Ensuring Reform is Ready, Accountable and Safe (DRO advocacy)

Submission 429 - Supplementary Submission

Australian Autism Alliance

Supplementary Submission

to the Senate Community Affairs Legislation

Committee Inquiry into the National Disability

Insurance Scheme Amendment (Securing the NDIS for

Future Generations) Bill 2026

Systems That Work:

Ensuring Reform is Ready, Accountable and Safe

Submitted by: Australian Autism Alliance

Contact: Jenny Karavolos, Chair, Australian Autism Alliance

www.australianautismalliance.org.au Date: 10/07/2026

Submission 429 - Supplementary Submission

Acknowledgements

We acknowledge the First Nations and Traditional Owners of the land, sea and waterways and pay respects to Elders past, and present and recognise those whose ongoing effort to protect and promote Aboriginal and Torres Strait Islander cultures will leave a lasting legacy for future Elders and leaders. We recognise the enduring connection that First Nations peoples have to land, waters, culture, and community. This land was, is, and always will be Aboriginal land.

We acknowledge the Individual and collective expertise of those with a living or lived experience of disability, as well as the lived experience of people who have been carers. We recognise their vital contribution at all levels and value the courage of those who share their unique perspective for the purpose of learning and growing together to achieve better outcomes for all.

Submission 429 - Supplementary Submission

Acknowledgements ………………………………………………………………………………………………………………………. 2 About the Australian Autism Alliance …………………………………………………………………………………………….. 4 1 Executive Summary ………………………………………………………………………………………………………. 5 2 Summary of Supplementary Recommendations ………………………………………………………………… 6 3 NDIS Amendments Bill 2026 - Testing the Bill’s legislative assumptions against established evidence ……………………………………………………………………………………………………………………………. 8 4 From evidence gaps to safer reform governance …………………………………………………………….. 13 5 NDIS Reform Dependence and Sequencing Problem ………………………………………………………. 16 6 The Missing Ledger: A Saving to the NDIS is Not Necessarily a Saving to Australia …………….. 21 7 Identifying Administrative Efficiency and Sustainable Savings …………………………………………… 25 8 Evidence of “Design First” Proposition: Preventing the Cost of Reform Failure …………………….. 29 9 Conclusion …………………………………………………………………………………………………………………. 32 Attachment A: Systems That Work Framework ……………………………………………………………………… 34 Attachment B – NDIS Amendment Bill 2026: legislation to evidence matrix ………………………………. 35 Attachment C – The Pre-Governance Assessment: before Parliament authorises reform ……………. 59 Attachment D – The Safeguards Index: protection before harm……………………………………………….. 60 Attachment E – Illustrative Administrative Efficiency Opportunities …………………………………………… 61

Submission 429 - Supplementary Submission

About the Australian Autism Alliance

www.australianautismalliance.org.au

The Australian Autism Alliance (the Alliance) is a funded national peak body Disability Representative Organisation (DRO) providing a Strong Unifying Voice for Autism, and authoritative policy expertise on autism working with Australian government at all levels to strengthen policy and service systems affecting Autistic Australians. Established in 2016, the Alliance works to improve the life chances of Autistic people and strengthen collaboration across the Australian autism community.

The Alliance contributes futures-focused adviser expertise to a range of national advisory and reform processes, including DHDA Disability Representative Organisation program, the NDIA

Autism Advisory Group, the NDIA DRCO Co-Design Advisory and Reform groups, NDIS

Commission Disability Sector Consultative group, the National Autism Strategy Oversight

Council, and National Health and Mental Health Roadmap for Autistic people. The Alliance is focused on system design solutions that enable disability systems to respond effectively to the diversity of Autistic support needs - “Change the System, Not the Person”.

Our membership represents a cohesive national network of key organisations with a diverse focus on autism – that is led by Autistic people, advocacy groups, peak bodies, service providers, and researchers. Together, this network provides a platform for the diversity of Autistic perspectives and lived experience across Australia. Through our members and communication channels, the Alliance reaches more than half a million people and supports Autistic people and their families across the lifespan. Most importantly, our work is informed by Autistic people and the Australian autism community.

Submission 429 - Supplementary Submission

1 Executive Summary

The Australian Autism Alliance (the Alliance) welcomes the opportunity to provide this supplementary submission on the National Disability Insurance Scheme Amendment (Securing the NDIS for Future Generations) Bill 2026.

This submission builds on, rather than repeats, the Alliance’s original submission. It responds to matters raised during the Senate Community Affairs Legislation Committee hearing on 10 June 2026, where the Alliance appeared as a witness, and presents additional evidence and analysis in six key areas:

a) Testing the Bill’s legislative assumptions against established evidence; b) The evidence and assurance Parliament should require before authorising reforms where legislative assumptions are unsupported, potentially inconsistent with, or contrary to established evidence, and where foreseeable consequences may be serious or irreversible (including the Pre-Governance Assessment and Safeguards Index);

c) The NDIS reform dependency and sequencing problem; d) The Missing Ledger – a saving to the NDIS is not necessarily a saving to Australia;

e) Identifying administrative efficiency and sustainable savings (“red tape reductions”); and f) Evidence of “Design First” proposition – preventing the cost of reform failure. The Australian Autism Alliance supports reform of the NDIS and recognises the importance of ensuring the Scheme remains sustainable for future generations.

However, Parliament’s responsibility extends beyond determining whether reform is desirable. It must also determine whether it has been provided with sufficient assurance that the proposed reforms are appropriately designed, supported by evidence, legally coherent, properly sequenced, safeguarded, capable of being implemented safely and delivering better outcomes for people with disability.

To assist that assessment, this submission applies the Alliance’s Systems That Work Framework—a practical governance methodology developed to help governments and parliaments assess whether major system reforms are sufficiently designed, evidenced, legally coherent, safeguarded, accountable and ready for implementation before they are authorised. The Framework is included at Appendix A for reference.

The Framework is not intended to determine whether reform should occur. Rather, it assists Parliament to determine whether sufficient assurance has been provided to responsibly authorise reform.

Applying the Framework, the Alliance concludes that significant gaps remain in relation to:

evidence supporting key legislative assumptions;
reform readiness and implementation sequencing;
accountability and governance arrangements;
safeguards for people with disability;
measurable human outcomes; and
whole-of-government economic and socioeconomic assessment. Accordingly, the Alliance respectfully submits that Parliament should require greater assurance in these areas before authorising reforms of this scale.

Submission 429 - Supplementary Submission

2 Summary of Supplementary Recommendations

Rec  Recommendation Summary                                     Section in

No   Topic                                                     Submission

Before authorising the highest-risk reforms, Parliament Section 3 should require Government to:

a. demonstrate the evidence supporting the contested legislative assumptions;

b. publish relevant validation and readiness Parliament evidence;

provided with             c.  undertake a whole-of-reform cumulative impact

RA        sufficient                  assessment;

assurance                d.  identify foreseeable harms and legislated

safeguards; and

e. establish measurable human outcomes and clear accountability arrangements before the highest-risk provisions commence.

Parliament should require application of the Pre- Section 4 Major reform Governance Assessment before authorising major

RB        authorisation         structural reforms to the NDIS.

Protection before    Parliament should require application of the Safeguards   Section 4

Harm Index before authorising major structural reforms where RC foreseeable harm may be serious or irreversible.

Parliament should require sufficient assurance that Section 5 critical dependencies are demonstrably ready before reforms that restrict access, reduce supports or Dependency and transition participants to alternative systems commence.

RD       Sequencing Issue   People with disability should not be required to

determine through lived experience whether replacement systems are ready in practice.

Before authorising major structural NDIS reforms, Section 6 Parliament should require publication of an independent

Whole-of-Government Economic and Socioeconomic

Impact Assessment that identifies whether projected

RE   Missing Ledger      reductions in NDIS expenditure represent genuine net

savings to Australia or the transfer of costs to other Commonwealth programs, State and Territory systems, families, unpaid carers and the broader economy.

