Submission 2079 — Mr Lee Knights (2079

The Impact Analysis and the size of the cuts

Note tabled by Lee Knights — witness in a private capacity, Community

Affairs Legislation Committee inquiry into the National Disability Insurance

Scheme Amendment (Securing the NDIS for Future Generations) Bill 2026. Perth, 6 August 2026. Submissions 2079 and 2079.1. Corrected version, 7 August 2026 — corrections listed at the end.

This note supports both of the asks in my statement: that the Bill should not pass in its current form, and that the Government publish a cost-benefit analysis before it proceeds. Everything in it comes from published Government documents — the Impact Analysis for this Bill (01.07.26 update), the Department’s answers to questions on notice IQ26-000082, and

the Scheme Actuary’s Annual Financial Sustainability Report 2024-25 —

quoted directly, so every point can be checked at source. I table it because a written note reaches the Committee whether or not there is time for questions.

Summary

1. The constraint is a growth rate. Nothing published turns that rate into these percentages.

The Impact Analysis says the package is required to meet National Cabinet’s target of scheme growth “between 5 to 6 per cent, or lower”. It states no required savings figure. A growth target establishes that growth must slow; it does not establish that social and community participation budgets must fall by half. The step from one to the other is not on the public record.

2. On the only published breakdown, participant growth is the smallest driver — and the projections the package relies on have not been broken down at all.

The Scheme Actuary’s most recent published projections attribute 1.4 to 2.0 per cent of annual growth to participant numbers, and 3.0 to 3.7 per cent to price indexation. Schedule 1 is directed at the smallest of the three components. Those figures are a June 2025 vintage; the Impact Analysis relies instead on later projections prepared for the 2026-27 Budget, which I have not found published with any comparable breakdown.

3. Inside the document, the size of the categories is evidenced and the size of the cuts is not.

The Impact Analysis carries 185 footnotes. The chapter disposing of the options the Department considered and rejected contains none — no footnotes, no data, no figures, no tables. Neither the 50 per cent nor the 10

per cent figure has a published derivation. Several footnotes that do appear cite NDIA internal analysis marked unpublished.

4. Utilisation is an unfair basis for a cut in one Government document and a reassurance about the same cut in another.

The Impact Analysis records that the Department considered reducing budgets in line with previous utilisation and rejected it, because low utilisation can mean services are not available rather than not needed. Its answer to question on notice IQ26-000082 then defends the 50 per cent reduction by pointing to average utilisation of 80 and 86 per cent. Both cannot hold.

5. The harms the Department calls disqualifying are concentrated by the option it chose.

A blanket 10 per cent reduction was rejected because it could cause regression in daily living skills, injury, neglect, social isolation, and lost employment and community engagement. Those harms follow from the kind of support being cut, not from how widely the cut is spread — and the measure adopted removes five times as much of that support from a narrower group.

6. The Department has named the risk this Bill runs, and then run it. Rejecting a more substantial eligibility change, the Impact Analysis says it would risk people with significant need being left without supports critical to their wellbeing and safety — “in the absence of adequate alternative supports outside the NDIS”. The Department therefore accepts that the scheme’s intent can be failed by tightening as well as by growth, and it names the danger precisely: people losing access before the alternatives exist. That is my family’s position, and it is why I asked to appear. Questions the Committee could put

7. What analysis takes the Government from an agreed growth rate to a 50 per cent and a 10 per cent reduction?

8. What is the component breakdown of the 2026-27 Budget projections the package relies on — what share of forecast growth is participant numbers, and what share is prices?

9. On what date were those projections finalised, and on what date were the 50 per cent and 10 per cent figures settled?

10. Will the Department produce the two pieces of NDIA internal analysis cited as unpublished, and any document showing how the two percentages were derived?

11. Which of its two positions on utilisation does the Government stand by?

6. Why do harms that disqualified a 10 per cent reduction across all categories not disqualify a 50 per cent reduction concentrated in the category where those harms arise?

