A survival-analysis study of 30.3 million property transactions covering 10.4 million Australian addresses from 1990 to April 2026. After censoring renovation-and-flip resales using the FIRST hold's reno flags, the Kaplan-Meier median hold is 8.61 years. About 34 per cent of buyers resell within five years and 55 per cent within ten. The 2022 to 2025 rate cycle pushed short-hold flips to a thirty-year high.
Microburbs Research whitepaper. Luke Metcalfe. 21 May 2026 at 13:16 AEST. Data as at April 2026.Abstract. We apply Kaplan-Meier survival estimation to the universe of property transactions in Australia from 1990 to April 2026. We measure the cumulative probability of resale by year of holding, by purchase cohort, by year of sale, and by tenure type. After removing renovation-and-flip resales, we find a national Kaplan-Meier median hold of 8.61 years. About 34 per cent of buyers resell within five years and 55 per cent within ten. NSW Treasury's 2022 holding-periods paper (TRP22-13) reports a 9.7-year median for NSW buyers. CoreLogic's Pain & Gain reports a national median hold period of resales typically around 8.8 to 9.0 years on a completed-holds basis. Both treat completed holds only and do not Kaplan-Meier-correct for the 36 per cent of properties that have not yet resold — which is why our K-M median sits at or just below them. We document a structural shift over the last two decades, with the annual resale rate of mature stock falling from a 2002 peak of 10.6 per cent to a 2023-25 trough of 5.4 per cent, the lowest on record. The 2022 to 2025 cash rate cycle pushed first-year resell rates for new buyers to 18 per cent, the highest mark since at least 1990. Investors earn about one percentage point per year less than owner-occupiers in the 2 to 15 year hold buckets, although investors outperform owner-occupiers in the 0 to 1 year flip bucket by 2.1 percentage points and in the 1 to 2 and 20-plus year buckets the gap is half a percentage point or less. The May 2026 budget changes to negative gearing and capital gains tax create a defined treatment date for further work.
Robustness. The headline is stable across sensitivity tests for dwelling type and for how aggressively renovation-and-flip resales are removed, holding within an 8.6 to 9.6 year band that brackets the NSW Treasury figure.
All figures use a renovation-stripped Kaplan-Meier estimate that corrects for properties not yet resold. The result agrees with the NSW Treasury 2022 holding-periods paper and CoreLogic Pain and Gain.
The financial industry quotes hold times of fifteen, twenty, or thirty years and recommends behaviour to match. The data shows that almost nobody actually holds that long. Half of all Australian property has resold within ten years of purchase. The median completed hold sits around five years for properties sold today. Investors who plan around long holds are planning for an outcome most buyers never reach. Recent buyers, especially those who entered the market in 2020 to 2022, are flipping within a year at the highest rate in three decades.
Investment guidance, tax modelling, and portfolio assumptions all need to reflect real hold behaviour rather than industry mythology. The strongest predictor of how long an investor will hold is not their stated intent at purchase, it is the cash rate trajectory over the years that follow.
The cohort survival grid below tracks every purchase cohort from 1990 to 2025. Each cohort is a group of properties first transacted in a single calendar year. We follow each cohort forward and measure what fraction had resold by year 1, 2, 3, 5, 10, 15, and 20.
Each line is what fraction of properties bought in year X had been resold by N years after purchase.
A property bought in 2005 had a 15 per cent chance of resale within twelve months, more than three times the 1995 rate of 4.6 per cent. The 2002 to 2007 cohorts hit 39 to 44 per cent resold within five years, the highest sustained mark in our sample. Rising prices created exit opportunities and the cash rate was low (4.25 to 6.75 per cent), encouraging refinance and trading up.
The 2009 cohort dropped 5.4 percentage points on the 5-year resell rate, the single largest year-on-year fall in our series. From 2011 to 2018, cohorts stabilised around 32 to 34 per cent resold within five years, eight to ten percentage points below the 2005 peak. APRA's December 2014 ten per cent investor lending growth benchmark and March 2017 thirty per cent interest-only cap appear to have reinforced the post-GFC slowdown for the investor cohort.
The 2022 cohort posted an 18.4 per cent 1-year resell rate, the highest 1-year mark in three decades. The 2023 cohort followed at 17.2 per cent. The timing tracks the cash rate. The RBA lifted from 0.10 per cent in May 2022 to 4.35 per cent in November 2023. Many recent buyers had purchased at historic-low rates and could not sustain repayments at the higher level. Forced sales drove the 2022 and 2023 short-hold spike. Easing started in February 2025 but has not yet restored turnover.
Distinct properties resold in a given year, as a share of properties first transacted at least five years earlier. Controls for address-data coverage growth over time. Vertical dotted lines mark policy and macro events (CGT 50 per cent discount 1999, GST 2000, FHB Boost 2008, APRA investor and interest-only caps 2014 and 2017, COVID emergency cuts 2020, RBA hikes 2022, Stage 3 tax cuts 2024, RBA easing 2025, 2026 budget reform).
Median completed-hold duration of properties that sold each year, post-2000 only because pre-2000 address-data coverage is patchy. This is the conditional median: of the properties that ARE sold in year Y, how long had they been held? Bottomed at 2.45 years in 2008 (forced sales during GFC), peaked at 5.25 years in 2022, falling back as the rate-shock cohort exits.