Before participant supports are reduced, Parliament Section 7 should require sufficient assurance that

Red Tape Savings   opportunities to reduce administrative costs, improve RF      and Efficiency                             decision-making and eliminate avoidable bureaucracy

have been comprehensively identified and exhausted.

Submission 429 - Supplementary Submission

Rec  Recommendation Summary                                     Section in

No   Topic                                                     Submission

Parliament should require a Design First Assurance Section 8 Standard for major NDIS reforms. Implementation should not commence until Parliament is satisfied that:

a. the reforms are supported by sufficient evidence to achieve their intended outcomes;

b. the legislation is legally coherent and consistent with Australia’s human rights obligations; and

Design-first

RG   assurance             c. appropriate governance, safeguards and

requirement accountability arrangements are in place for implementation.

This assurance should be proportionate to the potential consequences of error. The greater the potential severity and irreversibility of harm, the higher the threshold of evidence, design and assurance required before implementation

Table 1: Supplementary Recommendation Summary Table

Submission 429 - Supplementary Submission

3 NDIS Amendments Bill 2026 - Testing the Bill’s legislative assumptions against established evidence

In its initial submission, the Australian Autism Alliance identified a number of provisions where the proposed legislative settings appeared inconsistent with the lived realities of Autistic people, the heterogeneity of autism and established evidence-informed practice.

Following its appearance before the Committee, the Alliance undertook a more detailed review to test those concerns. Rather than starting from a position for or against individual provisions, we identified the underlying assumption each contested provision appears to rely upon and tested those assumptions against published research, authoritative reviews, Australian and international evidence, and established evidence-informed practice.

The resulting NDIS Amendment Bill 2026: Legislation to Evidence Matrix is provided at Attachment B.

The Matrix examines 11 legislative and cross-cutting issues. For each, it identifies:

the legislative assumption;
the relevant evidence;
the strength and direction of that evidence; and
the potential consequences if the assumption is wrong. Risk was assessed across three dimensions: likelihood, severity and reversibility.

The findings should be of significant concern to Parliament.

Across the 11 issues examined, the Matrix found:

seven areas where the legislative assumption was contrary to established evidence, either alone or in combination with an identified evidence gap;
three areas potentially inconsistent with established evidence or evidence-informed practice; and
significant evidence gaps regarding critical implementation assumptions, including the availability of autism-appropriate functional assessment methodology, the readiness of alternative support systems and the cumulative impact of simultaneous reforms.

Six of the 11 issues were assessed as presenting Extreme risk if the legislative assumption is wrong, three as Very High risk and two as High risk.

The most serious concerns relate to assumptions that:

families can absorb additional unpaid care indefinitely;
that autism-related impairment may resolve through treatment before disability support is required;
that a valid and autism-appropriate functional assessment methodology exists or will exist;

Submission 429 - Supplementary Submission

that population-level funding rules can fairly allocate supports across a population defined by heterogeneity;
that multiple reforms can be assessed independently; and
that harm arising from incorrect assumptions can be identified and corrected after implementation.

The matrix’s summary Table 2 sets out these findings together.

The evidence reviewed points in a materially different direction, demonstrating that:

a) families of Autistic people already carry significant and measurable caring, employment, financial and family impacts;

b) support can sustain rather than displace informal care; c) autism is a lifelong neurodevelopmental condition; d) early developmental opportunities can be time-sensitive; e) functioning is context-dependent and can be obscured by masking and fluctuating presentation;

f) autism is characterised by substantial heterogeneity; and g) support needs can change cumulatively rather than through a single unexpected event.

The evidence also reinforces the importance of:

individualised decision-making,
procedural fairness,
human oversight,
supported decision-making, and
effective review mechanisms where decisions affect access to essential disability supports.

The issue is not Government’s policy intent.

The Alliance supports NDIS reform and long-term sustainability.

However, legislation converts policy assumptions into legal decision-making rules. If those assumptions are wrong, the consequences are not theoretical—they directly affect people’s lives. It can determine:

whether a child receives support during a critical developmental period;
whether a family is legally presumed capable of absorbing additional care;
whether an Autistic person’s functional capacity is accurately understood;
whether a participant can seek reassessment as their circumstances progressively deteriorate; or
whether an individual can challenge a class-based reduction in funding. The Matrix identifies several areas where the consequences of error may be serious and, in some cases, irreversible. In particular, the evidence relating to developmental windows, cumulative deterioration, family sustainability and delayed support challenges an implementation model that assumes harm can be detected and corrected after commencement.

Submission 429 - Supplementary Submission

This raises a broader governance question for Parliament:

Has Government demonstrated sufficient evidence for Parliament to responsibly authorise major reform where key legislative assumptions appear contrary to established evidence, potentially inconsistent with evidence-informed practice, or dependent upon systems and methodologies that have not yet been demonstrated to be ready?

The Alliance supports reform. However, where foreseeable harms may include lost developmental opportunities, family breakdown, workforce exit, crisis escalation or other irreversible consequences, Parliament should require a higher evidentiary and assurance threshold before implementation.

Where the consequences of getting a legislative assumption wrong may be serious or irreversible, Parliament should test that assumption before authorising reform—not wait for people with disability to demonstrate the harm afterwards

Recommendation A: Before authorising the highest-risk reforms, Parliament should require Government to:

a. demonstrate the evidence supporting the contested legislative assumptions; b. publish relevant validation and readiness evidence; c. undertake a whole-of-reform cumulative impact assessment; d. identify foreseeable harms and legislated safeguards; and e. establish measurable human outcomes and clear accountability arrangements before the highest-risk provisions commence.

Submission 429 - Supplementary Submission

No     Bill provision             Legislative assumption                              Evidence direction         Risk rating *

1     s34(1G), (1H), (1J), (1K):  Families can absorb and sustain supervision, personal care,  Contrary to established      Extreme

parental responsibility,     transport, emotional support and behavioural support         evidence

informal supports and indefinitely, and capacity can be presumed rather than family capacity assessed

2      s24(5) and s25A:          Autism-related impairment may resolve with treatment, and   Contrary to established      Extreme

permanence and          treatment must be exhausted before access                evidence

appropriate treatment requirements

3     s9B and the functional    A valid, reliable, autism-appropriate assessment            Evidence gap and contrary   Extreme

assessment framework    methodology exists or will exist, and functioning can be        to established evidence

assessed as stable and context-free

4     s48A: restricted           Support needs change only through unanticipated,           Contrary to established      Very high

unscheduled                significant and ongoing events                            evidence

reassessment triggers

5     s34A, s33(2EA),           Population-level funding rules can fairly serve a population    Contrary to established      Extreme

s33(2EB): class-based     defined by its heterogeneity, without merits review           evidence

funding reductions and caps

6     s25B: alternative support   Alternative service systems are available, accessible,        Evidence gap and          Very high

systems rule-making       affordable and autism-capable now                              potentially inconsistent

power

Submission 429 - Supplementary Submission

No     Bill provision             Legislative assumption                              Evidence direction         Risk rating *

7      s34(1A), (1E), (1F):       Support effectiveness for heterogeneous individuals can be    Potentially inconsistent      High

comparable supports judged by generalisable population evidence, and limited and the evidence evidence justifies refusal hierarchy

8     Repeal of s31 planning     Individualisation, participant direction and choice and          Potentially inconsistent      High

principles control principles are redundant

9     Review rights,            Decisions will be accurate without independent correction or  Contrary to established      Very high

procedural fairness and   human oversight                                         evidence

automated decision making

10    Cumulative impact of     Each reform can be assessed in isolation and the combined  Evidence gap and contrary   Extreme

simultaneous reforms     burden is negligible                                             to established evidence

11    Foreseeable harm and    Harms will be detectable and correctable after               Contrary to established      Extreme

irreversibility (cross-       implementation                                          evidence

cutting)

Table 2: Summary of Legislation to Evidence Matrix

Submission 429 - Supplementary Submission

4 From evidence gaps to safer reform governance

Legislation should not turn an untested assumption into a legal rule and then rely on people with disability to provide the evidence of harm after implementation.