7. What evidence is there that adequate alternative supports will exist for the nine to fifteen age group before the access changes commence on 1 January 2028? The request

Publish the derivation of the 50 per cent and 10 per cent figures. If an analysis showing how they were arrived at exists, the Committee and the people affected should be able to see it before the Senate votes. If none exists, then two figures that will remove supports from hundreds of thousands of people were not derived from any published evidence, and the Senate is being asked to legislate them anyway. That is not a reason to amend the Bill at the edges. It is a reason not to proceed with it.

Detail

1. A rate target, and an unevidenced step to two percentages In 2023 National Cabinet committed to an annual NDIS growth target of 8 per cent by 1 July 2026. On 30 January 2026 it agreed to an updated target of 5 to 6 per cent, or lower. The Impact Analysis frames the package throughout as what is required to meet that target, and notes costs are forecast to grow well above it.

There is also a question of sequence that the published documents do not answer. The Impact Analysis rests its case on “the latest projections for the 2026-27 Budget”, and states that on those projections the 8 per cent target would not be met when it took effect on 1 July 2026. The Budget was delivered on 12 May 2026. The measures those projections are said to justify were announced on 22 April 2026, three weeks earlier, and the Bill was introduced on 14 May. National Cabinet had tightened the target to 5 to 6 per cent on 30 January 2026.

I do not assert that the projections post-date the decision. Budget projections are prepared over months and may well have existed internally well before April. That is exactly why it is a question and not a claim — and it is a question only the Department can answer.

My objection is not to the Government managing the cost of the scheme. It is that the operative constraint here is a rate, agreed in a forum that publishes communiqués rather than reasoning, and the tightening from 8 per cent to 5 or 6 per cent is stated as a decision with no published analysis deriving it. The Senate is being asked to legislate to a figure settled elsewhere, on reasoning it has not been shown. I would add, since the scheme’s worth is what is really in question underneath the rate: the three cost-benefit analyses commissioned of this scheme do not, between them, support the proposition that it delivers a negative return.

The most recent of them is the one that most needs stating accurately, because it is the one most easily misread in the Government’s favour. Taylor Fry’s 2023 analysis for the Independent NDIS Review measured benefits of $7.5 billion against net costs of $16 billion in 2022-23, so at the all-of scheme level its measured costs exceeded its measured benefits. But it reached that result on a deliberately constrained method. It expressly excluded justice and child protection outcomes, post-school education outcomes, community benefits, fiscal multiplier effects, the value of the scheme to the disability support sector, and economic utility and redistribution — and its authors described their own benefit figure as “most likely an underestimate”. The all-of-scheme result is therefore a product of what the method left out, and the authors said so.

At the cohort level, on that same constrained method, the analysis found net benefits exceeding net costs for children and families, and for the autism cohort with the lowest assessed support needs. That is the cohort this Bill’s access changes are aimed at, and it is the cohort the “low to moderate support needs” descriptor is built around. The Government’s own most recent commissioned analysis found the scheme paid for itself precisely where the scheme is now being cut — and it found that while understating the benefits, on its own account.

No cost-benefit analysis has been commissioned since 2023. And the most recent of the three did not attempt a full account of the benefits — it says so itself. So the reduction now proposed is being made without any current assessment in which the full range of costs and the full range of benefits are weighed in one frame.

There is also a mismatch between the instrument and the target. A growth rate target is met by measures that change the rate of growth. A one-off 50 per cent reduction to social and community participation budgets, and a 10 per cent reduction to capacity building daily activities, are level shifts. The Department itself works with this distinction elsewhere in the same chapter, rejecting more substantial eligibility change partly because it “would only have a short-term impact on moderating the growth of Scheme costs without changes to participant plan budgets.”

The question I would ask, if I were in a position to: the Impact Analysis says the constraint is a 5 to 6 per cent growth target and states no savings requirement, so what determines that the community participation cut is 50 per cent rather than 20, or 10, or nothing — and where is that analysis?

2. Where the growth actually comes from

The Scheme Actuary’s Annual Financial Sustainability Report 2024-25

projects annual growth in scheme expenses and breaks it into three components. Its Table 4:

Growth component        2025-26    2026-27    2027-28    2028-29    2034-35

Participant impacts         2.0%       1.7%       1.4%       1.4%       1.6%

Pricing impacts            3.0%       3.7%       3.5%       3.5%       3.5%

Real growth in payments     4.4%       1.0%       2.8%       2.6%       2.1%

Total growth             9.4%       6.4%       7.7%       7.5%       7.2%

Growth in the number of participants is the smallest of the three components, at 1.4 to 2.0 per cent a year. Price indexation runs at roughly double that. Schedule 1 is directed at who gets access and at what participants may spend.

