train stations:

Public transport mode shares fell away with distance from stations, and age above 20 (the 15-19 age band being an exception).

With VISTA data we can look at general travel mode share by home distance from a train station:

There’s clearly a relationship between PT mode share and proximity to stations, but there’s also a strong relationship between age and PT use, at all home distance bands from train stations.

Younger adults were also more likely to work close to a train station. Indeed 46% of them worked within about 1 km of a station:

And unsurprisingly people who work near train stations are also more likely to live near train stations:

The chart shows around 70% of people who worked within 1 km of a station lived within 2 km of a station. Also, 37% of people who worked more than 5 km from a station, also lived more than 5 km from a station.

But again, journey to work PT mode shares varied by both age and workplace distance from a train station:

For completeness, here is another matrix-of-worms chart looking at journey to work PT mode shares by age for both work and home distances from train stations:

PT mode share declined with age for most distance combinations, but this wasn’t true for the 15-19 age band, particularly where both home and work were within a couple of kms of a station. We know from part one that teenagers are much less likely to work in the city centre, so this might represent teenagers who happen to live near a station, but work locally and can easily walk or cycle to work.

If we take age out for a moment, here is the relationship between PT mode share of journeys to work and both home and work distance from train stations:

The relationship between PT mode share and work distance from a train station is much stronger than for home distance from a station.

So while home and work proximity to train stations influenced mode shares, it doesn’t fully explain the variations across ages. So what if we combine…

Work distance from the CBD, home distance from a train station

Work distance from a station is strongly related to work distance from the CBD, as the CBD and inner city has a higher density of train stations:

I expect workplace proximity to a train station to be a weaker predictor of mode share when compared workplace distance from CBD. That’s pretty evident when looking at journey to work PT mode share by place of work on a map:

And even more evident when you look at PT mode shares for both factors (regardless of age):

So perhaps work distance from the CBD, and home distance from a train station might be two strong factors for mode share? If we control for these factors, is there still a difference in PT mode shares across ages?

Time for another matrix of worms:

The chart shows that even when you control for both home distance from a station, and work distance from the CBD, there is still a relationship with age (generally declining PT mode share with age, with teenagers sometimes an exception). So there must be other factors at play.

Population density

Consistent with proximity to train stations and the CBD, younger adults are more likely to live in denser residential areas:

Higher residential density often comes with proximity to higher quality public transport. Indeed, here is the distribution of population densities for people living at different distances from train stations:

The next chart shows the relationship between residential density and mode shares – split between adults aged 20-39 and those aged 40-69:

The chart shows that both age and residential density are factors for journey to work mode shares. Younger adults had higher public transport mode shares for journeys to work at all residential density bands.

Similarly, VISTA data also shows PT mode shares vary significantly by both age and population density for general travel:

Technical note: data only shown where age band and density combination had at least 400 trips in the survey.

Curiously, people in their 60s living in areas with densities of 50-80 persons/ha were more likely to use public transport to get to work than those in their 40s and 50s living in the same densities (maybe due the presence of children?). For lower densities, PT mode share generally declined with increasing age (from 20s onward).

Population density is also generally related to distance from the CBD:

And here is a chart showing how PT mode share of journeys to work varied across both:

The chart shows home distance from the CBD had a larger impact on mode shares than population density. Indeed population density only seemed to have a secondary impact for densities above 40 persons/ha. However, as we saw in the first post, people living closer to the CBD were more likely to work in the city centre, and therefore more likely to use public transport in their journey to work.

Job density

Young adults were more likely to work in higher density employment areas in 2016, where public transport is generally more competitive (with more expensive car parking):

But yet again, there is a difference in mode shares between age groups regardless of work location job density:

So job density doesn’t fully explain the difference in PT mode shares across age groups.

I should add that job density is also strongly related to workplace distance from the CBD:

and workplace distance from train stations:

And putting aside age, PT mode shares for journeys to work are related to both workplace distance from the CBD and job density:

PT mode shares are also related to both job density and workplace distance from stations:

You might be wondering about the dot of higher job density (200-300 workers/ha) that is between 3 and 4 km from a train station. It’s one destination zone that covers Doncaster Westfield shopping centre – a large shopping centre on a relatively small piece of land (almost all of the car parking is multistory – see Google Maps)

Motor vehicle ownership

Are younger adults more likely to use public transport because they are less likely to own motor vehicles?