The Alliance’s legislation-to-evidence assessment raises a broader governance question. Where legislation is built on assumptions that are unsupported, inconsistent with, or contrary to available evidence—and where the consequences of error may be serious or irreversible—what should Government and Parliament be required to test before authority to proceed is granted?

This question is particularly relevant to the proposed NDIS reforms. The assessment identifies risks including lost developmental opportunities, increased reliance on families, under assessment of functional need, delayed responses to cumulative deterioration, and support exclusion based on alternative systems whose readiness has not been demonstrated.

These risks reinforce the need for a governance approach that tests whether reforms are genuinely ready before implementation, rather than relying on correction after harm has occurred.

Parliament’s role also does not end with authorising reform. It must also be able to determine, over time, whether the reforms have achieved the outcomes Parliament intended. Parliament cannot effectively hold Government to account unless reforms specify the outcomes they are intended to achieve and how success or failure will be measured and how those results will be transparently reported.

The Australian Autism Alliance has been developing the Systems That Work Framework as a practical governance methodology for major system reform.

It was developed to ask a simple but fundamental question: how do we know a system is capable of delivering better outcomes for people before people are required to rely upon it?

The questions raised through the Senate inquiry, and the findings of this evidence assessment, have reinforced the importance of that work and prompted the Alliance to further develop two components of the Framework: the Systems That Work Pre-Governance Assessment and the Safeguards Index.

4.1 The Pre-Governance Assessment: before Parliament authorises reform

The Pre-Governance Assessment examines whether sufficient design, evidence and assurance have been established before Parliament authorises major reform.

It tests whether:

the problem has been clearly defined and evidenced;
credible alternatives have been considered;
legislative assumptions are supported by validated evidence;
whole-of-society and cost-transfer impacts have been assessed; 13

Submission 429 - Supplementary Submission

dependencies and sequencing are understood;
receiving systems are ready;
foreseeable harms and rights impacts have been considered;
intended human outcomes are measurable;
accountability is clear; and
the reform can be paused, corrected or reversed if assumptions fail. Its final gate question is:

Has Government demonstrated enough evidence for Parliament to responsibly authorise major reform?

Applied to the Bill currently before the Committee, the Alliance submits that this question has not yet been adequately answered.

Recommendation B: Parliament should require application of the Pre-Governance Assessment before authorising major structural reforms to the NDIS.

The Pre-Governance Assessment is attached at Attachment B.

4.2 The Safeguards Index: protection before harm

The Safeguards Index addresses a different question:

What protects the person when a Government or system decision is wrong and the consequence may be serious or irreversible?

The Index focuses on the protections around the individual. It tests:

rights protection;
individual impact;
severity and irreversibility of potential harm;
supported decision-making;
whether evidence is required before support is reduced;
whether a safe alternative actually exists before existing support is withdrawn;
access to review while essential support is maintained;
cumulative harm;
heightened protections for people at greater risk; and
effective restoration and remedy. The evidence assessment demonstrates why this distinction matters.

Early developmental opportunities may be time-sensitive. Functional capacity may deteriorate cumulatively. Family capacity is not unlimited. Autism-related functioning is context-dependent and may be obscured by masking.

In these circumstances, monitoring harm after implementation is not, by itself, a sufficient safeguard.

Submission 429 - Supplementary Submission

A child cannot necessarily recover a missed developmental window.

A family may not readily reverse workforce exit or family breakdown.

A person who has deteriorated into crisis, hospitalisation or restrictive practice has already experienced the consequence of system failure.

Where foreseeable harm may be serious or irreversible, reform should be gated on demonstrated readiness rather than relying on remediation after implementation.

Human rights obligations extend beyond legislative drafting. Parliament should also be satisfied that implementation arrangements, safeguards, monitoring and accountability mechanisms are capable of progressively realising rights in practice. Without agreed outcome measures and public reporting, Parliament cannot determine whether reforms are improving outcomes, maintaining safeguards or progressively realising rights.

The Alliance therefore submits that:

Where a wrong decision can cause serious or irreversible harm, protection cannot come after the harm.

Recommendation C: Parliament should require application of the Safeguards Index before authorising major structural reforms where foreseeable harm may be serious or irreversible.

The Safeguards Index is attached at Attachment C.

Submission 429 - Supplementary Submission

5 NDIS Reform Dependence and Sequencing Problem

The proposed NDIS reforms represent a multi-year structural reset comprising a series of interdependent reforms implemented over different timeframes.

The Alliance mapped the Government’s publicly announced reform timetable against the key dependencies required for reforms to operate safely (refer Tables 3 and 4).

Our assessment considered not whether reforms were scheduled to commence, but whether the systems they depend upon are demonstrably ready, accessible and capable of meeting demand.

The mapping identifies a series of interdependent reforms in which several restrictive or expenditure-reducing measures appear to commence before, or at the same time as, the replacement systems people are expected to rely upon.

The issue is not the pace of reform. It is whether implementation risk is being transferred from Government to people with disability and their families during the transition.

If replacement systems such as Thriving Kids, foundational supports, navigation and mainstream pathways are still being established, unmet need does not disappear. It is likely to be transferred to families, health, education, housing, justice and other parts of Australian society.

Submission 429 - Supplementary Submission

Table 3: NDIS Reform and Sequencing Map

Submission 429 - Supplementary Submission

Dependency that should Apparent sequencing Government action logically precede it question

Are reassessment restrictions Tighter reassessment rules Valid assessment methodology; commencing before the functional commence shortly after workforce capability; review capacity methodology and

Royal Assent safeguards

thresholds are finalised?

Social/community What receiving system has been Mainstream/community inclusion participation budgets demonstrated as ready to absorb capacity reduced — Oct 2026 reduced participation support?

Service model; workforce; Does “rollout commenced” mean Thriving Kids rollout begins geographic coverage; referral families can actually access timely, — Oct 2026 pathways; outcomes framework appropriate support?

Are supports being excluded before Reasonable and necessary Mainstream role clarity and the alternatives are demonstrably tightening — Feb 2027 foundational supports operational?

Is a new planning architecture

Framework planning — Apr  Validated functional assessment     being rolled out only months after

2027                       methodology and trained workforce   the proposed threshold work is

finalised?

Is “at scale” being treated as New eligibility boundaries — Thriving Kids and foundational equivalent to accessible, sufficient Jan 2028 supports demonstrably operational and effective?

Why does the navigation Navigation/support Major participant transition begins architecture follow the major coordination — Jul 2028 Jan 2028 boundary transition by six months?

Table 4:Mapping Dependency and Sequencing to Government Action

The mapping identifies several recurring sequencing risks:

restrictive measures commencing before enabling systems are demonstrably ready;
replacement services being expected to operate from the same date that people transition into them;
major planning and eligibility changes preceding navigation and support arrangements; and
legislative authority being sought before key technical design work has been completed. A same-day dependency is not an assurance margin.

Submission 429 - Supplementary Submission

The proposed timing of Thriving Kids illustrates the issue. A system reaching “national rollout” or being “at scale” from January 2028 is not, in itself, evidence that it is accessible, sufficiently resourced, geographically available or capable of meeting demand before new eligibility boundaries commence.

A commencement date is not a readiness test.

“At scale” is not, by itself, evidence that services are accessible, sufficient, timely, geographically available, appropriate to individual need or capable of delivering better outcomes.

