Negative gearing and the 50% CGT discount cost the federal budget $12.3 billion last year in property-specific terms. The settings are working as written.

The two settings are routinely defended on three grounds: that they help ordinary Australians build wealth, that they fund new rental supply, and that any reform would push rents higher. Each of these claims has been tested. None of them survives the data.

The mechanism is simple. Negative gearing allows a property investor to deduct rental losses against any other income, including wages and salary, at the investor's marginal tax rate. Australia is one of the few comparable jurisdictions that does not quarantine these losses to rental income. The CGT discount halves the taxable portion of any capital gain on an asset held for more than twelve months, regardless of whether the gain is real or inflationary09. The two settings compound. An investor borrows to acquire a dwelling, deducts the holding losses against wages at the top marginal rate, then realises a gain that is taxed at half the headline rate.

Who actually benefits

The case for the current settings rests on three claims. The data on each is on the public record.

ACOSS, working from the most recent ATO Taxation Statistics, finds that 82% of the CGT discount value flows to the top 10% of households. 39% of negative gearing deductions also accrue to the top 10%. The wealthiest 10% of households hold roughly two-thirds of the value of all investment property04. VERIFIED

The CGT individual discount, item E15, is among the largest tax expenditures in the personal income tax system, with revenue forgone highly sensitive to the underlying income, capital gains, and macroeconomic parameters used in the estimate.

Treasury Tax Expenditures and Insights Statement 2024-25, item E15 (paraphrase)

The "mum and dad investor" framing collapses against this distribution. The two settings are most valuable to households on the highest marginal rates, with the most existing equity to leverage, and the longest time horizon to hold assets through a cycle. They are not designed to assist median-income households. They have not done so at any point in the available data.

The supply argument fails on the same data set. ACOSS analysis of investment loan data finds that 81% of investment property loans are taken out against existing dwellings, not new construction04. VERIFIED The policy transfers existing housing between owners. It does not build new housing.

The cumulative effect is visible in the published affordability series. The Demographia International Housing Affordability Survey records the median dwelling price as a multiple of gross median household income - the conventional "median multiple" measure used internationally. As recently as 1990, the Australian national ratio was approximately 3.0×. As of Q3 2023, it was 9.7× nationally and 13.8× in Sydney specifically; Sydney ranked second-least-affordable of 94 major markets surveyed globally that year06. ESTIMATED

What the independent modelling shows

The supply claim has been articulated directly by the Prime Minister.

"The problem is all of the analysis shows that a change to negative gearing will not assist supply."

Anthony Albanese, Sky News interview, 25 September 2024.

Variants of the formula have circulated through television and 2026 budget cycle commentary - that "the analysis" or "the research" shows reform would reduce supply16. The analysis is rarely identified.

Where the claim has been pinned to a specific paper, the contents of the paper do not match the framing it is given in public. The Australia Institute traced the Deloitte modelling cited publicly by the Prime Minister against reform back to its source. The modelling concluded that housing would become more affordable under the reform and that rents would be barely affected08. Both findings are incompatible with a public claim that supply would fall: a supply collapse would push prices and rents higher, not the other way. The headline framing of the modelling and the contents of the modelling were inverted at the point of public communication.

The "fewer houses" framing also rests on a category error. Investors do not manufacture dwellings. They hold them. When an investor sells, the dwelling does not leave the housing stock - it acquires a new owner. If the new owner is another investor, rental supply is unchanged. If the new owner is an owner-occupier, rental supply falls by one unit and rental demand falls by one unit simultaneously, because the buyer was previously renting. The two effects net at the level of the housing market. The dwelling stock is set by construction, demolition, and population. It is not set by the tax treatment of who holds the title.

The dominant claim against reform during the 2016 to 2019 federal election cycles was distinct: that any tightening of negative gearing would cause rents to rise sharply. The claim was anchored to industry-commissioned modelling and amplified through News Corp mastheads. Its funding source was not disclosed in most citations.

Independent modelling reaches the opposite conclusion. The Grattan Institute's Hot Property report finds that property prices would be up to 2% lower under proposed reforms, and that rents would not materially change07. ESTIMATED

"Property prices would be up to 2 per cent lower under our proposed reforms. Contrary to urban myth, rents won't change much, nor will housing markets collapse."