With census data, it is possible to measure motor vehicle ownership on an SA1 area basis by adding up household motor vehicles and persons aged 18-84 (as an approximation of driving aged people) and calculating the ratio. Of course individual households within these areas will have different levels of motor vehicle ownership.

Using this metric, young adults were indeed more likely to live in areas which have lower levels of motor vehicle ownership (in 2016):

But yet again, the PT journey to work mode shares varied between younger and older adults regardless of the levels of motor vehicle ownership of the area (SA1) in which they live:

Using VISTA data, we can calculate motor vehicle ownership at a household level. I’ve classified households by the ratio of motor vehicles to adults.

VISTA data shows PT mode shares strongly related to both age and motor vehicle ownership (I’ve shown the most common ratios):

You might be wondering why I didn’t calculate motor vehicle ownership at the household level for census data. Unfortunately it’s not possible for me to calculate the ratio of household motor vehicles to number of adults because ABS TableBuilder doesn’t let me combine the relevant data fields (for some reason).

The best I can do is the ratio of household motor vehicles to the usual number of residents (of any age). The usual residents may or may not include children under driving age – we just don’t know.

Nevertheless the data is still interesting. Here is how public transport mode shares of journeys to work varied across different vehicle : occupant combinations for households in Greater Melbourne:

Yes that’s a lot of squiggly lines – but for most combinations (excluding those with zero motor vehicles) there was a peak of PT mode share in the early 20s, and then a decline with increasing age.

The lines with green and yellow shades – where the ratio is around 1:2 or 1:3 – show a sharp drop around the mid 20s. I expect these lines are actually a mix of working parents with younger children, and working adult children living with their (older) parents. The high mode shares for those in their early 20s could represent many adult children living with their parents (but without their own car), while those in their 30s and 40s are more likely to be parents of children under the driving age. So the sharp drop is probably more to do with a change in household age composition.

If we want to escape the issue of children, the highest pink line is for households with one motor vehicle and one person (so no issues about the age of children because there are none present) – and that line has a peak in PT mode share in the mid 30s and then declines with age, suggesting other age-related factors must be in play.

But motor vehicle ownership levels aren’t only related to age. They are strongly related to population density,

..home distance from the CBD,

..and home distance from train stations:

And public transport mode shares are related to both motor vehicle ownership rates and population density (with motor vehicle ownership probably being the stronger factor):

Technical note: for these charts I’ve excluded data points with fewer than 5 qualifying SA1s to remove anomalous exceptions.

Public transport mode shares are also related to both motor vehicle ownership and home distance from the CBD:

And shares are also related to both motor vehicle ownership and home distance from a train station:

In all three cases, PT mode shares fell with increasing levels of motor vehicle ownership, but this effect mostly stopped once there were more motor vehicles than persons aged 18-84.

Drivers licence ownership

I’ve previously shown on this blog that people without a full car driver’s licence are much more likely to use public transport, which will surprise no one. So are younger adults less likely to have a driver’s licence?

VISTA data shows us that younger adults are indeed less likely to have a car driver’s licence, with licence ownership peaking around 97% for those in their late 40s and early 50s, and only dropping to 91% by age 75 (there is a little noise in the data):

So the lack of a driver’s licence by many young adults will no doubt partly explain why they are more likely to use public transport.

Consistent with VISTA, data from the BITRE yearbooks also shows that younger adults have become less likely to own a licence over time:

At the same time, those aged 60-79 have been more likely to own a licence over time.

But do public transport mode shares vary by age, even for those with a solo driver’s licence? (by solo, I mean full or probationary licence). The following chart shows public transport mode shares for age bands and licence ownership levels (data points only shown where 400+ trips exist in the survey data).

PT mode shares peaked for age band 23-29 for most licence ownership levels, including no licence ownership (there isn’t enough survey data for people older than 22 with red probationary licences – the licence you have for your first year of solo driving).

As an aside, there is a curious increase in public transport mode share for those aged over 60 without a drivers licence – this may be related to these people being eligible for concession fares and occasional free travel with a Seniors Card (if they work less than 35 hours per week).

So even younger adults who own a driver’s licence are more likely to use public transport.