A same-day dependency provides no meaningful assurance margin if implementation is delayed, workforce is insufficient, demand exceeds modelling or services are unavailable in particular communities.

The relevant question is: what period of demonstrated readiness will exist before people become subject to changed access boundaries?

The functional capacity sequence raises a distinct pre-governance concern.

A similar sequencing issue arises with the proposed functional capacity model. Legislative authority may be granted before the Technical Advisory Group has completed the threshold design, assessment methodology, validation and impact testing that will underpin future access decisions.

The apparent sequence becomes:

AUTHORISE → DESIGN → TEST → BUILD → TRANSITION

The Alliance submits that the safer sequence is:

DESIGN → TEST → BUILD → DEMONSTRATE READINESS → AUTHORISE → TRANSITION

Parliament should not be required to authorise the architecture of a high-stakes decision system before the technical body advising on its thresholds and assessment methodology has completed its work.

The cumulative impact of simultaneous reforms must be assessed

Between 2026 and 2028 participants, families, providers and the workforce may experience overlapping reforms to reassessments, planning, eligibility, provider regulation, payments, Thriving Kids, foundational supports and navigation.

The Alliance has not identified a publicly available cumulative impact assessment demonstrating how these reforms interact, where implementation risks may compound, or how those risks will be managed.

The relevant question for Parliament is whether Government has assessed the combined impact of these reforms on the same people, families, workforce and delivery systems—not merely each reform in isolation.

Implementation risk should sit with Government—not the person

The central safeguarding question is who bears the consequences when an enabling system is late, inaccessible or under-capacity.

Under the current sequencing, that risk appears likely to fall on people with disability and their families.

Submission 429 - Supplementary Submission

The Alliance submits that implementation risk should sit with Government, not the person whose support, access or pathway is changing. Where reforms depend upon another system being ready, commencement should be conditional upon demonstrated readiness, including evidence of capacity, workforce, access, safeguards and measurable outcomes.

A commencement date is not a readiness test.

Recommendation D: Parliament should require sufficient assurance that critical dependencies are demonstrably ready before reforms that restrict access, reduce supports or transition participants to alternative systems commence. People with disability should not be required to determine through lived experience whether replacement systems are ready in practice.

Submission 429 - Supplementary Submission

6 The Missing Ledger: A Saving to the NDIS is Not Necessarily a Saving to Australia Government’s financial modelling appears to focus primarily on projected reductions in NDIS expenditure. While Scheme sustainability is an important objective, NDIS expenditure represents only one part of Australia’s broader economic and social picture.

Reducing expenditure within one government program does not necessarily reduce expenditure across government as a whole. Where disability supports are reduced or become less accessible, costs may instead be transferred to health, mental health, education, housing, justice, income support, unpaid carers and families.

The Alliance describes these transferred, deferred and otherwise insufficiently measured costs as the Missing Ledger—the portion of economic and social impact that is not visible when reform is assessed solely through changes in NDIS expenditure.

Accordingly, the relevant question for Parliament is not simply:

Has NDIS expenditure reduced?

Or has Australia become better off?

6.1 A complete assessment requires a whole-of-government perspective

The Alliance submits that assessing value for money solely within the NDIS budget provides an incomplete picture of reform.

A complete assessment of value must examine not only what the NDIS spends, but what Australia gains or loses through changes in independence, education, employment, family capacity, taxation, income support, health and future service dependence. A complete assessment should consider impacts across:

Commonwealth and State expenditure
workforce participation, taxation and productivity
Disability Support Pension and other income support
unpaid family care
health, mental health, education, housing and justice
future support requirements
participant outcomes Without examining these broader impacts, Parliament cannot determine whether reductions in Scheme expenditure represent genuine savings or simply transfer costs elsewhere.

The Missing Ledger also extends beyond cost shifting to other portfolios. It also includes avoidable expenditure generated within the Scheme through repeated reassessments, duplicated

Submission 429 - Supplementary Submission

evidence requirements, external legal costs, Tribunal proceedings and administrative rework arising from poor system design

6.2 The Missing Ledger: Applying Real Case Studies

The following case studies illustrate how decisions that appear to reduce NDIS expenditure may generate substantially higher long-term costs elsewhere. They are intended to demonstrate categories of consequence rather than establish national causation.

6.2a Case Study One – Young Autistic Adult (21): the lifetime cost of a failed education pathway

This case illustrates how failure to maintain an effective education and developmental pathway can create substantially higher long-term costs for government and society.

As verified by the parents and using transparent assumptions relating to future daily supports, Supported Independent Living, reduced parental workforce participation and taxation, lost superannuation and the individual’s lost future economic contribution, the family estimated that failure to restore his pathway and his parents no longer providing the existing support safety net post 70 years of age, could result in additional lifetime costs to government and society of at least $16.5 million.

Component Estimated lifetime cost

Daily Supports $1,209,600

Parent workforce impacts $996,000

Supported Independent Living $13,500,000

Lost future taxation $810,000

Estimated Total $16,515,600

The estimate is not intended to predict the individual’s future with certainty. Rather, it demonstrates that when the broader economic consequences are considered, the costs associated with system failure extend far beyond the original education or disability budget.

6.2b Case Study Two – Young Autistic Entrepreneur Self Employed

The cost of withdrawing support that enabled economic participation

Case Study Two illustrates the potential cost of withdrawing the support scaffolding that enabled economic participation.

Submission 429 - Supplementary Submission

This Autistic individual successfully operated a small business for approximately eight years with more than $80,000 in recorded sales revenue over that period. Following a substantial reduction in NDIS supports (60%), the family reports declining business activity, reduced independence, deterioration in mental health/ well-being, decreased community participation, increased reliance on family support and an application for the Disability Support Pension.

The NDIS internal review confirms that additional employment support was requested and refused on value-for-money and effective-and-beneficial grounds.

The Alliance does not suggest that every subsequent outcome can be attributed solely to one funding decision, without further analysis. Rather, the case demonstrates the categories of value omitted when decisions are assessed only within the NDIS ledger.

A complete assessment would also consider:

the effect on employment and business activity
participant taxation
increased reliance on income support
additional unpaid family care
lost parental workforce participation
health and mental health impacts
review and administrative costs
the risk of greater future support dependence. A decision may therefore appear to save money within the NDIS while producing a larger net cost to Australia.

The Alliance therefore submits that a saving to the NDIS should not be regarded as a saving to Australia unless the broader economic and social consequences have also been assessed.

Recommendation E: Before authorising major structural NDIS reforms, Parliament should require publication of an independent Whole-of-Government Economic and Socioeconomic Impact Assessment.

That assessment should include:

Commonwealth expenditure impacts
State and Territory expenditure impacts 23

Submission 429 - Supplementary Submission

participant and family workforce participation
taxation impacts
income support impacts
health, education, housing and justice impacts
unpaid care
productivity impacts
future support costs
measurable human outcomes. 24

Submission 429 - Supplementary Submission

7 Identifying Administrative Efficiency and Sustainable

Savings

The Australian Autism Alliance supports measures that improve the long-term sustainability of the National Disability Insurance Scheme. Throughout the Senate Inquiry there has been considerable discussion about reducing Scheme expenditure. Comparatively little attention has been given to reducing avoidable costs generated by the administration of the Scheme itself.

The Alliance considers there remains significant opportunity to improve efficiency while maintaining participant outcomes. These opportunities should be comprehensively examined before reductions to participant supports are relied upon as the primary mechanism for achieving sustainability.

For example, substantial public expenditure is already incurred correcting, defending and reviewing decisions after they have been made. Public reporting indicates external legal expenditure associated with participant appeals exceeded $60 million in 2024–25, while broader NDIA legal expenditure has been reported at $75.4 million. These figures suggest that improving first-time decision-making and system design may provide significant opportunities to strengthen sustainability without reducing participant supports

Poorly designed reform also creates avoidable implementation costs through:

redesign and policy revisions
workforce retraining
complaints, reviews and appeals
crisis responses and remediation
rebuilding trust
participant churn
administrative rework Better governance reduces both participant harm and the financial costs of correcting reforms after implementation.

