Understanding the Residential Development Forecast data schema

The Residential Development Forecast records major residential development sites across Australia. Each record is one boundary that describes what is being built there, how far it has progressed and the dwellings it is expected to deliver each year to 2046.

Key concepts

Site identity

Site_ID – unique identifier for the site. Use this to join. (Data export only)
Reference – a second unique identifier, used in research records.
(Data export only)
Site_Name – the site's researched name. Not unique; separate sites can share a name.
Address – street address. Also not unique.
CensusYear – the census the forecast is based on.

Only Site_ID and Reference are unique. Joining on Site_Name or Address will duplicate rows.

Geography and area

SA4_Code – the ABS Statistical Area Level 4 the site sits in.
SA4_Name – the SA4 name.
Area_HA – site area in hectares.

Development type

Apartments – attached dwellings stacked vertically with shared access.
Greenfield – new development on previously undeveloped land, generally on the urban fringe.
Redevelopment – a net increase in residential density on already-developed or urban land.
Demolition – dwellings removed with no residential replacement, so a net dwelling loss.

Apartments excludes villas and townhouses. They sit side by side, each with its own access, so they are recorded under Greenfield or Redevelopment depending on the land they are built on.

Not every type appears in every extract. A state with no demolition sites will have three types in the file, not four.

Status

Status records the furthest milestone a site has reached, as at the publish date.

Complete – the site was finished as at the publish date. Dwellings it delivered earlier in the forecast period still appear in the annual series.
Under construction – building work is underway on site.
Site clearing – site preparation and early earthworks have begun.
Marketing – lots or dwellings are actively for sale. No construction has started.
Approved – planning approval has been granted. Nothing built or sold.
Pre-approved / Lapsed approval – an application has been lodged, or a previous approval has expired or been refused.
Structure plan / Under-utilised (med-high density) – future dwellings are signalled by a structure plan, or by land under-utilised for its density potential.
Under-utilised (low density) – land is under-utilised for its low-density potential.
Potential long-term supply – long-term or residential investigation sites.

Selling is what separates Marketing from Approved. A site with an approval that has started selling carries Marketing.

A lapsed approval is not a dead site. It still signals developer interest, and often needs only a reduced dwelling count to be resubmitted, which makes it more likely to develop than land with no application history.

Dwelling counts

Capacity – total dwellings the site is expected to deliver at full buildout, whenever that happens.
CensusCnt – dwellings already on the site at the census. Zero on most sites.
Remaining – capacity not delivered inside the forecast period. These dwellings sit beyond 2046.

The annual dwelling series

One column per year, y2021 to y2046. Each holds the dwellings the site is expected to deliver in that year.

They are annual figures, not running totals. Add them to get the site's contribution across the period.

Values carry decimals. These are modelled allocations across years rather than counted dwellings, so a site can show 4.6 dwellings in a single year.

How the counts reconcile

Capacity = CensusCnt + the annual series + Remaining.

The three fields divide a site's capacity into what was already there, what arrives inside the forecast period, and what sits beyond it.

Small differences are expected. Capacity is a whole number while the annual series carries decimals, so a per-site total can be out by a fraction and a state-wide total by a few dozen dwellings across a thousand sites.

Timing

Start_Year – the first year the site is expected to deliver dwellings.
End_Year – the year by which its forecast dwellings are delivered.

Both are text fields, and both hold 'N/A' where the site is not expected to deliver dwellings inside the forecast period. They will not cast to a number until you handle that.

Version number and publish date

Version_No – a five-part value recording which input data the forecast is built on. One value for the file.
PubDate – the date the land development assessment was completed. One value per SA4.

Publish date varies within a single file. Each SA4 is researched on its own cycle, so a state extract can carry several publish dates months apart. A site's status is current as at the publish date for its SA4, not the date the file was sent.

The version number tells you which forecast inputs were used. It does not tell you when a site was assessed. Use the publish date for that.

The file you receive

A shapefile, delivered as five files: .shp, .dbf, .shx, .prj and .cpg.

Geometry – polygon. Site boundaries, not points.
Coordinate system – GCS_WGS_1984, geographic degrees.
Character encoding – ISO-8859-1.

Some sites are multipart polygons, so one site can be several separate parcels. Count records, not parts.

Upcoming changes

From 1 September 2026, status no longer carries a year. Appending the year produced too many values to work with, so status is now a fixed set you can theme a map on.

Publish date has also been added, so you can see when the assessment behind the data was done.

From 1 October 2026, the status wording changes. The values themselves are unchanged.

Status today	From 1 October 2026
Complete	Complete
Under construction	Under Construction
Site clearing	Site Work Commenced
Marketing	Sales Commenced
Approved	Development Approved
Pre-approved / Lapsed approval	Development Proposed
Structure plan / Under-utilised (med-high density)	Identified Growth Area
Under-utilised (low density)	Under-utilised Residential Land
Potential long-term supply	Potential Long-Term Supply

If you match on status text in a report or extract, map the September and October changes together and update it once.

Frequently asked questions

Q: Which field should I join on?

A: Site_ID. Site_Name and Address are not unique, so joining on either will duplicate rows.

Q: How do I tell when a site's status was set?

A: Use the publish date for that site's SA4. Publish dates vary within a file, so check the value on the record rather than assuming one date for the whole extract.

Q: A site shows Complete but has dwellings in the annual series. Why?

A: Status is the site's position as at the publish date. A site that finished partway through the forecast period is complete now, and the years in which it delivered dwellings still carry those dwellings.

Q: Why don't Capacity, CensusCnt, Remaining and the annual series add up exactly?

A: Capacity is a whole number and the annual series carries decimals. The difference is rounding, and it grows with the number of sites you are totalling.

Q: Are Start_Year and End_Year numbers?

A: No. They are text fields and they hold 'N/A' where a site delivers nothing inside the forecast period.

Q: My filter on status returned nothing. What happened?

A: Status values changed on 1 September 2026 and change again on 1 October 2026. A filter looking for a retired value returns no rows rather than failing. Map both sets of values from the table above.