There’s also been a focus on the effective taxation of unrealised capital gains on fund assets and how this is contrary to the fundamental tax rule of only taxing realised gains. However, for this article, I’ll flip this and focus on the effect of unrealised losses and how the mechanics of calculating superannuation earnings for Division 296 purposes fail the policy intent of reducing tax concessions for those with a TSB in excess of $3 million.
I highlighted this in the case study of Carlos that I outlined at the Melbourne leg of the SMSF Professionals Day (May 27, 2025). Carlos has a TSB of $9 million – three times the Division 296 threshold, yet under the proposed rules, he is not subject to any Division 296 tax. How is this possible? The answer lies in the design of the formula, and it reveals a key weakness in the policy framework.
The case of Carlos
Here’s what we know about Carlos:
| 1 July 2025 opening balance (30 June 2025 TSB) | $9,000,000 |
| Add: share of fund’s taxable income (net of tax) | $600,000 |
| Less: share of decrease in market value of fund assets (after provision for deferred tax) | $600,000 |
| Less: withdrawals | $150,000 |
| 30 June 2026 closing balance (TSB) | $8,850,000 |
Under the proposed Division 296 tax, ‘superannuation earnings’ are calculated based on the annual change in an individual’s TSB across all of their super funds, with adjustments made by adding back any withdrawals and subtracting any contributions during the year. This approach captures both realised and unrealised investment gains, meaning even paper profits are included, not just actual income received. Conversely, both realised and unrealised investment losses will also be included, reducing the amount of the earnings. The calculated earnings are then apportioned so that only the amount corresponding to the portion of the individual’s TSB above $3 million is taxed at 15 per cent.
Using this approach, Carlos’s superannuation earnings for Division 296 purposes is calculated as follows:
- Closing TSB at 30 June 2026 of $8.85 million
- Less prior year TSB at 30 June 2025, being $9 million
- Add withdrawals made during the 2025–26 year of $150,000
This results in Division 296 superannuation earnings of Nil.
Consequently, despite his balance being well over the $3 million mark, Carlos has no Division 296 tax liability in respect of the 2025–26 financial year. That’s because his ‘superannuation earnings’ as defined under the legislation are zero for the year, owing to an effective unrealised loss in his investment holdings. Therefore, there is no income base on which to apply the extra 15 per cent tax.
Does this align with policy intent?
The stated objective of Division 296 is to reduce the concessional tax treatment on superannuation earnings for high-balance members. Yet, Carlos enjoys the full 15 per cent concessional tax rate on his actual share of fund taxable income, that is, $600,000 -there has been no reduction to the superannuation tax concessions to the taxable net earnings applicable to that portion of his TSB in excess of $3 million. This is because the additional Division 296 tax, the tax to reduce his tax concessions, isn’t triggered unless there are positive superannuation earnings under the statutory definition.
This outcome raises several questions:
- Should someone with $9 million in superannuation continue to benefit from the full tax concessions, simply because of unrealised investment losses?
- Is the policy achieving its stated aim, or does the calculation method create loopholes?
- Could this incentivise strategies to manage the timing of asset revaluations or withdrawals to minimise Division 296 exposure?
These questions highlight the potential for unintended consequences and inconsistency with the underlying principle: to ensure that those with significantly higher superannuation wealth receive less tax concessions.
The mechanics undermining the message
The calculation of superannuation earnings under Division 296 uses a formula that:
- Includes both realised and unrealised gains and losses.
- Adds back withdrawals.
- Excludes contributions.
- Applies a proportion of the individual’s TSB that exceeds $3 million to determine the amount subject to tax.
If the earnings figure is zero or negative, even a very wealthy individual with a significant portion of their balance above $3 million may not have the tax concessions pared back on their share of fund taxable net earnings allocated to their superannuation interest(s).
In Carlos’ case, the $150,000 withdrawal is added back to his closing balance of $8.85 million, effectively returning him to his $9 million starting point. The resulting superannuation earnings calculation is nil. With no positive earnings to apportion, the Division 296 mechanism falls away, despite Carlos still enjoying the benefit of a substantial concessional tax rate on any fund taxable income allocated to his interest.