But is this because they don’t necessarily have a car available to them? Let’s put the two together…

Motor vehicle and driver’s licence ownership

For the following chart I’ve classified households as:

  • “Limited MVs” if there were more licensed drivers than motor vehicles attached to the household,
  • “Saturated MVs” if there was at least as many motor vehicles as licensed drivers, and
  • “No MVs” if there were no motor vehicles associated with the household.

If there were any household motor vehicles I’ve further disaggregated by individuals with a solo licence and those without a solo licence (the latter may have a learner’s permit). I’ve only shown data points with at least 400 trip records in the category to avoid small sample noise (I am reliant on VISTA survey data).

Except for households with no motor vehicles, public transport mode share peaked for age band 18-22 or 23-29 and then declined with increasing age. So again there must be other age-related factors. However the impact of age is smaller than that of motor vehicle ownership and licence ownership.

Unfortunately driver’s licence ownership data is not collected by the census, so it is not possible to combine it with other demographic variables from the census.

Summary

So, what have we learnt in part two:

  • Younger adults are more likely to work and live near train stations, but that only partly explains why younger adults are more likely to use public transport.
  • Workplace distance from the CBD has a much bigger impact on public transport mode shares for journeys to work than home distance from a train station.
  • Younger adults are more likely to live in areas with higher residential density, but this only partly explains why they are more likely to use public transport.
  • Younger adults are more likely to work in areas with higher job density but this is highly correlated with workplace distance from the CBD, which is a stronger factor influencing mode shares.
  • Younger adults are more likely to live in areas with lower motor vehicle ownership (these areas are generally also have higher residential density and are closer to the city centre and to train stations), but this again only partly explains why they are more likely to use public transport. Motor vehicle ownership appears to be a stronger factor influencing mode shares than population density, distance from stations, or distance from the city.
  • Younger adults are less likely to have a driver’s licence, but again this only partly explains why they are more likely to use public transport.

While this analysis confirms younger adults tend to align with known factors correlating with higher public transport use, we are yet to uncover a factor or combination of factors that mostly explain the differences in public transport use between younger and older adults. That is, when we control for these factors we still see differences in public transport use between ages.

The next post in this series will explore the impacts on public transport use of parenting responsibilities, generational factors (birth years), and year of immigration to Australia.

1 Comment | Car ownership, drivers license, Employment density, Melbourne, Mode share, Urban density | Tagged: , | Permalink
Posted by chrisloader

How is density changing in Australian cities? (2nd edition)

Sun 21 April, 2019

[updated April 2020 with 2019 population data]

While Australian cities are growing outwards, densities are also increasing in established areas, and newer outer growth areas are some times at higher than traditional suburban densities.

So what’s the net effect – are Australian cities getting more or less dense? How has this changed over time? Has density bottomed out? And how many people have been living at different densities?

This post maps and measures population density over time in Australian cities.

I’ve taken the calculations back as far as I can with available data (1981), used the highest resolution population data available. I’ll discuss some of the challenges of measuring density using different statistical geographies along the way, but I don’t expect everyone will want to read through to the end of this post!

[This is a fully rewritten and updated version of a post first published November 2013]

Measuring density

Under traditional measures of density, you’d simply divide the population of a city by the area of the metropolitan region.

At the time of writing Wikipedia said Greater Sydney’s density was just 4.23 people per hectare (based on its Greater Capital City Statistical Area). To help visualise that, a soccer pitch is about 0.7 hectares. So Wikipedia is saying the average density of Sydney is roughly about 3 people per soccer field. You don’t need to have visited Sydney to know that is complete nonsense (don’t get me wrong, I love Wikipedia, but it really need to use a better measure for city density!).

The major problem with metropolitan boundaries – in Australia we use now Greater Capital City Statistical Areas – is that they include vast amounts of rural land and national parks. In fact, in 2016, at least 53% of Greater Sydney’s land area had zero population. That statistic is 24% in Melbourne and 14% in Adelaide – so there is also no consistency between cities.

Below is a map of Greater Sydney (sourced from ABS), with the blue boundary representing Greater Sydney:

One solution to this issue is to try to draw a tighter boundary around the urban area, and in this post I’ll also use Significant Urban Areas (SUAs) that do a slightly better job (they are made up of SA2s). The red boundaries on the above map show SUAs in the Sydney region.