This is the most recent decomposition the Scheme Actuary has published, and I want the Committee to have its limits along with its content. It is a June 2025 vintage, prepared before this package, and the report states that it assumes growth above the 5 to 6 per cent target with further reform needed. It is not evidence that growth reaches the tightened target on its own. Price indexation is also not a free alternative: reducing it is a provider price cut, and provider price cuts reach participants through thinner markets, which is the position my family is already in.

The Impact Analysis does not rely on these projections. It relies on later ones prepared for the 2026-27 Budget, which it says revised Scheme expenditure upward against the mid-year estimates, and which establish on its account that the 8 per cent target would not be met. Those later projections are what the case for this package rests on — and I have not been able to find them published with a breakdown between participant growth and price growth.

That is the gap. The public argument for this Bill is made in terms of participant numbers and participant budgets. The only published decomposition attributes most of the growth to prices. If the Budget projections change that attribution, the Committee should be shown how and by how much. If they do not, then the Bill is directed at the smallest driver of the growth it is meant to control.

3. What is evidenced, and what is not For capacity building daily activities the Impact Analysis builds its case carefully and cites as it goes: 752,454 participants hold this funding; the average annualised budget is $13,460; that is described as around 69 hours of therapy at the NDIS hourly rate of $193.99; total annualised budgets grew 11.5 per cent in the year to December 2025; a chart tracks the total from $6.9 billion to $10.1 billion across four years. Then, with no bridging analysis, the document states that budgets will be reduced by 10 per cent.

The 50 per cent figure is treated the same way. The only passage explaining how it was reached describes an approach that was not taken.

Footnotes in this part of the document cite the Department’s own unpublished work. Footnote 106 cites “NDIA (2026), Explore data; NDIA (2026), Internal analysis of distribution of participant budgets by select support categories at 31 December 2025, unpublished”, and footnotes 102 and 103 cite the same unpublished analysis. Footnote 129 cites internal analysis of the same kind, without that marking. The Department has named internal analysis, with dates, in a public document, and marked it unpublished.

There is a great deal of evidence in this document. All of it is evidence about how large these support categories are and how fast they have

grown. None of it is evidence for the size of the response, and the passage disposing of the alternatives carries no citation at all. Those are different questions. Establishing the first does not establish the second, and only the second is in issue here.

One further claim in the same section carries no source: “in most allied health disciplines there is little research evidence to support high volumes of therapy on a long-term basis.” That is a contested clinical proposition, asserted without a reference, and used to support reducing allied health. The comparisons offered alongside it — veterans’ care allied health allocated in cycles, and Medicare’s chronic disease management plan at up to five subsidised sessions a year — are placed beside the 69-hour figure to imply excess, with no analysis of whether the populations or the purposes are comparable. A chronic disease management plan and a lifelong disability support package are not the same instrument, and the document does not argue that they are.

4. The utilisation contradiction, in full Explaining how it settled on the reduction, the Impact Analysis records:

“In determining the extent of SCCP reductions, the department did consider whether reducing individuals’ budgets in line with previous utilisation may be an effective way to constrain growth, however this was ultimately not progressed due to the impacts on some NDIS participants where thin market issues are prevalent. For example, where a participant’s utilisation is low due to services not being available this may unfairly impact those participants more than participants with higher utilisation in areas where markets are more mature.”

That is the Department stating, in its own words, that low utilisation can mean the services are not there rather than that the support is not needed.

The Department’s answer to question on notice IQ26-000082 (Q4) then defends the 50 per cent reduction on precisely the basis it rejected, reassuring the Committee that a halved budget will not halve support because most participants do not use their full budget — citing average utilisation of 80 per cent for participants receiving supported independent living and 86 per cent for others.

