WA Tomorrow Population Forecast - Methodology

WA Tomorrow 12 uses a Cohort Component Model (CCM) to produce the age by sex population forecasts for Western Australia and its sub-regions. The CCM is the most widely used approach by demographers for preparing population projections and is internationally recognised as best practice.  

The WA Tomorrow 12 forecasts are based on a series of assumptions regarding future trends in the demographic components of births, deaths and migration. The assumptions for the population forecasts are informed by a detailed analysis of recent trends in these demographic components. The assumptions are then applied to the base population, by age and sex, to derive the forecast population, age-sex structure and estimated future number of births, deaths and migrants. The CCM uses the following demographic equation:

New population =

Base population + births – deaths + in-migration – out-migration

The main CCM steps are to:

• start with the base population of an area, disaggregated by age and sex to form cohorts (age groups, usually 0 to 4, 5 to 9, etc)  
• apply the fertility rate to calculate the number of births
• apply the mortality rate to calculate the number of deaths
• apply the migration flows and rate to calculate the number of migrants (overseas and internal)
• calculate the new population for the following time period by ageing the population, adding the number of births to the youngest cohort, removing deaths from the relevant cohorts and adding and subtracting the number of migrants
• This process is repeated for each year in the forecast.

The CCM has the advantage of providing age-sex distributions in addition to total population numbers. 

The base population for the WA Tomorrow 12 State population forecast is the Australian Bureau of Statistics’ (ABS) Estimated Resident Population (ERP) at 30 June 2022, which takes into account the 2021 Census of Population and Housing.

The ERP is the official measure of Australia’ population. It includes all people who usually live in Australia, except foreign diplomats. The WA Tomorrow forecasts refer to the population as at 30 June for each stated year. This is consistent with the ABS’ ERP.

For more information regarding ERP, please visit the ABS website: National, state and territory population methodology

The WA Tomorrow 12 assumptions are based on an analysis of the following range of demographic datasets. They are based on the best available data at the time.

• Fertility rates are based on historical ABS births data, by year of occurrence (of the birth), from 2006 to 2019. There is usually an interval between the occurrence and registration of a birth (referred to as a registration ‘lag’). This lag in the registration of births can cause data reported by occurrence to be incomplete for the most recent data – 2020 and 2021. For this reason, analysis of past trends excludes data for 2020 and 2021
• Mortality rates are based on historical ABS deaths data, by year of occurrence (of the death), from 2006 to 2019
• Future assumed levels of Net Overseas Migration (NOM) and Net Interstate Migration (NIM) are informed by ABS data on Overseas and Interstate arrivals and departures, Commonwealth Budget NOM forecasts (National and WA), and leading economic indicators 

The WA Tomorrow 12 State population forecast applies assumptions about future fertility, mortality and migration. These assumptions are applied to the population for each year over the projection period, underpinning the future size, structure and growth of the population.

Assumptions are informed by historical trends. The COVID-19 pandemic added uncertainty to recent estimates, specifically overseas migration, which has made recent historical data less reliable for forecasting. The effects of the pandemic on recent demographic trends were considered and addressed during the assumption setting process. 

Fertility

Future trends in fertility are important in shaping future population size, structure and growth. The WA Tomorrow model applies age-specific fertility rates (ASFR) and sex ratio at birth for each year of the projection period to determine the number of babies born. The ASFR measures the number of births occurring to mothers of a given age per 1,000 women of the same age in the same year. The WA Tomorrow 12 State forecast proposes a relatively stable ASFR across the forecast period. The forecast ASFR assumes a continued trend towards older mothers to 2024, after which the age-specific fertility rates will remain constant.  

The Total Fertility Rate (TFR), which is the sum of ASFR per thousand women, is forecast to decrease from 1.65 births per woman in 2023 to 1.63 births in 2024, remaining constant thereafter. Based on ABS births by year of occurrence data, this assumption reflects the medium-term trend which shows a steadily declining TFR from 2.00 births per woman in 2008 to 1.69 in 2019. For all years, the sex ratio at birth is assumed to be 105.8 male births per 100 female births based on current data. 