However SUAs they still include large parks, reserves, industrial areas, airports, and large-area partially-populated SA2s on the urban fringe. Urban centres are slightly better (they are made of SA1s) but population data for these is only available in census years, the boundaries change with each census, the drawing of boundaries hasn’t been consistent over time, they include non-residential land, and they split off most satellite urban areas that are arguably still part of cities, even if not part of the main contiguous urban area.

Enter population-weighted density (PWD) which I’ve looked at previously (see Comparing the densities of Australian, European, Canadian, and New Zealand cities). Population-weighted density takes a weighted average of the density of all parcels of land that make up a city, with each parcel weighted by its population. One way to think about it is the average density of the population, rather than the average density of the land.

So parcels of land with no population don’t count at all, and large rural parcels of land that might be inside the “metropolitan area” count very little in the weighted average because of their relatively small population.

This means population-weighted density goes a long way to overcoming having to worry about the boundaries of the “urban area” of a city. Indeed, previously I have found that removing low density parcels of land had very little impact on calculations of PWD for Australian cities (see: Comparing the residential densities of Australian cities (2011)). More on this towards the end of this post.

Calculations of population-weighted density can also answer the question about whether the “average density” of a city has been increasing or decreasing.

But… measurement geography matters

One of the pitfalls of measuring population weighted density is that it very much depends on the statistical geography you are using.

If you use larger geographic zones you’ll get a lower value as most zones will include both populated and unpopulated areas.

If you use very small statistical geography (eg mesh blocks) you’ll end up with a lot fewer zones that are partially populated – most will be well populated or completely unpopulated, and that means your populated weighted density value will be much higher, and your measure is more looking at the density of housing areas.

To illustrate this, here’s an animated map of the Australian Capital Territory’s 2016 population density at all of the census geographies from mesh block (MB) to SA3:

Only at the mesh block and SA1 geographies can you clearly see that several newer outer suburbs of Canberra have much higher residential densities. The density calculation otherwise gets washed out quickly with lower resolution statistical geography, to the point where SA3 geography is pretty much useless as so much non-urban land is included (also, there are only 7 SA3s in total). I’ll come back to this issue at the end of the post.

Even if you have a preferred statistical geography for calculations, making international comparisons is very difficult because few countries will following the same guidelines for creating statistical geography. Near enough is not good enough. Worse still, statistical geography guidelines do not always result in consistently sized areas within a country (more on that later).

We need an unbiased universal statistical geography

Thankfully Europe and Australia have adopted a square kilometre grid geography for population estimates, which makes international PWD comparisons readily possible. Indeed I did one a few years ago looking at ~2011 data (see Comparing the densities of Australian, European, Canadian, and New Zealand cities).

This ABS is now providing population estimates on a square km grid for every year from 2006.

Here is what Melbourne’s estimated population density looks like on a km square grid, animated from 2006 to 2019:

The changes over time are relatively subtle, but if you watch the animation several times you’ll see growth – including relatively high density areas emerging on the urban fringe.

It’s a bit chunky, and it’s a bit of a matter of luck as to whether dense urban pockets fall entirely within a single grid square or on a boundary, but there is no intrinsic bias.

There’s also an issue that many grid squares will contain a mix of populated and non-populated land, particularly on the urban fringe (and a similar issue on coastlines). In a large city these will be in the minority, but in smaller cities these squares could make up a larger share of the total, so I think we need to be careful about this measure in smaller cities. I’m going to arbitrarily draw the line at 200,000 residents.

How are Australian cities trending for density using square km grid data? (2006 to 2019)

So now that we have an unbiased geography, we can measure PWD for cities over time.

The following chart is based on 2016 Significant Urban Area boundaries (slightly smaller than Greater Capital City Statistical Areas but also they go across state borders as appropriate for Canberra – Queanbeyan and Gold Coast – Tweed).

Technical notes: You cannot perfectly map km squares to Significant Urban Areas. I’ve included all kilometre grid squares which have a centroid within the 2016 Significant Urban Area boundaries (with a 0.01 degree tolerance added – which is roughly 1 km). Hobart appears only from 2018 because that’s when it crossed the 200,000 population threshold.

The above trend chart was a little congested for the smaller cities, so here is a zoomed in version without Sydney and Melbourne:

You can see most cities getting denser at various speeds, although Perth, Geelong, and Newcastle have each flat-lined for a few years.