Accordingly, we recommend that Government undertake a comprehensive Red Tape and Administrative Efficiency Review to identify opportunities to reduce bureaucracy, administrative duplication and avoidable expenditure across the disability ecosystem.

Potential areas for reform include the following:

7.1 Reduce unnecessary evidence requirements Many participants with lifelong disabilities continue to experience repeated requests to prove permanent disability despite no material change in their condition.

These reforms would reduce participant burden while releasing allied health capacity for direct support rather than report writing.

Submission 429 - Supplementary Submission

Potential benefits include:

reduced participant burden;
reduced report-writing demand on allied health professionals;
lower administrative costs;
faster decision-making;
reduced internal reviews and Tribunal matters;
more clinician time available for treatment rather than paperwork. 7.2 Improve first-time decision quality Poor initial decisions generate substantial avoidable expenditure. Government should examine the cost of correcting decisions that should have been correct the first time and invest in workforce capability, decision support and quality assurance to reduce these downstream costs.

7.3 Reduce duplication across government Participants frequently provide similar information to multiple organisations. Government should examine opportunities for:

one collection of information with multiple authorised uses
interoperable systems
shared reporting
reduced duplication of compliance activities
streamlined governance arrangements. 7.4 Streamline reporting requirements Evidence provided to the Alliance indicates significant administrative burden associated with:
Behaviour Support reporting
duplicated restrictive practice reporting
repeated provider compliance reporting
fragmented plan management processes
duplicated information requests. Government should examine whether current reporting arrangements improve participant outcomes or simply duplicate information already available elsewhere.

Administrative effort should be proportionate to risk.

Submission 429 - Supplementary Submission

7.5 Improve system design Better system design offers opportunities to reduce unnecessary expenditure. Participants should spend their time receiving supports—not navigating bureaucracy.

7.6 Invest in prevention The Alliance considers there are significant opportunities to reduce long-term expenditure through earlier investment.

Reducing crisis is generally more efficient than funding crisis responses.

Similarly, supporting implementation often delivers greater value than repeatedly funding assessments and reports without ensuring recommendations are successfully translated into practice.

7.7 Improve reform sequencing Where reforms are introduced before replacement systems are operational, governments incur avoidable expenditure. Readiness should therefore be treated as an efficiency measure as well as a safeguard.

Better sequencing reduces implementation costs.

Refer Appendix E for Illustrative Administrative Efficiency Opportunities to the above areas.

Recommendation F: Before participant supports are reduced, Parliament should require sufficient assurance that opportunities to reduce administrative costs, improve decision-making and eliminate avoidable bureaucracy have been comprehensively identified and exhausted.

This includes identifying savings arising from:

reducing repeated evidence requirements
improving first-time decision quality
reducing internal reviews and ART proceedings
reducing duplicated assessments and reporting
reducing fragmented governance
simplifying participant pathways
improving workforce capability
reducing administrative churn
improving digital integration
improving reform sequencing
reducing implementation failures. 27

Submission 429 - Supplementary Submission

The Alliance supports genuine efficiency. Sustainable savings should be achieved by reducing waste, duplication and unnecessary administrative burden before reducing participant supports.

As part of this review, Parliament should require transparent reporting on the administrative costs associated with internal reviews and Administrative Review Tribunal proceedings. For a representative sample of cases, this should include:

the value of the funding decision or reduction under review;
the outcome of the internal review;
the final participant outcome following review or ART proceedings;
the costs of internal review and reconsideration;
external legal expenditure;
Tribunal costs; and
the costs of expert reports or additional evidence requested by either the NDIA or participants.

This information would enable Parliament to determine whether opportunities to reduce avoidable administrative expenditure and improve decision-making have been fully examined before reductions to participant supports are authorised.

Submission 429 - Supplementary Submission

8 Evidence of “Design First” Proposition: Preventing the Cost

of Reform Failure

8.1 Evidence The Australian Autism Alliance submits that major reforms should be designed, tested and demonstrated to be workable before people are required to rely on them.

This principle is well established across systems engineering, implementation science and quality assurance. While developed in different contexts, these disciplines consistently conclude that decisions made during design commit future costs, and that correcting fundamental problems becomes significantly more difficult and expensive once implementation has commenced.

NASA’s systems engineering guidance demonstrates that design decisions made early in a project determine most lifecycle costs, while the National Health and Medical Research Council similarly emphasises the importance of early planning, consultation, testing and implementation support. Although these principles cannot be mechanically transferred to disability policy, they illustrate an established cross-sector proposition:

Decisions made during design shape the cost, safety and performance of the system that follows. The later a fundamental design problem is identified, the more difficult and costly it becomes to correct.

This principle is highly relevant to the proposed NDIS reforms. If reforms are authorised before assessment methods, workforce capability, service pathways, safeguards, information systems and receiving services are demonstrably ready, foreseeable problems may only become apparent after they affect participants and families.

At that point Government may incur the cost of:

correcting decisions and repeating assessments;
redesigning operational guidance, systems and digital infrastructure;
retraining workforces and rebuilding implementation capability;
responding to complaints, safeguarding incidents, internal reviews and Administrative Review Tribunal proceedings;
restoring interrupted supports and remediating failed transitions; and
responding to broader social consequences, including crisis, hospitalisation, family breakdown and subsequent inquiries.

These are not only administrative costs. For people with disability, late correction may mean that the original outcome cannot be fully restored.

Submission 429 - Supplementary Submission

A delayed payment or administrative error may be reversible. A missed developmental opportunity, the loss of employment, deterioration into crisis, family workforce withdrawal or the breakdown of a trusted support arrangement may not be.

Design quality is therefore both a fiscal discipline and a safeguard.

Parliament cannot confidently determine whether Australia’s human rights obligations will be progressively realised unless it has been provided with sufficient assurance that the proposed reforms are supported by evidence, appropriate safeguards, measurable outcomes and accountable implementation arrangements.

A Design First approach does not require reform to be perfect before it begins. Rather, it requires sufficient assurance before implementation that:

the problem has been correctly defined;
evidence, alternatives and meaningful co-design have informed the reform;
assessment methodologies and decision-making processes are valid and reliable;
workforce capability, receiving systems and transition pathways are operational;
safeguards, review mechanisms and accountability arrangements are established; and
outcomes, early warning indicators and mechanisms for course correction have been defined.

This reflects the staged decision-gate approach used in complex systems, where progression occurs only when readiness has been demonstrated rather than assumed.

The cost of correcting reform after commencement

The Alliance recommends distinguishing four categories of reform cost.

Cost category Examples

Design and Research, co-design, options analysis, validation, pilots, impact prevention costs assessment, workforce preparation and safeguards

Implementation Training, systems, communications, commissioning, transition support costs and monitoring

Rework identified before people are affected, including correcting Internal failure costs guidance, systems or processes during testing

Complaints, reviews, litigation, crisis responses, service restoration, External failure participant harm and costs transferred to other systems after costs implementation

Submission 429 - Supplementary Submission

The policy objective should be to invest proportionately in design, validation and prevention to avoid significantly larger downstream correction costs.

The Alliance’s analysis of administrative efficiency identifies many of these failure costs already operating within the NDIS, including repeated evidence requests, duplicated assessments, administrative rework, poor first decisions, internal reviews, ART proceedings, fragmented reporting and repeated correction of problems arising from system design. The next phase of reform should not recreate these avoidable costs.

8.2 Early intervention demonstrates why timing matters The importance of Design First is particularly evident for children.

