Every resource, infrastructure investment, policy and planning decision at all levels of government is based on numbers like population statistics.
The numbers that drive government planning in NSW are built on a foundation that turns out to be more complicated than most people realise, and the further you dig, the more questions emerge about how reliable that foundation actually is.
In this New England Times investigation by RK Crosby, we look at how bad numbers, assumptions and guesses, and misrepresentations of old documents, have led to plans at all levels of government not being worth the paper they are printed on.
In the first story in this series, we started by looking at the NSW Government’s draft transport plan for the New England North West and it’s problematic assumptions.
There were two key numbers in that document that started our questions: the first was a projected a regional population of 198,000 by 2041, starting from a 2021 base of 187,000. Aside from being absurdly round figures, they just didn’t match up. The ABS census figure for the same region in 2021 was 185,560, and the adjusted ABS estimated resident population (ERP) was 186 972. By 2024, the ERP had already reached 190,289, or a growth rate more than double the NSW Government’s projections.
The jobs figures were more striking still. The transport plan put employment in the region at 87,000 in 2021, rising to 97,000 by 2041. The ABS jobs figure for 2022 was 157,467. A difference of some 70,000 is difficult to explain.
When we asked Transport for NSW to explain the difference, the answer pointed back to NSW’s Common Planning Assumptions.
So where do the Common Planning Assumptions come from, and how good are they?
The planning numbers explained
The Department of Planning, Housing and Infrastructure says the Common Planning Assumptions are “a whole-of-government planning tool and one of the ways used to support infrastructure and service planning across NSW,” and that they “are not intended to be directly compared with other statistics, such as ABS data.”
“Population and employment projections are developed using a range of data sources, including the Census, official population statistics, and local intelligence,” a department spokesperson said.
“They may differ from other datasets due to differences in definitions, geography and methodology.”
The department’s population projections are, it says, built on a substantial base. They use the ABS Estimated Resident Population as the starting point, which the department notes “differs from Census counts as it adjusts for undercounting” and is “the official measure of population.”
From there, they incorporate assumptions about the three drivers of population change: births, deaths and migration. They draw on ABS Census data, data from NSW and federal government agencies, consultation with expert demographers and NSW and local government stakeholders, and advice from local councils.
The department says the projections are developed using “internationally recognised, best-practice projection methods, with a combination of top-down and bottom-up modelling approaches,” peer-reviewed by demographic experts and other NSW government agencies, updated regularly to reflect new ABS data releases, economic conditions and land use planning changes, and validated against official population statistics after each update.
On the New England North West region specifically, the department notes that “overseas migration is the biggest driver of population growth” — a factor that would have been particularly volatile in the years around the COVID-19 pandemic, and one that is notoriously difficult to project.
The current projections, released in 2024, do account for disruptions to overseas migration flows from international travel restrictions, larger net losses of people from NSW to other states and territories during the pandemic, and changes to mobility patterns within NSW. But plans before then, such as the NSW Government’s New England North West Regional Plan 2041, are based on the assumption that the pandemic related city to country shift is temporary.
In simple terms, that means the NSW state government is building its investment and infrastructure decisions for the region on the expectation that the people from the city who came here during and since the pandemic will leave.
It is, in other words, planning for the regions to fail.
Those planning assumptions from the Regional Plan 2041 are then baked into all subsequent plans, like the transport plan. But the same issues with population, and the same error in “jobs” numbers was made in that document too. So where did that start?
Small language shift created big numbers problem
Employment projections follow a separate process from the population projections. The NSW Government engaged Victoria University to prepare them, providing “regional industry and occupation detail” and adding spatial context to the broader macroeconomic and demographic outlook set out in the NSW Intergenerational Report, NSW population projections and the NSW Budget. But those employment and jobs numbers also start from a Census base, so how did they get so far apart?
When we asked the ABS directly to explain the divergence between its jobs figures and NSW’s, the answer was succinct: “This would be a difference in methodology, data sources, and modelling.”
The low growth population assumptions appears to be in that analysis from Victoria University, but they in turn appear to be using a NSW Government source that we have not been able to identify.
But the cause of the NSW Government planning jobs numbers being wildly discordant from ABS figures is not so hard to track down, and has nothing to do with methodology or modelling.
The ABS measures jobs in the country from three different sources: businesses data by industry is from the Labour Account publication and employment data is from the 2021 Census or Jobs in Australia.
Another number available in the Census data set is employed persons. The number of people who work.
The number being referred to as ‘jobs’ in the NSW Government’s planing documents is employed persons, not jobs.
The two are not the same thing. Employed persons counts people, one per person regardless of how many jobs they hold. Jobs counts positions, so a person working two part-time roles counts twice. In a region where seasonal agricultural work, mining rosters and contract labour are common, the difference matters.
The root cause of the issue again seems to be within the Transport for NSW. In particular, the Travel Zone Projections 2019 document, which is referred to as the source in many other planning documents, repeatedly refers to jobs in the text while describing or using employed persons statistics.
It also, unhelpfully, refers to actual jobs as jobs, particularly the number of jobs expected to be created by certain projects. Meaning that it is impossible to determine which jobs figures are jobs, and which are really employed people, and whether the two very different statistics have been inadvertently combined at any point.
This error by Transport appears to have, to some degree, infected numerous other planning documents in NSW since being issued in 2019.
Real numbers issues: how the census is imperfect by design
The Australian Census is taken every five years. It is the most comprehensive count of people, households, employment, language, religion, health and living circumstances in the country. It underpins funding decisions at every level of government, from hospital allocations to school planning to road investment.
It is also, unavoidably, imperfect.