Perth’s population growth slowed at the end of the mining boom around 2013, and infill development all but dried up, so the overall PWD increased only 0.2 persons/ha between 2013 and 2018.

Canberra has seen a surge in recent years, probably due to high density greenfield developments we saw above.

How is the mix of density changing? (2006 to 2019)

Here’s a look at the changing proportion of the population living at different densities for 2006-2019 for the five largest Australian cities, using square km grid geography:

It looks very much like the Melbourne breakdown bleeds into the Sydney breakdown. This roughly implies that Melbourne’s density distribution is on trend to look like Sydney’s 2006 distribution in around 2022 (accounting the for white space). That is, Melbourne’s density distribution is around 16 years behind Sydney’s on recent trends. Similarly, Brisbane looks a bit more than 15 years behind Melbourne on higher densities.

In Perth up until 2013 there was a big jump in the proportion of the population living at 35 persons / ha or higher, but then things peaked and the population living at higher densities declined, particularly as there was a net migration away from the inner and middle suburbs towards the outer suburbs.

Here’s the same for the next seven largest cities:

Of the smaller cities, densities higher than 35 persons/ha are only seen in Gold Coast, Newcastle, Wollongong and more recently in Canberra.

The large number of people living at low densities in the Sunshine Coast might reflect suburbs that contain a large number of holiday homes with no usual residents (I suspect the dwelling density would be relatively higher). This might also apply in the Gold Coast, Central Coast, Geelong (which actually includes much of the Bellarine Peninsula) and possibly other cities.

Also, the Central Coast and Sunshine Coast urban patterns are highly fragmented which means lots of part-urban grid squares, which will dilute the PWD of these “cities”.

Because I am sure many of you will be interested, here are animated maps for these cities:

Sydney

Brisbane

Perth

Adelaide

Canberra – Queanbeyan

Sunshine Coast

Gold Coast

Newcastle – Maitland and Central Coast

Wollongong

Geelong

What are the density trends further back in time using census data?

The census provides the highest resolution and therefore the closest measure of “residential” population weighted density. However, we’ve got some challenges around the statistical geography.

Prior to 2006, the smallest geography at which census population data is available is the collector district (CD), which average around 500 to 600 residents. A smaller geography – the mesh block (MB) – was introduced in 2006 and averages around 90 residents.

Unfortunately, both collector districts and mesh blocks are not consistently sized across cities or years (note: y axis on these charts does not start at zero):

Technical note: I have mapped all CDs and MBs to Greater Capital City Statistical Area (GCCSA) boundaries, based on the entire CD fitting within the GCCSA boundaries (which have not yet changed since they were created in 2011).

There is certainly some variance between cities and years, so we need to proceed with caution, particularly in comparing cities. Hobart and Adelaide have the smallest CDs and MBs on average, while Sydney generally has larger CDs and MBs. This might be a product of whether mesh blocks were made too small or large, or it might be that density is just higher and it is more difficult to draw smaller mesh blocks. The difference in median population may or may not be explained by the creation of part-residential mesh blocks.

Also, we don’t have a long time series of data at the one geography level. Rather than provide two charts which break at 2006, I’ve calculated PWD for both CD and mesh block geography for 2006, and then estimated equivalent mesh block level PWD for earlier years by scaling them up by the ratio of 2006 PWD calculations.

In Adelaide, the mesh block PWD for 2006 is 50% larger than the CD PWD, while in the Australian Capital Territory it is 110% larger, with other cities falling somewhere in between.

Would these ratios hold for previous years? We cannot be sure. Collector Districts were effectively replaced with SA1s (with an average population of 500, only slightly smaller) and we can calculate the ratio of mesh block PWD to SA1 PWD for 2011 and 2016. For most cities the ratio in 2016 is within 10% of the ratio in 2011. So hopefully the ratio of CD PWD to mesh block PWD would remain fairly similar over time.

So, with those assumptions, here’s what the time series then looks like for PWD at mesh block geography:

As per the square km grid values, Sydney and Melbourne are well clear of the pack.

Most cities had a PWD low point in 1996. That is, until around 1996 they were sprawling at low densities more than they were densifying in established areas, and then the balance changed post 1996. Exceptions are Darwin which bottomed out in 2001 and Hobart which bottomed in 2006.