The evidence cited in the Alliance’s previous work indicates that timely, evidence-informed early intervention for Autistic children can produce substantial long-term benefits, with a conservative benefit-cost ratio previously estimated at 4.1. Delaying appropriate support is therefore not financially neutral—it can forgo developmental, educational, family and future participation benefits that may be difficult or impossible to recover later.

This illustrates the distinction between correcting a system and restoring a life outcome.

Government may later repair an assessment process or commission a missing service. It cannot necessarily return a child to an earlier developmental period or undo years of school disengagement, family workforce loss or deteriorating mental health.

The Alliance does not rely on an unverified claim that 80–85 per cent of costs arise from poor design. Rather, the available authoritative evidence supports the broader and more defensible proposition that system design decisions commit substantial future costs, and the cost of correction generally increases as failure is detected later in the lifecycle.

For disability reform, the consequences extend beyond financial rework. The later system failure is identified, the more costly it becomes to Government—and the more likely it is that harm experienced by people with disability cannot be fully reversed.

Design First is therefore not a reason to avoid reform. It is a governance principle that supports better reform.

The Alliance respectfully submits that Parliament’s role extends beyond determining whether reform is desirable. Parliament should require sufficient assurance that the reforms are supported by evidence, are legally coherent, are compatible with Australia’s human rights obligations, and are accompanied by appropriate safeguards, accountability and implementation arrangements before authorising reforms of this scale.

Well-designed reform enables Government to implement change once, safely and effectively, rather than repeatedly correcting the system and asking people with disability to bear the consequences of avoidable design failure.

Submission 429 - Supplementary Submission

Recommendation G: Parliament should require a Design First Assurance Standard for major NDIS reforms. Implementation should not commence until Parliament is satisfied that:

a. the reforms are supported by sufficient evidence to achieve their intended outcomes; b. the legislation is legally coherent and consistent with Australia’s human rights obligations; and c. appropriate governance, safeguards and accountability arrangements are in place for implementation.

9 Conclusion Parliament is not only being asked whether these reforms should occur, but whether it has been provided with sufficient assurance to justify authorising reforms of this scale.

The Alliance supports reform that improves outcomes, strengthens sustainability and protects rights. However, legislative authority alone is insufficient. Sustainable reform also requires demonstrable readiness, evidence, safeguards, accountability and measurable outcomes.

The Systems That Work Framework provides a practical governance methodology for assessing whether those conditions have been met.

The Alliance respectfully submits that Parliament should not authorise major structural reform until it has been provided with sufficient assurance that these conditions have been met.

Applying the Systems That Work Framework, Parliament should require sufficient assurance that the following conditions have been demonstrated before major reforms proceed:

Readiness

problem clearly defined
implementation capability demonstrated
critical dependencies addressed

Accountability

responsibilities clearly allocated
transparent reporting
corrective action mechanisms 32

Submission 429 - Supplementary Submission

independent implementation monitoring

Safeguards

protections operational
independent oversight
human rights protected

Savings

whole-of-government assessment
cost shifting assessed
implementation costs recognised

Evidence

evidence base published
assumptions tested
alternatives evaluated

Outcomes

success measures defined
monitoring arrangements established
public reporting arrangements established 33

Submission 429 - Supplementary Submission

Attachment A: Systems That Work Framework

Submission 429 - Supplementary Submission

Attachment B – NDIS Amendment Bill 2026: legislation to evidence matrix

A1. Purpose and how to read this matrix This matrix examines a number of the provisions of the Bill by stating the assumption the legislation makes, and tests that assumption against published research and evidence-informed practice. It follows the following structure:

Bill provision,
Proposed change,
Legislative assumption,
Population affected,
Evidence question,
Evidence found,
Evidence strength,
Evidence direction,
Risk if the assumption is wrong; and
Policy implication. Evidence direction uses five classifications: supports proposal, mixed, evidence gap, potentially inconsistent, or contrary to established evidence.

Risk is assessed as likelihood x severity x reversibility. The Alliance allocated risk ratings are marked with an asterisk [*].

The matrix sections are ordered by the Alliance’s assessment of risk.

Table 1 below is a summary of all the provisions at a glance.

Submission 429 - Supplementary Submission

No     Bill provision                   Legislative assumption                                  Evidence          Risk rating *

direction

1     s34(1G), (1H), (1J), (1K):         Families can absorb and sustain supervision, personal care,      Contrary to         Extreme

parental responsibility, informal    transport, emotional support and behavioural support             established

supports and family capacity       indefinitely, and capacity can be presumed rather than           evidence

assessed

2      s24(5) and s25A: permanence    Autism-related impairment may resolve with treatment, and       Contrary to         Extreme

and appropriate treatment        treatment must be exhausted before access                      established

requirements evidence

3     s9B and the functional        A valid, reliable, autism-appropriate assessment methodology    Evidence gap and   Extreme

assessment framework            exists or will exist, and functioning can be assessed as stable     contrary to

and context-free established evidence

4     s48A: restricted unscheduled     Support needs change only through unanticipated, significant     Contrary to         Very high

reassessment triggers          and ongoing events                                             established

evidence

5     s34A, s33(2EA), s33(2EB):        Population-level funding rules can fairly serve a population        Contrary to         Extreme

class-based funding reductions    defined by its heterogeneity, without merits review                established

and caps evidence

6     s25B: alternative support          Alternative service systems are available, accessible, affordable  Evidence gap and   Very high

systems rule-making power      and autism-capable now                                              potentially

inconsistent

7      s34(1A), (1E), (1F): comparable   Support effectiveness for heterogeneous individuals can be        Potentially          High

supports and the evidence       judged by generalisable population evidence, and limited          inconsistent

hierarchy evidence justifies refusal

Submission 429 - Supplementary Submission

No     Bill provision                   Legislative assumption                                  Evidence          Risk rating *

direction

8     Repeal of s31 planning             Individualisation, participant direction and choice and control       Potentially          High

principles                           principles are redundant                                           inconsistent

9     Review rights, procedural         Decisions will be accurate without independent correction or      Contrary to         Very high

fairness and automated        human oversight                                                established

decision making evidence

10    Cumulative impact of           Each reform can be assessed in isolation and the combined      Evidence gap and   Extreme

simultaneous reforms           burden is negligible                                              contrary to

established evidence

11    Foreseeable harm and         Harms will be detectable and correctable after implementation    Contrary to         Extreme

irreversibility (cross-cutting) established evidence

Table1: Summary of Legislation to Evidence Matrix

Matrix 1: Parental responsibility, informal supports and family capacity (s34(1G), (1H), (1J), (1K)) Bill provision Schedule 1 amendments inserting proposed subsections 34(1G) to 34(1K) into the reasonable and necessary test in the National Disability Insurance Scheme Act 2013 (Cth). Verified against the committee interim report, chapter 1, paragraphs 1.103 to 1.105 (link) and the National Legal Aid submission (link)

Proposed change       •  s34(1G): where a participant is a child, the CEO must take into account the presumption that parents are responsible

for substantial care and support. (link)

s34(1H): substantial care and support includes supervision, personal care, transport, emotional support and behavioural support. (link)
s34(1J): funding is barred where the primary purpose is reducing burden on parental time, improving household efficiency, or parental preference for provided support instead of parental care. (link)

Submission 429 - Supplementary Submission

s34(1K): protective considerations apply only where informal reliance would expose the person to a material risk of harm, abuse or neglect, or where informal supports are unsustainable. (link)

Legislative That families of Autistic children and adults can absorb and sustain the specified categories of support indefinitely, that assumption informal support is elastic and free, and that family capacity can be presumed as a legal default rather than assessed case by case.

