How Many Rental Homes Sit Empty, Microburb by Microburb

Abstract

Rental vacancy is normally published as one number per suburb, hiding the fact that a single suburb contains very different streets. This research resolves vacancy below the suburb, to the Microburb — an area of roughly thirty to sixty homes — for the whole country. Because a small area holds only a handful of rentals, we express the result the honest way: the likely number of rental homes empty right now, usually none or one, with a finer underlying reading shown as shading.

Accuracy, stated plainly

We check the work by adding each Microburb back up to its suburb and comparing against established suburb vacancy. Across 2,737 suburbs with a published figure to check against, the rolled-up estimate lands within range for about 764 of them (28%); for a typical suburb it currently reads about half the published rate. In plain terms: the method reliably tells you which streets in a suburb are tighter than others, but it currently understates the exact rate, and a calibration step is the next fix. We state this rather than dress the number up.

Key Findings

• Vacancy genuinely varies within a suburb. Streets a few hundred metres apart differ in how often their rentals sit empty. A single suburb-wide number erases that.
• Most small areas have none or one empty rental at any moment. Across the country the large majority of Microburbs are "likely 0 vacant right now", with a minority — almost always larger apartment blocks — at one or more.
• The ranking is reliable; the absolute level is not yet. Where a suburb has solid data the rolled-up figure sits close to the published rate; on a typical suburb it currently reads low. That points to a calibration step plus a sharper read on apartment-heavy and sparse areas.
• Capital growth impact. Lower vacancy reflects tighter rental demand, which supports rents and yields and reduces holding risk. The effect on capital growth is indirect and not separately measured here; we say so plainly rather than imply a number.

How It Works (at a level that protects the method)

For each rental home we estimate two everyday things: how often it comes up for rent, and how long it tends to take to find a tenant. Combined across the rental homes in a Microburb, those give the share of local rentals likely to be empty at any moment — and, multiplied by how many rentals the area holds, the likely number empty right now. We deliberately present the count, not a false-precise percentage. We validate by aggregation to the suburb, and we do not disclose data sources or the internal method.

Results

To illustrate, two well-sampled areas. In Subiaco (Perth) and Bondi (Sydney) the Microburb estimates roll up close to the established suburb figure while still showing real street-to-street variation underneath. These two are illustrations, not proof; the validation is the suburb-level reconciliation across 2,737 suburbs nationally.

Defence Against Criticism

"A small area is too small to measure vacancy." Correct for a precise rate, which is why we publish the likely count of empty homes, not a decimal percentage. "Half your suburbs are understated." True today, and stated up front; the understatement is systematic, which is exactly what a calibration step corrects, and the within-suburb ranking is unaffected by it. "This is just the suburb number spread around." No: each area carries its own reading from local activity, which is why areas within a suburb differ.

Limitations

The absolute rate currently understates established suburb vacancy on a typical suburb, driven by how the time-to-lease is estimated and by mixing houses with apartments; separating those and calibrating to the suburb benchmark are the next steps. Figures are built on data current to about mid-2025 and should be treated as a direction and a relative ranking, not a settled rate.

Conclusion

Rental tightness is one of the clearest signals a landlord has, and it has always stopped at the suburb boundary. Resolving it to the Microburb, nationwide, gives investors a street-level read the standard data cannot. The national rollout shows the ranking holds; calibrating the absolute level is the work ahead.