Both cannot be true. And the figures cited sit well above the new ceiling: a participant using 80 per cent of a budget that is then halved loses roughly a third of the support they actually use. This is also a blanket change, not an individual one. The Department has confirmed it is not a reviewable decision, and the Bill as it passed the House does not add these determinations to the list of reviewable decisions in section 99 — the items added there are decisions of the Chief Executive Officer, while a support

determination is a legislative instrument made by the Minister. A participant who does use and need the support has no way to say so.

5. The harms, and the option that concentrates them The rejected option was “a blanket 10 per cent reduction to every support category all participants can receive except SIL and those requiring 24-hour and intensive care.” The Department’s stated reason:

“More significant reductions in the volume of certain supports could result in participants experiencing regression in daily living skills, elevate the risks of injury, neglect and social isolation and ability to engage with employment and community activities. This would undermine the objectives of the NDIS, including the aim to provide reasonable and necessary supports to participants.”

The two options are not like for like. The rejected one was broader, across every category, and part of the Department’s concern was that breadth — including its effect on provider viability and on thin markets. But the participant harms it names are not breadth-specific. Regression, isolation, and lost employment and community engagement are harms of removing community participation support, and the adopted measure removes five times as much of it from a narrower group.

The Department also states the design intent plainly: the reduction would “encourage more uptake of group based services at more efficient prices… A shift away from individualised supports.”

That is a significant change to how the scheme works, and it is stated as a consequence rather than argued for. It also assumes the group-based services are there to be taken up. In the same chapter the Department accepts they are not, everywhere — that is the whole basis on which it rejected the utilisation approach. A participant in a thin market is not choosing individualised support over group support at a more efficient price; there is nothing else to choose. Describing that as encouraging uptake presents an absence of options as a preference.

6. A sequencing concession worth the Committee’s attention Rejecting more substantial eligibility change, the Impact Analysis says it “would not be consistent with the original intent of the Scheme” and “would also create too great a risk that people with significant need would be left without access to supports critical to their wellbeing and safety (in the absence of adequate alternative supports outside the NDIS).”

I draw two things from that. The first is that the Department accepts the scheme’s intent can be failed by tightening as well as by growth — that the test runs in both directions, and that there is a point past which narrowing access is itself inconsistent with what the scheme is for. The second is in the

parenthesis, which concedes the risk precisely: the danger is people losing access before adequate alternatives exist outside the scheme. That is my family’s situation and the reason I asked to appear.

Closing

Three cost-benefit analyses of this scheme have been commissioned in fifteen years, and none since 2023. A cost projection is not a cost-benefit analysis. Without one, nobody can say what these reductions cost once the consequences land on families, on state services, on health, justice and schools, or simply go unmet. The same gap runs on the other side of the ledger. The Department’s stated reason for rejecting a more substantial eligibility change was the risk of people losing access before adequate alternatives exist outside the scheme. Funding has been committed to those alternatives. I have not found anything published showing what that funding covers, how it was costed, or whether it is sized to the need it is meant to meet — and for the nine to fifteen age group there is no named program for it to fund.

That is why both of the asks in my statement are asks about evidence: publish the derivation of these figures, and publish a cost-benefit analysis of the scheme as it stands and as proposed, before the Senate votes. Corrections to the version tabled on 6 August 2026

On checking the tabled note against the source I found two errors, and I have corrected them here rather than leave them on the record. Neither affects the argument.

1. The tabled version said the Impact Analysis carries roughly 584 footnotes. It carries 185. I have corrected the figure. The point the sentence makes is a contrast between a heavily footnoted document and an options chapter carrying no citation at all, and that contrast is unaffected: on a direct count, sections 3.1.3 and 3.1.4 contain no footnote markers, tables or figures at all, while section 3.1.2, which sets out the option the Department adopted, carries a dozen or more.

2. The tabled version attributed a single quotation ending in the word “unpublished” to both footnote 106 and footnote 129. Footnote 106 carries that word; footnote 129 does not. I have set out what each footnote says. Footnotes 102 and 103 also cite the same unpublished internal analysis, so the substance of the point is if anything wider than the tabled version stated.

This document was prepared with the assistance of an AI writing tool (Anthropic’s Claude). The substantive arguments and factual claims are mine; the AI assisted with structuring, language polish and source-tracking. All citations have been independently verified against the originals.