The data shows rapid densification in Melbourne and Sydney between 2011 and 2016, much more so than the square km grid data time series. But we also saw a significant jump in the median size of mesh blocks in those cities between 2011 and 2016 (and if you dig deeper, the distribution of mesh block population sizes also shifted significantly), so the inflection in the curves in 2011 are at least partly a product of how new mesh block boundaries were cut in 2016, compared to 2011. Clearly statistical geography isn’t always good for time series and inter-city analysis!

How has the distribution of densities changed in cities since 1986?

The next chart shows the distribution of population density for Greater Capital City Statistical Areas based on collector districts for the 1986 to 2006 censuses:

You can more clearly see the decline in population density in most cities from 1986 to 1996, and it wasn’t just because most of the population growth was a lower densities. In Hobart, Canberra, Adelaide, Brisbane and Melbourne, the total number of people living at densities of 30 or higher actually reduced between 1986 and 1996.

Here is the equivalent chart for change in density distribution by mesh block geography for the capital cities for 2006, 2011, and 2016:

I’ve used the same colour scale, but note that the much smaller geography size means you see a lot more of the population at the higher density ranges.

The patterns are very similar to the distribution for square km grid data. You can see the how Brisbane seems to bleed into Melbourne and then into Sydney, suggesting a roughly 15 year lag in density distributions. This chart also more clearly shows the recent rapid rise of high density living in the smaller cities of Canberra and Darwin.

The next chart shows the 2016 distribution of population by mesh block density using Statistical Urban Area 2016 boundaries, including the smaller cities:

Gold Coast and Wollongong stand out as smaller cities with a significant portion of their population at relatively high densities, but a fair way off Sydney and Melbourne.

(Sorry I don’t have a mesh block times series of density distribution for the smaller cities – it would take a lot of GIS processing to map 2006 and 2011 mesh blocks to 2016 SUAs, and the trends would probably be similar to the km grid results).

Can we measure density changes further back in history and for smaller cities?

Yes, but we need to use different statistical geography. Annual population estimates are available at SA2 geography back to 1991, and at SA3 geography back to 1981.

However, there are again problems with consistency in statistical geography between cities and over time.

Previously on this blog I had assumed that guidelines for creation of statistical geography boundaries have been consistently applied by the ABS across Australia, resulting in reasonably consistent population sizes, and allowing comparisons of population-weighted density between cities using particular levels of statistical geography.

Unfortunately that wasn’t a good assumption.

Here are the median population sizes of all populated zones for the different statistical geographies in the 2016 census:

Note: I’ve used a log scale on the Y-axis.

While there isn’t a huge amount of variation between medians at mesh block and SA1 geographies, there are massive variations at SA2 and larger geographies.

SA2s are intended to have 3,000 to 25,000 residents (a fairly large range), with an average population of 10,000 (although often smaller in rural areas). You can see from the chart above that there are large variances between medians of the cities, with the median size in Canberra and Darwin below the bottom of the desired range.

I have asked the ABS about this issue. They say it is related to the size of gazetted localities, state government involvement, some dense functional areas with no obvious internal divisions (such as the Melbourne CBD), and the importance of capturing indigenous regions in some places (eg the Northern Territory). SA2 geography will be up for review when they update statistical geography for 2021.

While smaller SA2s mean you get higher resolution inter-censal statistics (which is nice), it also means you cannot compare raw population weighted density calculations between cities at SA2 geography.

However, all is not lost. We’ve got calculations of PWD on the unbiased square kilometre grid geography, and we can compare these with calculations on SA2 geography. It turns out they are very strongly linearly correlated (r-squared of over 0.99 for all cities except Geelong).

So it is possible to estimate square km grid PWD prior to 2006 using a simple linear regression on the calculations for 2006 to 2018.

But there is another complication – ABS changed the SA2 boundaries in 2016 (as is appropriate as cities grow and change). Data is available at the new 2016 boundaries back to 2001, but for 1991 to 2000 data is only available on the older 2011 boundaries. For most cities this only creates a small perturbation in PWD calculations around 2001 (as you’ll see on the next chart), but it’s larger for Geelong, Gold Coast – Tweed Heads and Newcastle Maitland so I’m not willing to provide pre-2001 estimates for those cities.

The bottom of this chart is quite congested so here’s an enlargement:

Even if the scaling isn’t perfect for all history, the chart still shows the shape of the curve of the values.