Implications and takeaways
The Carlos example spotlights a key inconsistency in the policy:
- Carlos enjoys the full benefit of tax concessions on a very large super balance
- Division 296 fails to impose tax, not because he isn’t wealthy, but because the earnings calculation is zero
This loophole could be exploited, intentionally or otherwise, by those able to manage valuations or investment timing, creating disparities between taxpayers with similar balances but differing short-term investment outcomes.
In contrast to cases like Jess, Why Division 296 Tax Is Not Just a 30 per cent Tax: Breaking Down the Math Misconception where unrealised gains dramatically inflate the Division 296 tax liability (resulting in effective tax rates exceeding 50 per cent when aligned to taxable income), Carlos is entirely exempt from the Division 296 tax for 2025–26, despite having a TSB well over $3 million and more than double Jess’.
A Call for Refinement
If the objective is to genuinely reduce the concessional treatment of high-balance superannuation accounts, then basing Division 296 on a broader and more consistent measure of income, or anchoring it to actual taxable income, might better serve that goal.
As it stands, the design of Division 296 creates inequity by taxing not just earnings, but paper gains, while allowing large balances to escape tax altogether when investment performance slips.
The Carlos conundrum and the case of Jess are not just extreme cases. They are both signals that the Division 296 framework needs a second look.
If a SMSF can manipulate valuations, there is a corresponding rouge SMSF Auditor in play. In other words, the gatekeeper will possibly catch this so it is just sensationalism to base this as the “unintended consequences” etc.
One of the problems with the formula is that losses are quarantined to this tax. So can only be used against future Div 296 gains. Just as a tax is levied in a year if there are taxable earnings, a loss should be refunded in a year incurred.
And further, this “transferable negative superannuation earnings” as it has been named, will be logged via another ATO ledger. So yet another data point that advice givers require in order to provide advice but, unless they are tax agents, don’t have access.
The list of errors in the drafting of this law is long. A classic example of the end being decided and the how built to achieve.
Thank you for your thoughtful comment Kym. I agree that in practice, it would be highly unlikely for a significant downward market valuation to go undetected, especially with SMSF auditors acting as gatekeepers. The point of my article was not to suggest that widespread manipulation is likely, but rather to illustrate how the Government’s proposed methodology for Division 296 can produce unintended consequences—even when all parties act appropriately. The “Carlos” example shows that, depending on timing and market movements, the approach to calculating “earnings” subject to the additional 15 per cent tax can result in outcomes that may not align with the measure’s intended policy objectives.
You’re absolutely right to point out that losses are quarantined, meaning they can only offset future Division 296 gains. As you noted, this differs from other areas of the tax system, where losses might be refunded or otherwise applied, and does create an additional complexity that advisers need to consider. The introduction of “transferable negative superannuation earnings” and its tracking via another ATO ledger will certainly add to the data points advisers must access—potentially increasing the compliance burden, especially for non-tax agent professionals.
As you suggest, there are several drafting issues and practical challenges that arise under the current rules. I hope the article contributes to the discussion on refining these measures before they are implemented (but I’m not holding my breath).
Mark
I have been saying this for a long time – if I can get a valuer to value my commercial property for double its market value and then keep on reducing it for the next 20 years with the income of the property – I will pay no Div 296 tax
Puting aside how retarded taxing unrealised capital gains is for one minute as on it’s own that discussion could go all week. This just highlights how poorly thought out Div 296 is. If you already have substantial assets in your SMSF you can calculate your likely paper profits and draw down enough bofore each financial year ends to offset the calculation.
Yet again this is shows how established older retirees can potentially avoid any additional tax obligations while younger ones under the preservation age in the future will not be afforded the same luctury.
The PM and treasurer will be long gone when this happens of course but the silly smirk they both give anytime anyone asks a question now might come back to haunt them both much sooner.
“If you already have substantial assets in your SMSF you can calculate your likely paper profits and draw down enough bofore each financial year ends to offset the calculation.”
That draw down is a withdrawal, consequently it’s added back to calculate your earn kings and you’ll still have to pay the tax.
Any taxpayer with losses in the current year or brought forward from a prior year will pay less or no tax, depending upon the amount of loss and the amount of current assessable income. This is not an issue specific to Div296 it is normal and correct application of taxation law.
So the author is saying its ok to pay tax on unrealised gains, but not ok if there is an unrealised loss? I don’t think so.