Division 296: Australia’s New Tax on Super Balances Above A$3 Million

The three-year fight over this measure was one of the most instructive tax debates in recent Australian history, because the original design was defensible in principle and unworkable in practice. Taxing unrealised gains on illiquid assets inside a superannuation fund creates liabilities on paper profits that a member may not be able to fund, and freezing thresholds in nominal terms guarantees that a measure aimed at the very wealthy eventually reaches ordinary savers. Both problems were fixed. This article explains what the final law does.

What problem was this meant to solve?

The uneven distribution of superannuation tax concessions. Earnings inside superannuation are taxed at 15%, well below most marginal income tax rates, and the value of that concession rises with the size of the balance and the member’s marginal rate. A person with A$20 million in superannuation receives a very large annual subsidy for savings they will never need for retirement income.

That outcome is a side effect of a system designed for a different purpose. Superannuation exists to reduce reliance on the age pension, and once a balance far exceeds any plausible retirement need, the concession is simply a tax shelter. Successive reviews identified this as the least defensible feature of the system.

Earlier measures had already tightened the edges — contribution caps, the transfer balance cap limiting how much can move into the tax-free retirement phase, and the Division 293 surcharge on contributions for high earners. Division 296 targets the accumulated balance itself, which is the part earlier reforms did not reach.

How does the calculation work?

Proportionally. The tax applies not to all of a member’s earnings but to the proportion of earnings attributable to the balance above each threshold. A member with A$4 million has one quarter of their balance above A$3 million, so roughly one quarter of their earnings falls into the additional-tax tier.

Total superannuation balance is measured across all interests — APRA-regulated funds, self-managed funds and defined benefit interests, including amounts in the retirement phase. Defined benefit interests are valued using prescribed methods rather than an account balance, and the legislation introduces a new “TSB value” concept applied to each interest.

Critically, only realised earnings are captured: interest, dividends, rent and realised capital gains. The original proposal to tax unrealised movements in asset values was removed from the tabled bill. That single change transforms the measure for self-managed funds holding property or unlisted business assets, which previously faced tax bills on paper gains with no cash to pay them.

How is the liability actually paid?

By the individual, from wherever they choose. The ATO calculates the liability using data reported by superannuation funds and issues a notice of assessment to the member. The member may pay it from personal funds or elect to have the amount released from their superannuation account.

That structure was deliberate. Making it a personal tax rather than a fund tax avoids placing the burden on trustees, avoids complicating fund accounting, and preserves the ability to use superannuation assets to meet the liability where the member prefers. Members with defined benefit interests may be eligible to defer the liability into a debt account rather than paying immediately.

The practical consequence for self-managed funds is a reporting one. Trustees must track and report investment activity in sufficient detail for the ATO to calculate realised earnings, which raises the compliance bar for funds that previously reported at a summary level. Anyone approaching the threshold should confirm their administrator can produce the required data.

What are the strategic responses?

Fewer than the commentary suggests, and most involve trade-offs. The obvious response is to hold less inside superannuation, either by ceasing voluntary contributions or by withdrawing amounts where a condition of release has been met and investing outside. That works arithmetically but sacrifices the concessional environment on the balance below the threshold.

Splitting balances between spouses is the most common legitimate strategy, since the threshold applies per individual rather than per household. Contribution splitting, spouse contributions and re-contribution strategies where eligibility permits can equalise balances over time, and doing so early is far more effective than doing it late.

Asset location matters more than it used to. Because only realised earnings are taxed, the timing of capital gains inside a large fund becomes a genuine planning variable. None of this is advice for any individual, and the interaction with contribution caps, the transfer balance cap and estate planning is complex enough that professional guidance is warranted for anyone near the threshold. Our overview of the superannuation system covers the surrounding rules.

Why did the original design fail politically?

Because it taxed unrealised gains, and that single feature made an otherwise defensible measure indefensible in practice. A self-managed fund holding a farm, a commercial property or shares in a private business could be assessed on a paper increase in value with no corresponding cash to pay the bill, forcing an asset sale to fund a tax on a gain that had not been received.

The absence of indexation compounded it. A threshold fixed in nominal terms captures a steadily larger share of the population over time, so a measure sold as affecting a very small number of very wealthy people would eventually reach ordinary savers — a point opponents made repeatedly and accurately.

Both features were removed before the legislation passed, which is a reasonable outcome for a democratic process and an expensive one in delay. The lesson for tax design is that a measure creating liabilities disconnected from cash flow will attract opposition disproportionate to the number of people it affects, because the unfairness is easy to describe and hard to defend.

What else is changing in superannuation from 2026?

Payday superannuation is the largest operational change. From 1 July 2026 employers must pay superannuation contributions at the same time as wages rather than quarterly, which improves compounding for employees and removes a long-standing working capital float for employers. It also makes unpaid superannuation far easier for the ATO to detect.

The low income superannuation tax offset is being enhanced, raising thresholds and payment limits so that lower-income earners are not effectively taxed more heavily on their superannuation contributions than on their wages. Superannuation is also now payable on government-funded parental leave pay, addressing a long-criticised gap that disproportionately affected women’s retirement balances.

Taken together, the package moves in two directions at once: less generous at the very top through Division 296, more generous at the bottom through the offset and parental leave changes. That is a deliberate rebalancing of who receives the system’s concessions, and it is the clearest statement of policy intent since the guarantee rate was legislated to 12%.

How should SMSF trustees prepare?

By confirming their data will support the calculation. Because only realised earnings are captured, the ATO requires sufficient detail from funds to distinguish realised gains, dividends, interest and rent from unrealised movements in asset values. Funds reporting at a summary level will need to upgrade their record keeping before the first assessment period.

Asset valuation methodology also matters more than it did. Total superannuation balance now uses a defined TSB value for each interest, and for funds holding property, unlisted business assets or collectables, the valuation approach determines whether the threshold is crossed at all. Documented, defensible valuations obtained annually are the practical requirement.

Finally, liquidity planning becomes a live issue. A fund holding a single illiquid asset with modest cash reserves may face an assessment it cannot comfortably meet from within the fund, and the option to pay personally exists only if personal funds are available. Modelling the cash position under a realistic earnings scenario is worth doing well before June 2027, and specific advice is warranted given how much turns on individual circumstances.

How does this fit the broader tax debate?

It is the first substantial narrowing of superannuation tax concessions in years, and it establishes a principle rather than settling a question. The principle is that concessions should be limited by reference to the size of the balance rather than only by reference to contributions, which is a different and more durable design than caps on what goes in.

That principle has obvious extensions, which is why the debate will continue. The thresholds could be lowered, the rates raised, or the same logic applied to other concessional structures. Nothing in the legislation commits a future government to anything, and the political difficulty of the original passage suggests the next step would be equally contested.

For planning purposes, the sensible assumption is that the direction of travel is one way. Superannuation concessions represent a very large annual cost to the budget, the ageing population increases pressure on that budget, and measures targeting a small number of large balances are the least politically costly available. Anyone building a long-horizon strategy around current settings should stress test it against tighter ones.

Frequently Asked Questions