The $3 million super line rises less often than the word indexed suggests

The three million dollar line moves in jumps of a hundred and fifty thousand. That is five per cent. At inflation inside the target band, five per cent takes roughly two years to accumulate — and until it does, the line does not move at all while balances keep compounding towards it.

That line is the threshold for Division 296, which became law on 13 March 2026 and applies from this financial year. If everything you hold in super adds up to more than three million dollars on 30 June, an extra 15 per cent tax applies — not to the whole balance, only to the share of your investment earnings relating to the amount above the line. Above ten million a further amount applies. It is a personal liability, not the fund’s, it is not deductible, and it can be paid from outside super or by releasing money from it. First assessments land in the year beginning July 2027.

Most of the original objections have been answered. This is a marginal structure, not a cliff. It taxes a proportion of earnings rather than the whole balance. The worst feature of the first design is gone.

The word carrying the weight

Both lines are indexed. Indexing means a number is written to rise automatically with the cost of living, so that a threshold set today does not quietly catch ordinary people in fifteen years, so that a threshold set today does not quietly catch ordinary savers in fifteen years. That single fact did most of the work in defending the measure against the charge that it would quietly capture ordinary balances over time, the way the old superannuation surcharge did.

Look at how that indexing actually works. The large balance threshold moves in increments of $150,000. The very large threshold moves in increments of $500,000.

A threshold that only moves in $150,000 steps is a threshold that spends most of its time not moving.

The arithmetic is not subtle. $150,000 against a three million dollar threshold is five per cent. At inflation inside the target band, cumulative price growth takes roughly two years to reach five per cent — and the threshold does nothing at all until it gets there. In the intervening period the nominal line is fixed while balances compound.

Why this matters more than it sounds

The cohort sitting just below three million is not small, and it is growing toward the line from beneath. Every year the threshold stays nominally still, some of that group crosses it through nothing but market returns and contributions they were always going to make.

That is a defensible policy outcome if it is the intended one. It is much less defensible if the measure was sold on being indexed and people heard that the line keeps pace, when what it actually does is catch up late, roughly every second year.

The design is also asymmetric in an interesting way. From 2026–27 onward, the balance used to test the threshold is the greater of the figure immediately before the financial year and the figure at the end of it. That is sensible anti-avoidance — it stops a member ducking under the line with a well-timed June withdrawal. But note the direction of travel: the test reaches for the higher number, while the threshold it is tested against reaches for the lower one until an increment triggers.

The case against this argument

There is a serious defence, and it has been made well. Division 296 is a proportionate response to concessions flowing to balances far beyond any plausible retirement purpose, and much of the campaign against it has misrepresented what the policy actually does. On that reading, quibbling about increment size is another instance of exactly that: technical objection standing in for opposition to the principle.

The practical case for moving in steps is also real. Raising the line smoothly would shift it by a few thousand dollars a year, which every fund and every adviser would have to track for a cohort where the difference is trivial against the balances involved. Other superannuation limits already move in steps rather than smoothly. This is not an innovation.

And the effect may simply be small. If inflation runs near the middle of the band, steps arrive often enough that the frozen periods never bite hard.

What would settle it

Two things, both observable. How often the large balance threshold actually steps between now and 2030, and what happens to the count of members assessed under Division 296 across those years. If the threshold moves at least every second year and the assessed population stays roughly flat as a share of members, the increment size never mattered. If the threshold sits still for three years while the assessed count climbs, the word “indexed” was doing work the mechanism could not support.