Population forecasts in uncertain times

In the last couple of years I've lost count of the number of times that I've been told that population forecasts are useless because of the impacts brought about by COVID-19. Certainly I've written many blogs that describe the demographic impact of the pandemic, but does this render population forecasts useless? We do live in uncertain times, but I argue that they remain as relevant as ever. Read on to find out why.

Population forecasts, particularly for small geographic areas, are fraught with uncertainty at the best of times. There's a lot of misunderstanding about what they mean. They are not targets, or predictions - they are the modelled outcomes of a set of assumptions about the future direction of the components of demographic change. For example, how many births and deaths? How much migration from overseas and interstate? For small geographic areas, building activity and household formation becomes important. Demographers can only make assumptions based on what is known at the time forecasts are produced. They cannot include implicit assumptions about future economic conditions or even pandemics. 

The level of uncertainty is tempered somewhat by the inherent momentum in some demographic variables. For example, the number of births relates to the size of the female population of child-bearing age. Other factors such as access to birth control, population health and workforce participation are also important. The fertility rate indicates the propensity to have children, but the number of births can still be high even when the fertility rate declines. The number of births and deaths does not change significantly from year to year and this provides demographers with a degree of certainty when preparing forecast assumptions.

Migration on the other hand is highly volatile. This is because it responds to economic drivers, as well as government policy. A great example is the closure of the international border, which had an immediate impact on the level of overseas migration. In 2020-21, Australia recorded a net overseas migration (NOM) loss of 88,760 people ie more departures than arrivals. Given the high level of NOM driving population growth in the decade or so beforehand, this was a highly impactful and sudden departure from this trend. The feeling is that it will rebound quickly, as the 2021 Intergenerational Report assumes that NOM will recover to 235,000 people per annum by 2024-25. 

Interstate migration trends have also changed considerably in the last two years. Until 2020, Victoria gained population from other states, but the situation has changed dramatically and the state now loses population to other parts of Australia. The volume of interstate migration gain in Queensland has increased, and there have also been increases in Western Australia and South Australia. These trends weren't necessarily a result of the pandemic, but it certainly gave them a boost. For example, net interstate migration gain in Victoria was declining in the years beforehand, but accelerated rapidly from early 2020.

Intrastate migration trends have also shifted considerably. The property market in regional Australia has been turbocharged in the last two years due to increased demand (and possibly less supply). Much of this has been attributed to working from home trends, and how this breaks the nexus between location of home versus employment. The extent to which this is occurring, and its longevity, is still not quantified. The 2021 Census will inform the story somewhat via the question on usual residence 1 year and 5 years ago. Other data, such as that held by councils and other government authorities may also prove useful.

Building activity data is largely overlooked, and this is available for local areas on a frequent basis. The chart below shows building approvals for selected SA2s in the Geelong and Bellarine Peninsula region since 2016. This is one of the regions often cited as a destination for flee changers, but what does the data tells us? 

Since 2016 there have been more than 19,300 building approvals in this region. While not all will become dwellings on the ground, it's still indicative of strong residential construction. The Grovedale and Ocean Grove - Barwon Heads SA2s accounted for almost half of all approvals in this period, with Clifton Springs a distant third. These regions contain new housing estates, such as Armstrong Creek, that are generally attractive to first home buyers. There was a distinct increase in the number of approvals after March 2020. Given current migration trends, it's likely that these are not only from within Geelong itself, but from Melbourne as well. Government incentives to stimulate the building sector as part of COVID-19 recovery strategies would have also influenced this trend. This sort of information, and the spatial patterns, informs dwelling assumptions for small area population forecasts.

Building approvals tell one part of the story, sales of existing residential properties tell another. The median house price in Greater Geelong has increased rapidly since March 2020, reaching $765,000 in the December 2021 quarter. This is around 30% lower than the median price in metropolitan Melbourne. From a housing affordability perspective, this is a strong attractor, and another important indicator of population change. Although a lot of houses are being built, and there is a clear demand for regional properties, the question remains - who is living in them and how are they occupied. 

A key question for demographers is how these trends will play out in the future. Although we live in interesting and uncertain times, population forecasts are still a key tool for anyone involved in planning services and infrastructure. Planners need the data to inform their work, so it's important to consider the following:

• users should ensure they understand the assumptions underpinning the forecast. Do they reflect reality? What has changed in your local area?
• when were the forecasts prepared? If they predate the pandemic, users should take note of how more recent trends may impact future population, particularly the extent to which overseas migration drives population change. In some situations, frequent updates may be appropriate.
• what timeframe is required for planning purposes? Regardless of the pandemic, longer time frames are less likely to be reliable.
• are there other data available that may inform the assumptions? For small area forecasts, data held by Councils can be an untapped resource. This includes rubbish bin allocations (particularly in new residential areas), and kindergarten enrolments.
• using population scenarios to reflect demographic changes. The Queensland state government forecasts produce high, medium and low scenarios, which provides some flexibility with regard an appropriate assumption for a user's needs. 

Concluding comments

Despite the uncertainties arising from the COVID-19 pandemic, there is still enough demographic and other data available to make informed assumptions about future population trends. Some of the shifts we have seen have started to become more of a long term trend, such as the movement of people out of Victoria. The release of the 2021 Census data (from mid 2022) and the 2021 Estimated Resident Population (late March 2022) will further inform the demographic story. This is particularly true for assessing migration trends and household occupancy. Small area population forecasts have not lost their relevance - they remain a critical tool for planning a wide range of services.