Census crystal ball gazing: the results
Following the first release of data from the 2021 Census, demographer and Census expert Glenn Capuano revisits the predictions he made about what this Census would show.
Every Census, around Census day, I look into my crystal ball to come up with a few key trends that I think the Census results will show when they are finally released. In 2021, I made 12 specific predictions about the Census data. Now the data are (partially) out, it’s time to see how accurate (or inaccurate) these predictions were.
1. Slower population growth: Australia’s population growth for June 30th, 2021 revised to 25,675,000
The final 2021 population was revised down by just over 50,000 people, to 25,688,079 people – just 13,000 out from my estimate. That’s less than 0.05% out!
2. Census count much closer to Estimated Resident Population: count of 25,400,000 being a gap of only 275,000 from the population estimate
The actual Census count of usual residents in Australia was 25,422,788 – less than 0.1% different from my prediction. The ERP shown the first prediction is only 265,291 people higher than the Census count, almost exactly 1% different, and within 10,000 people of the gap I predicted. This difference is down from almost 800,000 in 2016. The improvement is partly is due to better coverage of the population (the Census website didn’t go down on Census night for starters!) but more so due to far less people being overseas. (Our borders were closed and you had to get a special exemption to leave the country, remember that?) People overseas are counted in the population but not in the Census count, so when there are less people overseas, the gap between Census and ERP is less.
3. 80% of Australians will fill in the Census online
Well OK, this was only 78.9%, but definitely closer to my 80% number than the 75% which was the ABS’s target range for online form response. So I’m counting this one as a hit as well, despite being 1.1% out. Higher online response has also led to lower “Not Stated” rates for many characteristics, as the online form doesn’t allow respondents to skip questions and guides you through the form to only those questions which are relevant to you (based on age, employment etc.)
4. More people counted at home: 97.2% of the population at home on Census night due to lockdowns
Result: Partial hit
This was a very specific prediction with a decimal point, so I’m a little off with that. While the proportion of people counted at home did increase as predicted, it didn’t increase as much as I thought; it went from 95.1% to 96.1%, missing my 97.2% by more than 1%. Maybe I was a bit over-enthusiastic about how many people would be following lockdown orders!
It’s been widely reported (from the ABS’s own media releases) that 2 million more people were counted at home on Census night this time. That is true – but the Census population count overall went up by more than 2 million! A good example of how statistics can be used to show things that, while technically true, aren’t as amazing as they may first seem.
5. Less overseas visitors: no more than 25,000 people from overseas in Australia on Census night
Result: Partial hit
Again, I got the direction but not quite the magnitude right . There were certainly far fewer overseas visitors in 2021, due to closed borders and strict quarantine requirements, but somehow more than I expected were still visiting Australia. The count of overseas visitors declined from 315,000 in 2016 to 62,000 in 2021 – still more than twice as many as I predicted.
6. Changes to country of birth: UK remains the largest but China and India become 2nd and 3rd, jumping ahead of New Zealand
There are still well over 1 million UK-born in Australia (though the ABS persists in reporting this by the individual UK countries, England, Scotland etc.). Both China and India have leapt ahead of New Zealand. The only thing I got wrong (and didn’t fully specify) was the order of those two. Indian migration has far outpaced China in the last 5 years, now standing at more than 673,000 India-born population, in 2nd place behind the UK only. China is more than 100k behind with 549,000.
7. Changing migration patterns: only 60% of the 5-year population growth will be within Greater Capital City areas, down from 78% in 2016
Again, I was a little over-exuberant here. Capital City growth did decline as a percentage of total. Of Australia’s 2.415 million growth between the Censuses, 76.2% was in Greater Capital City areas and 23.8% was in regional Australia. This is only slightly less than last Census. Perhaps my perspective was coloured by living in Melbourne – the two largest cities, Melbourne and Sydney, have a lot of net movement out to regional areas but this is less evident in other parts of Australia, which were less affected by lockdowns. (And remember that Census covers 5 years and lockdowns only affected the last 18 months.) I will revisit when the ERP data comes out, as this may change things a little.