Mortality

WA Tomorrow applies age-specific mortality rates  (ASMR) to determine the number of deaths that will occur for each year of the projection period.  WA Tomorrow 12 assumes mortality improvement across all ages, except for persons aged 100 years and over, for both males and females over the forecast period. The ASMR for males and females aged 100 years and over is held constant from 2023.

For each age, mortality rates for females are assumed to be lower than for males, except for females aged 99 years and over. This aligns with the higher life expectancy for females.

While ASMRs are assumed to continue to improve over the forecast period, the total number of deaths is expected to grow as the population ages. This is reflected by the Crude Death Rate (CDR). The CDR is derived from the total number of deaths occurring over a given period per 1,000 of the population. Derived from the WA Tomorrow forecast, the CDR is expected to increase from 5.67 deaths per 1,000 people in 2023 to 6.21 in 2036, due to the ageing demographic. 

Migration

The WA Tomorrow 12 State forecast applies assumptions on future levels of NOM and NIM. These assumptions have been developed in consultation with the WA Department of Treasury.  

NOM is the most volatile component of population change. This volatility has been heightened by the loss of overseas migration due to the COVID-19 pandemic, and its subsequent strong recovery.  The WA Tomorrow 12 State forecast has considered these recent trends, and long-term averages prior to  the pandemic.

NOM has recently been boosted by a sharp recovery in migrant arrivals, particularly students, and slow recovery in migrant departures. The WA Tomorrow State forecast assumes that NOM will peak in 2022-23 and return close to the long-term historic average from 2024-25 as migration flows normalise. This will include migrant departures increasing as recently arrived students complete their studies. Any potential future impacts from the Commonwealth Government’s Migration Strategy released in December 2023 have not been incorporated.

NIM is expected to remain elevated to 2023-24, before gradually returning to long term average by 2026-27, and remaining stable thereafter. 

Table 1: Summary of assumptions used in WA Tomorrow 12 State Forecast.

The WA Tomorrow 12 Sub-state population forecasts, by age and sex, have been produced for two geographic scales - the Statistical Area Level 2 (SA2s), derived from the ABS' Australian Statistical Geography Standard (ASGS) (July 2021), and Local Government Areas (LGA). The sum of each respective small area forecast equals to the WA Tomorrow 12 State Forecast (subject to rounding)

For SA2s, WA Tomorrow 12 uses a hybrid geography that reflects as closely as possible the current ASGS 2021. The SA2s that contain little, or no population, have been amalgamated with adjacent SA2s, leaving only 239 Sub-state forecast SA2 areas from the standard 265 SA2s in WA as identified by ABS. This geography is referred to as the WAT12 SA2s.  

Full information relating to the WAT12 SA2s can be found in the Geography tab in the SA2 forecast data Excel file, including the full list of SA2 amalgamations.  

LGA populations have been derived via a geographical and population concordance of SA2 forecasts. The LGA forecasts have not been directly forecast using the CCM. The SA2 forecasts by age-sex are disaggregated to SA1 by age-sex, based on ratios derived from ABS data. The SA1 by age-sex are then aggregated to derive LGA forecasts by age-sex. 

The WA Tomorrow 12 Sub-state population forecasts have also been produced using the CCM methodology for WAT12 SA2s. The Sub-state model uses a region-specific set of fertility, mortality and migration assumptions for each WAT12 SA2 based on an analysis of past trends. The assumptions are applied to the base population, by age and sex, to derive the forecast population. The Sub-state WAT12 SA2 and LGA forecasts was distinct from the WA Tomorrow 12 State forecast CCM, though the Sub-state forecasts were constrained to the State forecast by age, sex and components.  