Consistent with the CD data, several cities appear to have bottomed out in the mid 1990s. On SA2 data, that includes Adelaide in 1995, Perth and Brisbane in 1994, Canberra in 1998 and Wollongong in 2006.

Can we go back further?

If we want to go back another ten years, we need to use SA3 geography, which also means we need to switch to Greater Capital City Statistical Areas as SA3s don’t map perfectly to Significant Urban Areas (which are constructed of SA2s). Because they are quite large, I’m only going to estimate PWD for larger cities which have reasonable numbers of SA3s that would likely have been fully populated in 1981.

I’ve applied the same linear regression approach to calculate estimated square kilometre grid population weighted density based on PWD calculated at SA3 geography (the correlations are strong with r-squared above 0.98 for all cities).

The following chart shows the best available estimates for PWD back to 1981, using SA3 data for 1991 to 2000, SA2 data for 2001 to 2005, and square km grid data from 2006 onwards:

Technical notes: SA3 boundaries have yet to change within capital cities, so there isn’t the issue we had with SA2s. The estimates based on SA2 and SA3 data don’t quite line up between 1990 and 1991 which demonstrates the limitations of this approach.

The four large cities shown appear to have been getting less dense in the 1980s (Melbourne quite dramatically). These trends could be related to changes in housing/planning policy over time but they might also be artefacts of using such a coarse statistical geography. It tends to support the theory that PWD bottomed out in the mid 1990s in Australia’s largest cities.

Could we do better than this for long term history? Well, you could probably do a reasonable job of apportioning census collector district data from 1986 to 2001 censuses onto the km grid, but that would be a lot of work! It also wouldn’t be perfectly consistent because ABS use dwelling address data to apportion SA1 population estimates into kilometre grid cells. Besides we have reasonable estimates using collector district geography back to 1986 anyway.

Melbourne’s population-weighted density over time

So many calculations of PWD are possible – but do they have similar trends?

I’ve taken a more detailed look at my home city Melbourne, using all available ABS population figures for the geographic units ranging from mesh blocks to SA3s inside “Greater Melbourne” and/or the Melbourne Significant Urban Area (based on the 2016 boundary), to produce the following chart:

Most of the datasets show an acceleration in PWD post 2011, except the SA3 calculations which are perhaps a little more washed out. The kink in the mesh block PWD is much starker than the other measures.

The Melbourne SUA includes only 62% of the land of the Greater Melbourne GCCSA, yet there isn’t much difference in the PWD calculated at SA2 geography – which is the great thing about population-weighted density.

All of the time series data suggests 1994 was the year in which Melbourne’s population weighted density bottomed out.

Appendix 1: How much do PWD calculations vary by statistical geography?

Census data allows us to calculate PWD at all levels of statistical geography to see if and how it distorts with larger statistical geography. I’ve also added km grid PWD calculations, and here are all the calculations for 2016:

Technical note: square km grid population data is estimated for 30 June 2016 while the census data is for 9 August 2016. Probably not a significant issue!

You can see cities rank differently when km grid results are compared to other statistical geography – reflecting the biases in population sizes at SA2 and larger geographies. Wollongong and Geelong also show a lot of variation in rank between geographies – probably owing to their small size.

The cities with small pockets of high density – in particular Gold Coast – drop rank with large geography as these small dense areas quickly get washed out.

I’ve taken the statistical geography all the way to Significant Urban Area – a single zone for each city which is the same as unweighted population density. These are absurdly low figures and in no way representative of urban density. They also suggest Canberra is more dense than Melbourne.

Appendix 2: Issues with over-sized SA1s

As I’ve mentioned recently, there’s an issue that the ABS did not create enough reasonably sized SA1s in some city’s urban growth areas in 2011 and 2016. Thankfully, it looks like they did however create a sensible number of mesh blocks in these areas, as the following map (created with ABS Maps) of the Altona Meadows / Point Cook east area of Melbourne shows:

In the north parts of this map you can see there are roughly 4-8 mesh blocks per SA1, but there is an oversized SA1 in the south of the map with around 50 mesh blocks. This will impact PWD calculated at SA1 geography, although these anomalies are relatively small when you are looking at a city as large as Melbourne.

1 Comment | Australian Cities, Brisbane, Melbourne, Sydney, Urban density | Permalink
Posted by chrisloader

« Previous Entries