8. Further rise in ‘No Religion’: those having no religious affiliation to rise to 35% of the total population
Result: Partial hit
Again, the direction was correct but not the magnitude. I wasn’t prepared for the very large shift from 30% to almost 39% of the population having no religion in 2021. This rise is larger in percentage terms than the one from 2011–2016, and in that period the question wording changed and “No Religion” was moved to the top of the response list on the Census form. This is a very big shift in Australian society away from religious affiliation and will be the subject of another blog. In just a decade the proportion of Australia’s population professing no religion has almost doubled.
9. Increase in work from home: 25% of the population of Australia will work from home on Census day
Result: Unknown until 2nd release
Census was conducted in the middle of lockdowns in Melbourne and Sydney, so this is likely, but unfortunately we don’t have that data yet. The second release of Census data is due in October this year, and will include everything to do with employment, industry, occupation, journey to work and method of travel to work. So we’ll have to wait a few months for this!
10. Further increase in renting: renters to rise to 30.9% of all dwellings
Well, I bought the hype here. Tenure types have barely moved in the 2021 Census, with renting as a percentage going up only from 29.4% to 29.5% of households in 5 years. The other two categories – full ownership and mortgage – also increased in percentage terms to 29.9% and 33.2% respectively, both within 1% of their 2016 percentages. The only reason it’s possible for all the categories to increase in percentage is a big decline in the “Not Stated” category, from 7.7% to 5.4%. (This is due to better quality Census results.) In fact, if you exclude this, renting as a percentage has gone down. That’s right – more people have entered the housing market than before. Probably due to low interest rates. The second part of my prediction was a big increase in the percentage of 25–34 year olds who are renters. This will have to wait until ABS TableBuilder is available, as we don’t have this data in any of the standard datasets at present.
11. Unoccupied dwellings: increase to 11.2% but a lot of variability with holiday areas down and inner city areas up
Again, despite a lot of misleading media talk about record vacant dwellings, unoccupied dwellings as a percentage actually went down to 9.6% of total dwellings. (The absolute number was remarkably consistent at just over 1 million, despite another million dwellings being added to Australia’s dwelling stock.) I did say this one would be hard to predict, and it is highly variable from area to area. Typical beachside holiday areas in southern Australia like Eurobodalla Shire in NSW (29.6% down to 25.3%) and Bass Coast Vic (44.3% down to 37.3%) still had the bulk of unoccupied dwellings but generally fell as more people occupied those dwellings in a lockdown winter, while inner-city areas like the City of Melbourne had a huge increase in unoccupied (12.8% to 25.2%).
What this does not show is large number of investors deliberately keeping dwellings vacant for negative gearing benefits (this is not allowed under the current tax system). Remember also that unoccupied doesn’t necessarily mean vacant long-term. The occupants could just be away on Census night.
12. Long-term health conditions: 43% with a LTHC and arthritis the most common
I was quite a bit too high on the percentage here, with only 32% nationwide reporting a long-term health condition in the 2021 Census. This was a bit of an unknown and I based this number off the national health survey. It’s quite likely that there is more of a tendency to under-report this in a self-responded Census questionnaire compared to an interviewer based survey. But it’s still going to be a very useful dataset for comparisons at a local level. I wasn’t far off with arthritis as the most common LTHC: it was reported by 8.5% of people, and more common among the elderly, but the most commonly reported condition was actually mental health, at 8.8% of the population.
So overall, a mixed bag for the predictions this year! Of the 12 predictions, I had:
- 4 hits
- 3 partial hits
- 4 misses
- 1 unknown until 2nd release.
I hope you’ve enjoyed following these predictions, which are based on our understanding of the underlying trends in the population. These are national trends, of course, and the great thing about Census is it brings out all the local trends – where the real stories are. Every place is different, and we’re rolling out the 2021 Census data as quickly as possible to allow all our users Local Government and beyond to tell that story for the places within their local areas. We’ll also be out and about in the coming months presenting these results and running training and development in using the .id toolkit at your council. If you’d like to organise this for your area, or have any queries about the Census data, please don’t hesitate to get in touch: firstname.lastname@example.org.