How the NIEIR economic model works

The State of the Regions dataset is prepared by the National Institute of Economic and Industry Research (NIEIR) and published by .id (Informed Decisions). As detailed economic data is not available annually at the local government area level, these indicators are modelled using a regional economic model aligned to the national accounts framework, drawing on a range of data sources. These include Australian Bureau of Statistics (ABS) data - such as the Census, the Labour Force Survey and the State and National Accounts - and administrative data sources such as ATO income tax estimates and Centrelink payments.

Key concepts

Modelled, not surveyed

No agency surveys the whole economy of every local government area every year. A model fills that gap by combining the data that does exist - tax records, survey estimates, Census benchmarks, administrative payments - into a consistent annual picture, reconciled so that local estimates add up to official ABS state and national totals. The model is updated each year as new and revised data becomes available, so estimates for all years, including GRP and local jobs, may be revised between releases. Revisions are typically greater for the most recent years.

Bottom-up and top-down

There is no standard method for estimating GRP below state level in Australia, and providers choose between two broad approaches. A top-down model takes a state total and shares it out to local areas, assuming every worker in an industry produces the state average. A bottom-up model builds each area up from its own local data, giving each area and industry its own production relationships, then reconciles the results to state and national totals. NIEIR's model is bottom-up.

The plain-language version .id has used to explain the difference: a Coles supermarket in Port Phillip trades at a higher level than a Coles in Preston, and a financial services head office in Docklands is a very different business from a branch in a regional town. A bottom-up model driven by ATO statistics picks that up. A top-down method assumes both are average.

An input-output model of every local economy

At .id, we use NIEIR's dynamic inter-regional input-output model, which covers every Australian local government area. Each area has its own input-output and household income sub-model, linked to every other area by trade flows between industries and the journey-to-work patterns of commuters. For each industry the model carries employment, hours of work, incomes, exports and imports, consumption and investment, so a change in one part of one local economy flows through to the rest.

The main data inputs

ATO income tax statistics by postcode - employment totals, household income and the local productivity differences used to value production. These arrive with a lag of 12 to 18 months.
ABS Census of Population and Housing - journey-to-work patterns, employment benchmarks and household structure, every five years.
ABS Labour Force Survey - the industry structure of employment between Censuses, plus hours worked.
ABS state and national accounts - the control totals the regional estimates reconcile to.
Centrelink payments by postcode and the Small Area Labour Markets series - recent-period employment and unemployment.
Industry-specific sources such as agricultural values and Tourism Research Australia expenditure, used for calibration.
Building approvals and floorspace completions - inter-Census estimates of industry capacity and the journey-to-work update.
Local knowledge - where a council advises .id of a plant closure, a major development or a specific employer decision, that information is substituted for the model-based estimate.

Frequently asked questions

Q: Why should we trust modelled data over ABS published data?

A: The two are not rivals - the model is built on ABS data and reconciles to ABS state and national totals. What the model adds is an annual, industry-level local picture the ABS does not publish, using local tax data to reflect how your area's industries actually perform rather than assuming state averages.

Q: How does the model fill the gaps between Censuses?

A: The Census provides five-yearly benchmarks. Between them, annual tax data, the quarterly Labour Force Survey, Centrelink payments and building approvals keep the estimates current, and each new Census re-anchors the series.

Q: The figures don't match what local businesses are telling us. What should we do?

A: Tell us what you are seeing, with as much specificity as you can - the business, the industry, the approximate number of jobs and the timing. Local knowledge is a documented model input: .id asks councils each year whether they have observed closures or major employment losses, and feeds the answers into the quality assurance of the next release.

Q: Why do the numbers change between annual releases?

A: The dataset is remodelled every year, not just extended, so every year in the series can move as new data arrives. See 'Why economic figures change between annual releases'.