Three events produce clean, identifiable shifts in the resold rate or the short-hold share.
Two events did not produce identifiable shifts in our data. The September 1999 CGT 50 per cent discount produced no visible step, though address-data coverage of the 1990s is patchy enough that a real effect could have been swamped. The May 2017 foreign investor changes also did not register at the national resale rate.
The May 2026 budget changes to negative gearing and capital gains tax create a defined treatment date for further work. The reform restricts negative gearing to new builds for purchases from 12 May 2026 onwards, with the CGT change taking effect 1 July 2027. Grandfathering means existing investors are not forced to sell but are unlikely to add to their position. We expect to see the lock-in effect (existing investors holding longer) in the data over the next 24 months.
Beyond the national curve, we score every Australian property for how likely it is to sell in the years ahead. The score is built only from place and property data, with no access to any individual owner's circumstances, so we read it as a ranking of areas and property types rather than a forecast for any single household.
We measure accuracy on a year the model never sees while it is being built. On that hold-out year it reliably ranks the properties that go on to sell above those that do not, and the ranking carries through to the full national stock. A property's own characteristics and its location do most of the work.
Aggregated to suburb level (minimum 500 properties), the score identifies where turnover is most and least likely over the next five years. The top of the list is dominated by coastal NSW and regional Queensland short-stay markets. Cliftleigh (NSW) tops the list at 48.2 per cent of homes likely to sell, followed by River Heads (QLD) at 45.1, Tallwoods Village (NSW) at 44.6, Queenstown (Tas.) at 44.4, and Russell Island (QLD) at 43.3. The lowest-turnover end is dominated by Perth new-build land-release suburbs and a handful of Sydney apartment clusters at 15 to 18 per cent.
We linked brochure-style feature tags for 1.8 million listed properties to their actual hold durations, to see which kinds of home change hands soonest.
Properties with modern, upscale, lifestyle features hold for the shortest periods. Butler's pantry properties hold for 6.54 years against an 8.18 year baseline. Media rooms hold 6.56 years. Gated communities 6.64 years. Theatre rooms 7.40 years. Golf course views 7.34 years. LED lighting 7.37 years. These are typically new-build or recently-renovated houses in master-planned communities, bought by dual-income professionals or lifestyle downsizers who trade more often.
Properties with older, character, or development-potential features hold the longest. Renovation potential 10.16 years. Gas heater 9.91 years. Powered workshop 9.84 years. Double brick 9.47 years. Period features 9.43 years. Cul-de-sac 9.49 years. One bathroom 9.47 years. These properties skew toward original owners who bought decades ago and stayed for life events rather than market timing.
The forward sale-likelihood scores show the same pattern. Properties tagged "gated community" have a mean 5-year likelihood of selling of 37.8 per cent against a 31.5 per cent baseline. Beachfront 37.2, waterfront 36.2, new construction 35.7, ocean views 34.7, dual living 34.8, butler's pantry 34.8. Long-holders include hydronic heating 29.4 per cent, open fireplace 29.3, pergola 29.4, gym 28.7, lock-up garage 28.9, roller shutters 28.0, ducted evaporative air-conditioning 27.8. The forward scores and the historical holds agree independently.
For investor decision-making, the read is that modern lifestyle homes carry a built-in five-year exit timeline and period houses carry a ten-plus-year holding profile. If the investment plan needs flexibility to exit at 5 years, new turn-key houses fit. If the plan is a long-term hold compounding capital growth, older character properties match the typical holding pattern of the cohort that already owns those houses.
We use Kaplan-Meier survival estimation rather than a simple average of completed holds, because about a quarter of properties bought before 2020 have not yet resold. Ignoring those unsold properties would understate true hold times. The survival method corrects for it and answers, without bias, what share of buyers have sold by year N.
The result holds up against the authoritative external benchmarks. Our median sits just below the 9.7-year figure published by NSW Treasury and the 8.8 to 9.0 years reported by CoreLogic, exactly as expected once the correction for not-yet-resold properties is applied. The boom-era turnover of 2005 to 2008 and the rate-shock flips of 2022 to 2025 also match RBA and AHURI commentary on the same periods.
Two practical implications follow from these findings.
Plan around the 5-year horizon. Industry guidance to hold for fifteen or twenty years is not what most people do. Half of all Australian property has changed hands within ten years of purchase. Investors and homeowners should model returns assuming a 5-year exit window and treat anything longer as a bonus.
Watch the cash rate trajectory. The strongest predictor of short-hold behaviour over the last three decades is the cash rate trajectory. Boom-and-low-rate cohorts traded fast (2005-2008). High-rate-shock cohorts also traded fast, but for the opposite reason (2022-2024). Cohorts that bought into stable mid-cycle markets (1995-2000, 2011-2018) held longest.
The 2026 negative gearing reform is unlikely to be the dominant driver of hold-time behaviour over the next five years. Cash rate movements still matter more, and property selection within a suburb dominates either of these. See the companion family-home versus investor work for that comparison.
The CGT exit pattern is real but moderate. Suburbs in the top quintile of historical annualised return show 3.6 percentage points more 5-year sale likelihood than bottom-quintile suburbs. People crystallise gains. The pattern holds for owner-occupiers as well as investors.
A plain-English summary of these findings is on the companion landing page.