The CCM forecasts for WAT12 SA2s were projected at five-year intervals for 2026, 2031 and 2036. Due to the limited historic data available at the SA2 level, the underlying input data for Sub-state geography was only available at 5-year age cohorts. Therefore, the forecasts are calculated at five-year intervals, as each 5-year age cohort completely transitions to the next. For example, the 5 – 9-year-old age cohort in 2026 will transition to the 10 - 14-year-old age cohort in 2031. The WA Tomorrow Sub-state population forecasts for intercensal years were derived by interpolating the forecast population between future Census years. 

The base population for the WA Tomorrow 12 Sub-state population forecast is the ABS’ final rebased SA2 ERP as at 30 June 2021.

This differs to the WA Tomorrow 12 State forecast, which used the more recent 30 June 2022 ERP as the base population. This difference in base populations was due to challenges around data availability at the SA2 and LGA level which is disaggregated by age and sex. The 2021 base for the Sub-state forecasts means that the CCM outputs are projected at five-year time intervals which align with future ABS Census years.  

The WA Tomorrow 12 Sub-state assumptions are based on an analysis of a range of demographic datasets. They are based on the best available data at the time.  

1. Age-specific fertility rates are based on historical ABS births data at SA2 level, by year of occurrence (of the birth), from 2006 to 2021.  
2. Period cohort mortality rates are based on historical ABS deaths data at the state level, by year of occurrence (of the death), 2017-2021 (historical period); and WA Tomorrow 12 State forecast deaths estimates from 2021 to 2036.  
3. Sub-state migration rate assumptions are informed by ABS data on Internal and Overseas arrivals and departures for SA2s - Regional Internal Migration Estimates (RIME) and Regional Overseas Migration Estimates (ROME), 2017 to 2022. Sub-state migration forecasts are constrained by the WA Tomorrow 12 State forecast levels of Net Overseas Migration (NOM) and Net Interstate Migration (NIM). 

The WA Tomorrow 12 Sub-state population forecasts apply region-specific assumptions about future fertility, mortality and migration. These assumptions are applied to each five-year age-sex cohort over the projection period, underpinning the future size, structure and growth of the forecast population for each Sub-state area.

Fertility

The WA Tomorrow 12 Sub-state model applies age-specific fertility rates (ASFR) for each region for relevant female five-year age cohorts and the sex ratio at birth, for each five-year projection period to estimate the number of male and female babies born.  

For all future periods, sex ratio at birth is assumed to be 105.8 male births per 100 female births based on current data. This assumption is consistent with the WA Tomorrow 12 State forecast.  

The sum of WAT12 SA2 births is constrained to the State level forecast.  

Mortality

The WA Tomorrow 12 Sub-state forecast matches the equivalent State forecast assumption for period cohort mortality rates (PCMR) by sex across all WAT12 SA2s. The PCMR is used to determine the number of deaths that will occur for each five-year interval of the projection period (2021-26, 2026-31, and 2031-36). The WA Tomorrow 12 Sub-state forecast assumes mortality improvement across all ages for both males and females over the forecast period, except for persons aged 100 years and over where no mortality improvement for 100+ years is assumed.  

While overall mortality is assumed to continue to improve over the forecast period, the total number of deaths is expected to grow as the population ages.

The sum of WAT12 SA2 deaths is constrained to the State level deaths.  

Migration

The WA Tomorrow Sub-state forecast applies internal and overseas migration rates and flows by age and sex for each WAT12 SA2. Internal migration is the sum of interstate and intrastate migration and represents the net movement of people between each small area and the remainder of Australia. Intrastate migration is much more common than interstate migration.

The Sub-state migration rate assumptions are informed by historical SA2 level data on internal and overseas arrivals and departures. This data is highly variable, with a short time series that has been highly impacted by unusual migratory flows during the COVID-19 pandemic. This data has also not been revised as part of the ABS’ ERP rebasing process following the 2021 Census of Population and Housing.

To overcome these limitations, the WA Tomorrow 12 Sub-state approach uses a synthetic migration approach for forecasting internal and overseas migration flows. This method combines the strengths of directional migration and local area net migration patterns, whilst minimising input data requirements. It also enables the unique age-sex profiles of each local area’s migration pattern to be maintained.  