Grattan Institute, Hot Property, 2016

The folk memory beneath the modelling debate is older. Industry briefs in the 2026 budget cycle have revived the claim that rents rose 30% nationally when the Hawke government quarantined negative gearing between July 1985 and September 1987. The ABS rental price series for the same window does not support the national framing.

The ABS Consumer Price Index publishes the Rents subgroup at the eight-capital-cities aggregate. Across the quarantine window, Q3 1985 to Q3 1987, that index rose approximately 21.7% in nominal terms1415. The All Groups CPI for the eight-capital-cities aggregate over the same window rose approximately 17.8%1415. The implied real rent growth nationally across the entire 27-month quarantine was approximately 3.3% - meaningfully positive, but a fraction of the 30 per cent figure being recycled into the 2026 budget framing. ESTIMATED

The ABS does not publicly publish the Rents subgroup by individual capital city for this period. The September 1987 issue of Catalogue 6401.0 prints subgroup figures only at the eight-capital-cities aggregate (Table 7); the disaggregated city series exists in ABS internal time-series stores but is not exposed in any public table, time-series spreadsheet, or current dataflow1415. The per-city nominal rent figures below are reproduced from authors (Eslake, Grattan, MacroBusiness) who accessed the underlying ABS series via the Information Consultancy service. The geographic disaggregation is original analytical work by those authors, not a re-quoted public table. The CPI inflation figures alongside them are drawn directly from the ABS Data Explorer All Groups series by capital city for the same window15.

The 1985-87 quarantine ran during a high-inflation period: general CPI rose roughly seven to eight per cent per year across the window, far above the two to three per cent inflation typical of the 2010s and 2020s. Nominal rent figures from this era have to be deflated to be comparable to anything contemporary. Once they are, even Sydney and Perth - the cities the industry case rests on - show real rent rises of roughly nine to ten and a half per cent over the entire 27-month period. That is a fraction of the 30 per cent figure circulating in 2026, and roughly a third of the nominal rises being quoted out of context.

Negative gearing was withdrawn nationally. If the policy change had been the cause, rent rises should have appeared in every capital. They appeared in two. Both were running record-low vacancy rates entering the period: Sydney was just below 2%, Perth slightly above 2%12. VERIFIED The vacancy rate in Sydney had been driven down by interstate migration. The vacancy rate in Perth had been driven down by the second half of the WA mining cycle.

"Negative gearing was withdrawn nationally, so if there was to have been any impact as a direct result of that change it should have shown up in all cities since they were all affected by that policy change, but they only showed up in Perth and Sydney because in those two cities the vacancy rate was unusually low."

Saul Eslake, Negative gearing and the policy promise, address to Prosper Australia, February 2016.

The Grattan Institute reached the same conclusion on the historical record in 2015: real rents did not rise in Melbourne, Brisbane or Adelaide; the Sydney rise was driven by other factors13. The folk memory survives because it is convenient to those who benefit from the current settings. The data underneath it is on the public record.

What reform actually proposes

The reform on the policy table now, as proposed by the Parliamentary Budget Office, Grattan, the Australia Institute, and a series of independent economists, takes the following shape. All existing arrangements are grandfathered. New investors may deduct rental losses only against rental income, not against wage and salary. The CGT discount is reduced from 50% to 25% for new contracts only. Existing holdings remain on the current 50% discount until disposal. Recovered revenue is hypothecated to social and affordable housing07.

The grandfathering is the load-bearing feature. No existing investor loses access to current settings. The change is purely prospective. Steady-state revenue recovery takes a decade or more, which is the principal economic cost of the proposal as drafted. It is not a shock. It is the slowest plausible reform of the policy.

The settings as currently written are doing what the Income Tax Assessment Act 1997 was amended to permit them to do09. VERIFIED The argument that reform threatens existing investors is, on the face of the draft proposal, false. The argument that reform would push rents higher is contradicted by the independent modelling that exists on the question0708. The argument that the current settings help ordinary Australians is contradicted by the distributional data published by the ATO every year04.

The case against reform is, on the public record, made by the people the reform would touch. That is not, by itself, a reason to dismiss the case. It is a reason to read the case with attention to who paid for it.