Why .id's population forecasts differ from state government forecasts

State planning departments publish their own population forecasts, such as Victoria in Future (VIF), WA Tomorrow, the NSW Department of Planning's forecasts and South Australia's PlanSA figures. These often differ from the .id forecast for the same area, sometimes by a widening margin in later years. The difference is not an error in either forecast. It reflects different purposes, methods, assumptions and timing, and understanding those differences lets you use both sets of figures with confidence.

Key concepts

Top-down and bottom-up forecasting

A top-down forecast starts with national, state and regional trends and distributes them across areas. State governments generally forecast this way. A bottom-up forecast starts with what is happening on the ground in a local area, such as development sites and local migration patterns. At .id, we combine the two through our National Forecasting Program: top-down demographic constraints at the national, state and regional (SA4) level, with detailed bottom-up development and land supply assumptions at the local level.

Constrained and unconstrained forecasts

Our forecast for any area must add up. Every local area sits within a regional total, every region within a state total and every state within a national total, so growth assumed in one place cannot also be counted somewhere else. A state model is built for its own state and is not constrained to a national total in the same way, which is one reason a state figure can sit above ours for the same area over the same period.

Our forecasts add up. Each local area sits within a regional total, each region within a state total, and each state within a national total. Growth counted in one place cannot be counted again somewhere else. A state government forecast is built for its own state and is not tied to a national total in the same way. This is one reason a state figure can be higher than ours for the same area over the same period.

Base year and data vintage

Forecasts are prepared at different times from different inputs. The Australian Bureau of Statistics publishes Estimated Resident Population (ERP) about a year in arrears, and each ERP vintage is revised later, with a fuller revision after each Census. For example, the 2025 ABS ERP release did not just give the population figure for 2024 but revised the 2022 and 2023 ERP figures too. Two forecasts prepared a year or two apart can start from different base populations and different views of migration, so part of any gap is simply timing. Checking when each forecast was prepared, and from which base year, is the first step in explaining a difference.

Why the figures differ

The usual sources of a gap between an .id forecast and a state government forecast come down to four things:

Components of population change. We may make different assumptions about net overseas migration, interstate migration and natural increase.
Land supply and dwelling capacity. We assess how much development an area can realistically carry and how long greenfield areas take to build out. In established areas this includes whether medium and high-density development is feasible to deliver, not only whether planning controls allow it. Both factors can cause the State model to carry growth in an area longer than supply or feasibility allows.
Household size and vacancy rates. Different assumptions can produce a different population from a similar number of dwellings.
Preparation dates and base years, as above.

These differences flow from purpose. State forecasts support state-level planning and are often the mandated reference for statutory processes. Our forecast is built to support local decisions, such as service planning, infrastructure staging and facility networks, so it invests in local land supply evidence and small-area detail, and it is presented as a single most likely outcome rather than a range of scenarios.

What to do when the two diverge

Start by comparing vintages: a forecast prepared after a Census rebase or a major migration shift will differ from one prepared before it. Then compare assumptions. The assumptions behind your .id forecast, including the residential development assumptions, are published with your forecast site so the basis of each figure can be examined side by side, and state departments publish their assumptions too. In most cases the gap comes down to a small number of identifiable assumption differences, and naming them is usually enough for a report or a briefing.

Where state guidance directs you to a state figure for a statutory purpose, use it for that purpose. The two sets of figures answer different questions, and councils routinely cite the state figure for statutory planning while using the .id forecast for service and infrastructure planning, where the small-area detail matters. If a gap for your area looks larger than assumption differences can explain, raise it with us: forecasts are reviewed on a regular cycle, your feedback on development assumptions feeds into the next review, and we can walk through the comparison with you.

Where this appears in Population Forecast

Your Population Forecast site presents the .id forecast for your local government area and its small areas, with the assumptions behind it available alongside the results. The residential development assumptions show the development expected in each area, which is where differences in land supply views become visible. Because the forecast is a single most likely outcome, you will not find high and low scenarios to match against a state model's range: the comparison is between the state figure you are using and the one .id forecast.

Frequently Asked Questions

Q: The state government's forecast for our area is higher than .id's. Which one should we use?

A: It depends on the decision. Where a state process mandates the state figure, use it for that process. For local service and infrastructure planning, the .id forecast gives small-area detail and development-level evidence the state figure does not. Many councils use both, and say so: the state figure for statutory alignment, the .id forecast for local planning.

Q: Why is the .id forecast often lower in our growth areas?

A: Usually because of land supply and build-out assumptions. We assess the capacity of each area and how quickly development can realistically be delivered, and we constrain total growth to the regional and state outlook. A model without those constraints can carry strong growth in an area for longer than the land, identified capacity, and the building industry can support.

Q: Does .id take the state government's forecast into account?

A: We prepare our forecasts independently rather than calibrating to state figures, but we draw on the same foundation data, including ABS ERP. Our development assumptions are underpinned by our Residential Development Layer, which is built from planning information such as planning instruments, growth strategies and development approvals, and they are reviewed on a regular cycle with council feedback incorporated, so changes in those settings reach the forecast at its next review.

Q: The gap between the two forecasts grows over time. Why?

A: Small annual differences in assumed migration, fertility or dwelling delivery compound. A forecast that assumes slightly faster growth every year will sit well above a more constrained one by the end of a 25-year horizon. That widening pattern is expected whenever two models carry different growth rates, and it is why comparing the underlying assumptions is more informative than comparing the end-year totals.