The sum of WA Tomorrow 12 Sub-state internal and overseas migration flows were constrained by the WA Tomorrow 12 State forecast levels of Net Overseas Migration (NOM) and Net Interstate Migration (NIM).  

The WA Tomorrow Sub-state forecast method applies an independent total population forecast for each WAT12 SA2 based on historic ERP trends. This acts as a constraint to adjust the migration component flows in the synthetic migration CCM (with no adjustment made to fertility or mortality rates) so the CCM age and sex forecasts sum to the independent total population forecast.  

This process also applies an estimate of dwelling and population capacity to areas across Greater Perth. Future population growth was constrained in some areas where limited further developable land is available to support additional dwelling construction. The demographic forecasts were also considered and brought into alignment with the strategic outlook contained in Perth and Peel @ 3.5 million, and emerging growth areas.  

All forecasts comprise uncertainty, across the base input data, assumptions and model. Statistical techniques are used to overcome some uncertainty in the central forecast and are also used to generate forecast bands to reflect the inherent uncertainty.

WA Tomorrow 12 has applied different techniques to estimate higher and lower forecasts (Bands) for the State and Sub-state forecasts.  

The WA Tomorrow 12 State forecast uses probabilistic methodology to produce a forecast range. This range is presented as five forecast bands indicating five probable futures. ‘Band A’ and ‘Band B’ represent the lower forecasts, ‘Band C’ is the median forecast and ‘Band D’ and ‘Band E’ represent the higher forecasts.

To derive the WA Tomorrow 12 State forecast bands, a deterministic forecast model based on a single set of assumptions (rather than a set of different scenarios) is first produced to form ‘Band C’. The forecast range is then generated from 100,000 possible variations of this model, which are equally divided into the five bands or population forecasts. Each of the bands represent one-fifth of the 100,000 variations which denote different probability levels.

‘Band C’ represents equal 50% probabilities that the actual population will be less than, or higher than, this forecast. ‘Band B’ represents a 30% probability that the population will be less than or equal to ‘Band B’, and a 70% probability that it will be higher. Conversely, ‘Band D’ represents a 30% chance that the population will be higher or equal to ‘Band D’, and a 70% chance of it being lower. ‘Band A’ represents a 10% probability that the population will be less than or equal to Band A, and a 90% probability that it will be higher. Conversely, ‘Band E’ represents a 10% chance that the population will be higher or equal to Band E, and a 90% chance of it being lower.

It is important to understand the differences between the Bands and to choose the Band most suitable for the user’s needs. For the user who requires the best estimate of future population size based on past trends, ‘Band C’ should be used.  The other Bands should be used if they are required for other decision-making purposes. 

For the WA Tomorrow 12 Sub-state forecasts, the uncertainty of each SA2 and LGA population forecast is estimated with an ‘Upper Band’ and ‘Lower Band’. Based on past forecast errors of Sub-state populations across Australia [1], the upper and lower bands for each area represent an 80% certainty that the actual population will lie between these forecast limits.  

The 'Upper Band' represents a 10% probability that the population will be more than or equal to the Upper Band, and an 90% probability that it will be lower. Conversely, the ‘Lower Band’ represents a 10% chance that the population will be lower or equal to the Lower Band, and an 90% chance of it being higher. The Central Forecast, like the WA Tomorrow State forecast ‘Band C’, represents the best estimate of the future population based on past trends.  

It is important to note that the smaller the population of an area, the wider the bands. This model will calculate a larger range to an area with a smaller baseline population, representing a greater variance in the population forecast itself. The sum of the WAT12 SA2 upper and lower bands is wider than that of the LGAs due to the higher number of small area WAT12 SA2 populations and therefore greater variance in average percentage error. It is also wider than Bands A and E of the WA Tomorrow 12 State forecast.  

The Sub-state forecast upper and lower bands give the same 80% confidence range as the Band A and Band E from the State forecast.