Annual rental turnover across 313 NSW postcodes has fallen from 41.1% in the year to December quarter 2018 to 34.0% in the year to June quarter 2026. Turnover fell in 299 of 313 postcodes measured against themselves. Nobody publishes this figure; both counts it is built from are published every quarter.
Beside every rent figure, NSW publishes bonds lodged during the quarter and bonds held at the end of it. One is a flow and the other a stock, so together they close an accounting identity: every bond that existed and no longer does was closed. Turnover is lodgements over the pool they entered. It is arithmetic on two published counts, not a model.
The obvious objection is that this is COVID. The aggregate invites it: turnover sat near 41.1% until mid-2021 and stepped down during the Delta lockdown. So every window is shown below, including the ones that flatter the finding least, with each postcode compared only against itself.
| Window | Span | Postcodes that fell | Share | Median move |
|---|---|---|---|---|
| The whole spaneverything the identity can be computed over | December quarter 2018 → June quarter 2026 | 299 of 313 | 96% | -8.50pp |
| Before COVIDends the quarter before the Delta lockdown began | December quarter 2018 → June quarter 2021 | 199 of 313 | 64% | -3.31pp |
| Across the 2021 stepthe year containing the lockdown | June quarter 2021 → June quarter 2022 | 224 of 313 | 72% | -2.39pp |
| Since COVIDbegins after restrictions ended, so no lockdown quarter is inside it | June quarter 2022 → June quarter 2026 | 249 of 313 | 80% | -3.77pp |
| The last three yearsmore recent still | June quarter 2023 → June quarter 2026 | 276 of 313 | 88% | -3.48pp |
| The last two yearsthe shortest window shown | June quarter 2024 → June quarter 2026 | 230 of 313 | 73% | -2.02pp |
Turnover fell in a majority of postcodes in every window, including the years before the pandemic and the four years since restrictions ended. Taking only June quarter 2022 to June quarter 2026 — no lockdown quarter inside it — turnover still fell in 249 of 313 postcodes, a median of -3.77pp. The pandemic is visible in the series and is not the cause of it.
Turnover counts bonds, not tenancies. A tenancy that changes hands by bond transfer — a housemate replaced, a change of tenant on the same bond — produces neither a lodgement nor a closure, so the real churn is higher than this and this figure is a floor.
It is tempting to divide one by turnover and report an average tenancy length. We do not publish that number. It is a steady-state identity, true only where turnover is constant, and the finding here is precisely that it is falling — so the division gives the duration tenancies would reach if today’s churn persisted, not the length of any tenancy that exists. At 34.0% that would be about 2.9 years, up from 2.4; it holds only if nothing changes further, which is the one thing the chart argues against.
Source: NSW Department of Communities and Justice Rent and Sales Reports, used under CC BY 4.0. Computed on a fixed panel of 313 postcodes reporting both counts in every quarter — a panel chosen on the stock alone yields nothing, because at least one stock-complete postcode has a suppressed lodgement count in all 36 quarters. Suppressed cells are never imputed. The series starts at December quarter 2018 because DCJ restated the bonds-held column between the September and December quarters of 2017, a −7.4% step that moved 99% of postcodes together, so no change is calculated across it. Figures are trailing four-quarter rates: every December quarter shows the pool falling and the following March jumping, which is partly Christmas and partly a bonds registry working through a holiday backlog, and four quarters cancel both exactly with no seasonal model. The live figures · Sources and licences.