Population Autistic children and their parents, most directly. Also adult participants who rely on informal supports, siblings, and ageing affected carers. AIHW data show informal carers, 88 per cent of them mothers, provided regular care for 84 per cent of specialist disability service users with autism (link), and ABS data show 43.8 per cent of Australia’s 1.2 million primary carers have disability themselves (link)

Evidence question What does research show about the consequences of assuming parents and informal supports can absorb supervision, personal care, transport, emotional or behavioural support; and about the impact of increasing unpaid care on family functioning, carer employment, health and long-term support sustainability?

Evidence found • Parenting load and stress. Hayes and Watson (2013), meta-analysis: parenting stress in families of Autistic children versus typically developing children showed a large effect size, larger than for Down syndrome or cerebral palsy comparisons. (link)

Carers Australia, Caring Costs Us (2022): carers have a 52.2 per cent employment to population ratio against 75.9 per cent for non-carers, and around 28 per cent of primary carers provide more than 60 hours of care weekly. (link)
Family functioning. A 28-year longitudinal cohort found elevated parental divorce risk in autism families (23.5 per cent versus 13.8 per cent in a matched comparison), persisting into the child’s adulthood. (link)
Shivers and colleagues (2019), meta-analysis: siblings of Autistic people show significantly lower functioning than siblings of people with other disabilities or no disability. (link)
Scale and fragility of informal care. Deloitte Access Economics (2020): replacement value of informal care was 77.9 billion dollars in 2020, and demand for informal care among people with severe restriction is projected to grow 23 per cent to 2030 while carer supply grows only 16 per cent. (link)
Carers NSW 2024 National Carer Survey (over 10,000 respondents): 19.5 per cent of carers experience four or more types of financial stress, a significant increase on previous years. (link)
Formal supports sustain informal care. Harper and colleagues (2013): each additional weekly hour of respite was associated with half a standard deviation improvement in marital quality in couples raising Autistic children. (link)

Submission 429 - Supplementary Submission

Verbakel (2018), 19-country analysis: less generous formal care was associated with fewer informal carers overall. Cutting formal support does not produce more family care. (link)
AIHW: short periods of alternative care ‘can sustain the informal carer in their caring responsibilities’. (link)
Current law. The NDIS (Supports for Participants) Rules 2013, r 3.4, already requires individualised consideration of whether a child’s care needs are substantially greater than other children of a similar age, alongside risks to family wellbeing. The Bill replaces an individualised inquiry with a statutory presumption. (link)
Interim report. The committee heard concerns that increased reliance on informal supports or parental responsibility risks overburdening families, and a submitter’s evidence that ‘when parental responsibility becomes the legal default for everything, parents leave the workforce’. (link)

Evidence strength Strong. Two meta-analyses, national statistics (ABS, AIHW), commissioned economic modelling on ABS microdata, a longitudinal cohort, and Australian primary studies, all pointing one direction.

Evidence direction Contrary to established evidence. No verified source supports the assumption that family capacity can be presumed; the comparative literature shows these families already carry measurably higher load, health, employment and relationship costs.

Risk if assumption * Likelihood: high, the presumption applies to every child participant decision. Severity: severe, carer workforce exit, carer is wrong health decline, family breakdown and safeguarding failures. Reversibility: partly irreversible, workforce exit and family breakdown are rarely undone. Overall: extreme.

Policy implication Supports Attachment A issue 10: remove s34(1H); amend the remainder so that family capacity is never presumed and the legislation requires explicit consideration of willingness, sustainability, carer health, workforce participation, sibling impacts, ageing carers and changed circumstances, with the sustainability of caring arrangements reviewed over time.

Submission 429 - Supplementary Submission

Matrix 2: Permanence and appropriate treatment requirements (s24(5) and s25A) Bill provision Item 89 inserts s24(5); item 92 inserts s25A. Verified against the interim report, chapter 1, paragraphs 1.111 to 1.116 (link)

Proposed change • Impairments are taken not to be permanent, or likely to be permanent, unless the person has undertaken all appropriate treatment. (link)

s25A defines appropriate treatment. Financial and geographic circumstances are not relevant to whether a person has undertaken all appropriate treatment. (link)
Secondary legal analysis: access will only be granted when all appropriate treatment for an impairment has been undertaken and the impairment is likely to be lifelong. (link)

Legislative That autism-related impairment may resolve with treatment, that treatment should be exhausted before disability support assumption is provided, and that capacity to access treatment is universal regardless of cost or location.

Population affected All prospective participants, with Autistic children in developmental windows most exposed. AIHW reports 65 per cent of Autistic people have profound or severe core-activity limitation (link)

Evidence question Does evidence support a legislative assumption that autism-related support needs are temporary or curable through treatment, and what happens when disability support is delayed while treatment is pursued?

Evidence found Autism is lifelong. American Psychiatric Association (DSM-5-TR, 2022): ‘autism is considered a lifelong condition’ while the need for services and supports varies among individuals. (link)

WHO autism fact sheet: some Autistic people live independently while ‘others have severe disabilities and require life-long care and support’. (link)
NICE maintains a standing clinical guideline (CG142) for diagnosing and managing autism in adults, reflecting persistence across the lifespan. (link)
AIHW: autism is ‘a persistent developmental disorder, characterised by symptoms evident from early childhood’. (link)
Australia’s National Autism Strategy 2025 to 2031 aims to improve inclusion, support and life outcomes for almost 300,000 Autistic Australians, a support and participation framing consistent with a lifelong disability. (link)

Submission 429 - Supplementary Submission

The ‘optimal outcome’ literature does not support temporariness. Fein and colleagues (2013), the flagship study, states autism spectrum disorders ‘are generally considered lifelong disabilities’ and that only a minority lose the diagnosis; the study was small, selective and cross-sectional. (link)
Delay causes lasting harm. Pickles and colleagues (2016), PACT randomised controlled trial long-term follow-up in The Lancet: the first evidence of long-term symptom reduction after an RCT of early autism intervention, roughly six years after treatment. (link)
Sandbank and colleagues (2020), Project AIM meta-analysis of 130 samples: significant positive effects for developmental and naturalistic developmental behavioural interventions, holding under RCT-only analysis. (link)
Whitehouse and colleagues (2021), Australian randomised controlled trial (JAMA Pediatrics, CliniKids and Telethon Kids Institute): pre-emptive intervention in infancy reduced clinical autism diagnoses at age three by two thirds, with the authors noting the first two years of life are when the brain is rapidly developing. (link)
Autism CRC evidence synthesis (commissioned by the NDIA): effective intervention during childhood plays an important role in promoting learning and participation in everyday life. (link)
Support and treatment are concurrent, not sequential. Lai and colleagues (2019), Lancet Psychiatry meta- analysis of 96 studies: co-occurring mental health conditions are far more common in autism (ADHD 28 per cent, anxiety 20 per cent), and mental health assessment ‘should be integrated into clinical practice’. (link)
WHO: the health care needs of Autistic people ‘are complex and require a range of integrated services’. (link) Evidence strength Very strong on permanence: the two international diagnostic classifications, WHO, NICE, AIHW and an NDIA commissioned umbrella review align. Strong on delay harms: RCT with long-term follow-up plus meta-analysis. One honest nuance: Sandbank and colleagues report effect sizes attenuate under strict bias controls, so the accurate claim is that developmental and naturalistic developmental behavioural approaches have the strongest evidence, not that all early intervention is equally proven.

Evidence direction Contrary to established evidence. No authoritative source supports treating autism as temporary or as resolvable through treatment exhaustion, and the delay evidence runs directly against a treatment-first gate.

Risk if assumption is * Likelihood: high, the provision operates at every access decision. Severity: severe, denial or delay of support during wrong critical developmental periods, described in the Alliance submission as a loss of life chances. Reversibility: irreversible where developmental windows are missed. Overall: extreme.

Submission 429 - Supplementary Submission

Policy implication Supports Attachment A issue 1: retain recognition of autism as a lifelong neurodevelopmental condition and ensure access does not depend on exhausting treatment options, therapies, educational interventions, family supports or mainstream services.

