Editor's note: On Friday morning Treasury released the exposure draft of the Innovative Business CGT Concession, and with it abandoned three of the four conditions it had attached to the startup carve-out in June. This edition tells the four-month story of a government discovering the shape of its own innovation economy by walking into its walls, names the wall still standing, and sets out how much further the demolition should go (the answer has been sitting in a Treasury review since 2010, which readers of our May editions will find unsurprising).

Innovation Spotlight: The Startup Carve-Out

Good evening Reader,

In the space of four months, the Australian Government has held four distinct public positions on how much tax a person who builds a company should pay. On Friday it arrived at its fourth, and best, and still unfinished.

The exposure draft legislation released by Treasury scraps the $10 million lifetime cap on the startup investor concession, extends company eligibility from ten years to fifteen across all sectors, and cuts the minimum holding period from five years to three. One condition survives: the $50 million turnover ceiling. Consultation runs until September 28. To understand why this is a retreat worth applauding and finishing, you have to watch the demolition in order.

The demolition, in order

In May, the Budget abolished the 50 per cent CGT discount from 1 July 2027, replacing it with cost-base indexation¹ and a 30 per cent minimum tax on real gains, and breaking an election commitment to do no such thing. We covered the anger and the arithmetic at the time, and made two arguments that have aged well: that the discount was functionally a property-and-timing subsidy whose defence on founder grounds asked the country to keep paying a $40 billion bill to preserve a by-product, and that the genuinely unprotected people were founders and early backers in the twenty-million-to-a-billion exit band. Treasury's own factsheet conceded the limit case and promised consultation.

In June, the consultation produced the carve-out's first draft: a concession for companies under ten years old, with a five-year minimum holding period, a $10 million lifetime cap per individual (a maximum lifetime benefit, on Piper Alderman's arithmetic, of $2.4 million for the people the policy was ostensibly designed to keep in the country), and a $50 million turnover ceiling. Four gates, each one a job for lawyers and a trap for founders. In August the broader abolition passed into law, and nearly 2,000 submissions arrived at the Senate and Treasury to explain what the gates would do.

On Friday, three of the four came down. Fifteen years, all sectors. Three-year hold. No lifetime cap. Every demolished condition is one the ecosystem said, in writing, in advance, should never have been built. The government's response to each backlash has been correct, which is to its credit, and reactive, which is the problem. Policy by successive concession means the final design is determined by whoever complains most effectively, rather than by anyone's model of how company formation actually works.

The wall they left standing

The $50 million turnover ceiling is the wrong wall, and it is wrong in a specific way. Turnover is a proxy for success, and the concession already contains an instrument for its actual target: an innovation test. A ceiling on revenue does not filter out rent-seekers; the innovation criteria do that. What the ceiling filters out is the exact cohort our May analysis identified as the unprotected segment, the companies crossing from promising to consequential. It draws a cliff at precisely the moment a startup proves the thesis its early investors risked capital on, and cliffs produce behaviour: structuring to stay under the line, exiting before crossing it, or incorporating somewhere without one.

And there is a second wall, quieter than the first. The innovation test that justifies removing the turnover ceiling does not yet exist. The government has promised a legislative instrument allowing businesses to self-assess whether they qualify, contents to be named later. Founders are being asked to plan five-to-fifteen-year equity positions against a definition of innovation that Canberra has not written down. Self-assessment against unpublished criteria is not certainty. It is homework with the marking scheme withheld.

How much further

The direction of Friday's retreat is right, so finish it. Four moves, none of them novel, all of them already argued in these pages or costed by Treasury.

Drop the turnover ceiling and let the innovation test carry the gate. One instrument, doing the job it was designed for, instead of a proxy punishing the outcome the policy exists to produce.

Publish the innovation instrument with the bill, not after it. A concession whose central definition arrives by delegated legislation at an unspecified date is not a concession, it is an option over one.

Build the rollover. Our May proposal stands: for the exit band above the small business concessions, a founder-cohort rollover into qualifying productive assets (re-investment vehicles, ESVCLP units, R&D-active companies) beats any gated exemption, because the gain is never crystallised, the cost-base problem dissolves, and the capital recycles into the next cohort instead of extracting. The exposure draft's own R&D Tax Incentive expansion shows the government can design for reinvestment when it chooses to.

And stop improvising. The reason this took four months and five documents is that the government has been rediscovering, gate by gate and backlash by backlash, conclusions that sit fully costed in the Henry Review's 138 recommendations, of which three have ever been implemented. We called Henry the tax switchboard in May. Friday's exposure draft is what it looks like to rewire a house by touching wires until one sparks. The switchboard is on the wall. Use it.

Consultation closes September 28. That is seventeen days in which the pen is, unusually, held out to the people who build things. The last round produced 2,000 submissions and demolished three walls. The remaining one is load-bearing for nothing.

¹ Indexation, for a founder whose cost base is one dollar, lifts it to roughly one dollar forty. The forty cents is the incentive.

P.S. Share this on LinkedIn/X to spark conversations with your colleagues.