The ABS openly acknowledges this. The official population figure it produces is not the census count itself, but a derived figure called the Estimated Resident Population, or ERP. The ERP adjusts for underenumeration, the technical term for the gap between the number of people who should have been counted and those who actually were.
Underenumeration occurs when people are missed entirely, when they do not complete a form, or when they are counted in the wrong place. The ABS runs a Post-Enumeration Survey after every census to estimate how many people were missed and adjust the figures accordingly. In theory, this should correct for most of the problem.
In practice, some groups are consistently harder to count than others. Like people in regional areas.
Narrabri Mayor Darrell Tiemens raised the issue publicly in March, ahead of the upcoming census in August, saying anecdotal evidence suggests Narrabri Shire was undercounted in 2021 and calling for the ABS to increase its efforts in regional areas.
“The census is one of the most important datasets used by all levels of government and many service providers to determine funding allocations, infrastructure investment and service delivery,” Mayor Tiemens said.
“When regional communities are undercounted, it means they are effectively ripped off when it comes to hospitals, roads, housing and other essential services.”
He pointed to several groups in the shire that are difficult to capture: itinerant agricultural workers, international temporary workers, backpackers, contractors working across large properties, fly-in-fly-out and drive-in-drive-out mining workers on roster arrangements, interstate truck drivers, people in temporary or informal accommodation, and people without a fixed address.
“Our Shire covers an enormous geographic area, and our population is far more dynamic than many people realise,” Mayor Tiemens said. “
As Australia’s second largest agricultural economy, we have a highly mobile workforce that must be accurately captured.”
The ABS says its census is based on usual residence, defined as the address at which a person lives or intends to live for six months or more. By design, this approach does not capture the temporary or service population that Mayor Tiemens is describing.
“The usual residence concept is not intended to measure temporary or service populations, or to tell us how many people there are in areas at different points in time,” an ABS spokesperson said.
That is not an admission of failure but it is a statement of limitation. For communities whose economic activity depends heavily on mobile workforces, the official population count may consistently understate both the demand for services and the scale of economic activity.
Who else doesn’t count
The regional undercount problem is part of a broader pattern of groups that the census struggles to accurately capture.
The census has historically had difficulties counting people who do not speak fluent English, and the form itself is only offered in English. Refugee and recently arrived migrant populations, such as the Ezidi and Congalese people in Armidale, are some of the fastest-growing communities in parts of regional NSW, are among those most likely to be undercounted.
Religion figures have long been considered unreliable. Research from Monash University’s Australian Centre for Jewish Civilisation has documented the tendency of Jewish Australians to under-identify in the census, a reluctance with roots in the use of census data by Nazi Germany to identify and round up Jewish communities during the Holocaust. A multi-faith group is this year running a project to encourage people of faith to identify their religion in the census, particularly given rising tensions and reports of religious bigotry, particularly towards Jews and Muslims.
Attempts to improve the 2026 census by adding questions on LGBTQI identity and switching from ancestry to ethnicity as a category were tested and then dropped. Advocacy organisations have argued that without better data on these populations, services and funding cannot be properly targeted to meet their needs.
Health data collected in the census carries its own problems. For example, questions about chronic illness rely on pick lists of specific conditions, which tend to reflect the most well most funded ailments rather than the full picture of community health.
Migraine Australia have campaigned hard on their ongoing exclusion for years, releasing a statement again last year condemning the ABS for excluding migraine from the 2026 census long-term health conditions question. They say the decision “continues a pattern of exclusion that has left migraine unrecognised in national health statistics for years.”
The ABS stated in its Outcomes of the 2026 Census Topic Review that migraine will not be included because “migraine” may not be consistently understood or answered correctly by respondents.
This design decision to exclude one of the most common chronic illnesses in the world from census figures, like the misattribution of employed persons as jobs in the NSW planning figures, does not stay confined to census data. It infects every other data set related to chronic illness. This includes reports from the Australian Institute of Health and Welfare, in every state and territory and federal health department reports and assessments, independent academic research, and it warps public understanding, media coverage, and funding and policy decisions related to chronic illness for years.
“If migraine isn’t counted, it isn’t funded. It isn’t researched. And people living with migraine are left to suffer in silence,” Migraine Australia said.
Fixing the front door
The federal government is trying to improve census participation rates through a new myGov subscription option, which allows people to register for census updates and receive a direct link to the online form when census night approaches.
“The census is Australia’s national stocktake: a count that helps us plan for the future,” said assistant minister for productivity, competition, charities and Treasury Andrew Leigh. “It helps governments, businesses and communities work out where schools, roads, hospitals and other services are needed.”
Mr Leigh said the myGov option was an additional access point, not a replacement. “Most people will still complete it online through the Census website, and paper forms will still be available for those who prefer their Census with a pen.”
The myGov option is available now at my.gov.au/census. People who subscribe will receive a link to the census form in their secure myGov inbox ahead of census night.
The NSW Department of Planning, Housing and Infrastructure also says they are planning to improve their work.
“The Department is reviewing all regional plans across NSW as part of a broader update to strategic planning, including the New England North West Regional Plan 2041.”
“The review will lead to the development of new regional plans that will consider updated planning regions, housing needs and opportunities for employment growth. The plans will aim to improve alignment between infrastructure delivery and regional development, and ensure key transport, freight and logistics corridors are appropriately identified and protected.”
This review will also reset the plans for a future without Inland Rail, as all NSW Government plans have had that assumption baked in as a key driver of regional growth and improvements.
“As part of this process, the Department will consider recent State and Commonwealth policy and investment decisions,” the spokesperson said.
The Minister for Transport has been asked for a comment regarding the apparent error around ‘jobs’ in Transport for NSW key planning documents.
Read more from the Bad Numbers investigation
Got something you want to say about this story? Have your say on our opinion and comment hub, New England Times Engage