The Widow-Maker

In February 2016, Jonathan Tepper of Variant Perception and John Hempton of Bronte Capital toured Sydney’s mortgage-broker belt undercover. Brokers, they said, coached them to lie on loan applications; banks “rarely verified payslips.” Tepper came home and declared “one of the biggest housing bubbles in history,” tipped falls of 30–50% in Sydney and Melbourne, and told clients to short the big four.[3][2] Bronte started shorting that month.

By mid-2016 the trade was crowded — about A$6.5bn of shorts across the four banks.[3] The logic was clean. The execution was a bloodbath.

Here is why it bled. A short-seller has to pay the banks’ fat, fully-franked dividends to whoever lent them the stock. The banks spent roughly A$7bn buying back their own shares.[1] And Commonwealth Bank simply went up — from A$5.40 at its 1991 float to about A$192 by mid-2025.[7] A correction that never comes is the most expensive thing in finance. The trade earned a nickname borrowed from the traders who spent decades shorting Japanese bonds: the widow-maker.

One myth is worth killing first. Steve Eisman, the investor made famous by The Big Short, never shorted Australian banks. His 2019 bet was on Canada.[4] The Australian short was its own home-grown obsession.

By June 2026 the short was bigger than it had ever been — a record ~A$11bn, roughly double in six months.[5] And in May, for the first time in years, it bit: CBA fell 10.4% in a single session on 13 May 2026, its worst day in 34 years of listing.[6] Whether that is the reckoning or another false alarm depends on the machine we are about to take apart.

Start with the tax code. Negative gearing lets an investor deduct the losses on a rental — when the rent doesn’t cover the interest and costs — against their wage income. Most countries ring-fence rental losses against rental income. Australia lets you write them straight off your salary.

The scale is enormous. In 2022–23, 1.12 million Australians — 49% of all property investors — ran their rentals at a deliberate loss, claiming A$10.4bn in net rental losses.[8] The benefit flows up the income ladder: roughly two-thirds of it lands with the top 30% of earners.[9]

It is not new, and it has been touched exactly once. In 1985 the Hawke government quarantined negative gearing; in 1987 it caved and brought it back.[11] Labor carried a reform plan — limiting it to new builds — into the 2016 and 2019 elections and lost both. The settings survived because millions of voters had built their retirement on them.[10]

If negative gearing lit the fire, the 1999 capital gains tax discount poured petrol on it. On 20 September 1999, the Howard government scrapped inflation-indexed capital gains and replaced them with a flat 50% discount on anything held longer than a year.[14]

Pair that with negative gearing and you get something close to a tax-arbitrage machine. The Reserve Bank spelled it out: the discount’s effect “is amplified if the asset can be purchased with leverage, because the interest deductions are calculated at the full marginal rate while the subsequent capital gains are taxed at half” — and “property is especially affected.”[12]

The data did what the incentives told it to. In the four years after the discount, real house prices jumped almost 50%.[23] Investors, who had been a fifth of new mortgage lending in the early 1990s, became a third to nearly a half of it.[15]

Note what the chart does not say. Tax policy is a multiplier, not the only cause — supply was throttled, rates fell, the population grew. But every honest account of why Australian housing detached from the rest of the developed world circles back to these two settings.[13]

Tax policy creates the desire to buy. The banks supply the means. House prices are set at the margin by what the next buyer can borrow — and for two decades the big four have been willing to lend more, against less, to more people.

The result is one of the most indebted household sectors on earth. Household debt climbed from 67% of income in 1990 to a peak near 187% in 2017–18 — among the highest in the advanced world.[18][19] The big four now carry roughly A$1.9 trillion of mortgages between them, about 77% of a A$2.475 trillion system book.[17][22]

How powerful is the lever? The Reserve Bank’s own housing model implies a sustained one-percentage-point fall in the real mortgage rate lifts real prices by around 28% over the long run.[16] That is the single most important number in Australian housing — and you can feel it for yourself.

This is why APRA, not the RBA, became the de-facto housing regulator. When prices ran too hot it capped investor-credit growth (2014), throttled interest-only loans (2017), and forced banks to test borrowers against a three-percentage-point buffer (2021).[17] Each time the brakes worked — for a while.

Bears are not wrong that this can end. They are wrong about how fast. Three other markets show the range of outcomes — and only one of them looks like the crash the shorts keep pricing in.

Japan is the nightmare. After a 1980s mania built on land collateral, real house prices peaked in 1991 and fell 47% over the next eighteen years.[23][24] They are still a third below that peak today, more than three decades on. The banks that had lent against the land spent a decade as zombies.

America is the fast version. Subprime lending inflated a bubble that peaked in 2006; real prices fell 37% by 2011 before clawing back — but it took until 2021 to reclaim the old high.[23][27] The crash that broke the global financial system still took five years to find a bottom.

China is the one happening now. Property and its tentacles are roughly a quarter of GDP;[25] after Beijing’s “three red lines,” Evergrande defaulted in December 2021 on more than US$300bn of debt,[26] and real prices have fallen about 20% since — and are still falling.[23]

And Australia? On the same inflation-adjusted basis, its worst stretch this cycle was a dip of about 6% in 2022–23, erased within two years.[23] Put the four on one axis and the contrast is the whole argument: three markets fall off a cliff from their peak; Australia bends, then makes new highs.

The May 2026 selloff — CBA’s worst day in 34 years, the big four shedding tens of billions in a fortnight — is the most serious crack in years.[6] Short interest sits at a record.[5] The bear case has never been more quantitatively true: debt is extreme, affordability is the second-worst in the world after Hong Kong,[21] and the policy props are, for the first time in 40 years, under genuine political threat.

But all of that has been true before. The machine is still running: the tax incentives are intact, the banks are still lending, and every previous dip turned out to be a buying opportunity. The honest answer is the one no short-seller wants to hear — the bet that Australian housing ends like Japan or America is probably right about the destination and hopeless about the timing. Which is precisely what makes it a widow-maker.

Watch the banks. They are the market’s purest expression of the housing thesis — and the place the reckoning, if it comes, will show up first.