Submission 429 - Supplementary Submission

Matrix 3: Functional capacity definition and assessment framework (s9B and rules) Bill provision Item 4 inserts proposed s9B defining functional capacity and substantially reduced functional capacity, with thresholds to be set by category A rules expected to commence from 1 January 2028, informed by a Technical Advisory Group with consultation from August 2026. Verified against the interim report, chapter 1, paragraphs 1.43 to 1.48 (link) and the Bills Digest (link)

Proposed change Parliament is asked to legislate an assessment architecture before the methodology, validation, workforce capability and safeguards have been published or independently evaluated. The Alliance submission (Attachment A issues 3 to 5) opposes implementation before these are demonstrated. The functional capacity definition may permit assessment without regard to supports and environment.

Legislative That a valid, reliable and fair functional assessment methodology for Autistic people exists or will exist by assumption commencement, and that functional capacity is a stable, context-free attribute that a point-in-time assessment can measure.

Population affected All participants and prospective participants subject to functional assessment for access or budgets, with Autistic people, women and girls who mask, and people without intellectual disability whose functioning is overestimated, at particular risk.

Evidence question What constitutes a valid, reliable and fair functional assessment methodology for Autistic people, including inter-rater reliability, autism validation, communication differences, masking, fluctuating presentation, context dependence and assessor competency? What are the consequences of using an insufficiently validated methodology for high-stakes resource allocation?

Evidence found • Professional standards. The AERA, APA and NCME Standards for Educational and Psychological Testing treat validity for each intended use, reliability and fairness as fundamental obligations for high-stakes testing. (link)

The I-CAN’s published base is narrow. Arnold, Riches and Stancliffe (2014): internal consistency 0.73 to 0.91 (n

and criterion validity kappa 0.94 (n 49), in general disability samples; the authors describe support needs measurement as ‘still in its infancy’. (link)

Arnold and colleagues (2015): the I-CAN’s funding-prediction evidence rests on one study using a brief research version with participants already stable in their supports. (link)
Verified absence. Live literature searches undertaken by AA could not locate published autism-specific validation, inter-rater reliability or test-retest study of the I-CAN.

Submission 429 - Supplementary Submission

Adaptive behaviour tools mislead in autism. Alvares and colleagues (2020, Australian register, n 2,225): Vineland scores fall significantly below IQ in Autistic children without intellectual disability; intelligence is an imprecise predictor of functional abilities. (link)
Park and colleagues (2019): the WHODAS 2.0 has one published autism validation, 109 Australian Autistic adults without intellectual disability, self-report only; its psychometrics ‘have not been explored extensively’ in autism. (link)
No accepted tool exists. Bölte and colleagues (2019), international ICF core sets consensus: ‘no standardized, internationally accepted tools exist to assess autism spectrum disorder-related functioning’. (link)
Masking defeats observation. Hull and colleagues (2017): camouflaging is effortful and costly, with consequences including exhaustion and threats to self-perception. (link)
Cook and colleagues (2021), systematic review of 29 studies: camouflaging is measurable and higher camouflaging is associated with worse mental health. (link)
Hull and colleagues (2020, n 778): Autistic women report higher camouflaging than Autistic men, making under- assessment a gendered fairness issue. (link)
Cassidy and colleagues (2018): camouflaging and number of unmet support needs each uniquely predicted suicidality in Autistic adults. (link)
Context dependence. WHO ICF, the international standard: functioning and disability occur in a context, and environmental factors are part of the classification itself. (link)
Raymaker and colleagues (2020) and Higgins and colleagues (2021, Australian): Autistic burnout involves loss of skills and reduced daily living skills, so capacity measured on one day is a moving target. (link)
Australia has run this experiment. Joint Standing Committee on the NDIS, Independent Assessments report (October 2021): the proposed tools ‘were not designed to specifically assess functional capacity to inform funding decisions or plans’ and lacked relevance, sensitivity and specificity; opposition was almost universal; independent assessments were abandoned on 9 July 2021. (link)
International analogue. Barr and colleagues (2016): the UK Work Capability Assessment reassessed over one million people using a functional checklist and is the subject of peer-reviewed research into associations with suicides, mental ill-health and antidepressant prescribing. (link)

Evidence strength Strong on the gap: the tools’ own published validation record is the evidence. Strong on masking and context (systematic review plus replicated studies, and the international consensus that no accepted tool exists). The Australian 2021 precedent is a full parliamentary inquiry record.

Submission 429 - Supplementary Submission

Evidence direction Evidence gap on autism-validated methodology (nothing exists to support the framework as safe), and contrary to established evidence on assessing functioning without regard to supports, environment, masking and fluctuation.

Risk if assumption is * Likelihood: high, every access and budget decision would run through the framework. Severity: severe, systematic wrong under-assessment concentrated on those who mask, with a UK analogue linked in peer-reviewed research to serious harms. Reversibility: partly reversible in individual cases through reassessment, but harms accrued in the interim (crisis, hospitalisation, lost schooling) are not recovered. Overall: extreme.

Policy implication Supports Attachment A issues 3, 4 and 5: publication of methodology, independent clinical validation, autism-specific and cohort testing, inter-rater reliability, workforce capability assessment, access to raw outputs, review rights and parliamentary scrutiny before implementation; assessment must include real-world environments, masking, burnout and fluctuating presentation, consistent with the WHO ICF.

Matrix 4: Restricted unscheduled reassessment triggers (s48A) Bill provision Item 21 inserts s48A setting conditions before the CEO can conduct a reassessment; s48A(3) requires an unanticipated, significant and ongoing change in living, education or work arrangements or in the informal support network. Only participants, their plan nominee or guardian can request an unscheduled reassessment. Verified against the interim report, chapter 1, paragraphs 1.58 to 1.61 (link) and the departmental fact sheet (link)

Proposed change A restrictive gate on unscheduled reassessments. Framing correction for the supplementary submission: the Bill inserts ‘unanticipated’ as a new requirement; it does not remove an existing trigger. National Legal Aid criticises the word because it may preclude reassessment for needs that develop gradually (link)

Legislative That support needs change only through discrete, unanticipated events, and that foreseeable or cumulative deterioration assumption does not warrant plan reassessment.

Population affected All participants whose needs change between scheduled reassessments: Autistic people experiencing burnout, students excluded from school, people losing carers through ageing, illness or death, and people whose plans were inadequate from the start.

Submission 429 - Supplementary Submission

Evidence question Can support need materially change without a new impairment or an unanticipated event, and what are the risks of delaying reassessment when plans are no longer sufficient?

Evidence found • Needs change foreseeably and cumulatively. Raymaker and colleagues (2020): Autistic burnout arises from ‘life stressors that added to the cumulative load’ with chronic exhaustion, loss of skills and reduced tolerance to stimulus, absent any new impairment. (link)

Arnold and colleagues (2023, Australian, n 141): burnout episodes are both short and long, recur, and are frequently misread as psychiatric illness, evidencing delayed presentation of distress. (link)
WHO ICF: functioning changes when environments change, independent of impairment. (link)
Foreseeable third-party and life-course changes. Graham inquiry (SA, 2020): suspensions jump 58.9 per cent between Year 7 and Year 8, and exclusions are used against students with disability who were not provided adequate reasonable adjustments. (link)
AIHW: around 7 in 10 Carer Payment recipients are aged 45 and over, so loss of informal support is foreseeable at population scale. (link)
Delay causes documented harm. Commonwealth Ombudsman (2018): NDIA review delays ‘pose a particular risk to those who may be at risk of losing services or experiencing deterioration’, with failure to triage urgent cases. (link)
Winkler (2022): more than 1,430 NDIS participants were stuck in hospital awaiting funding decisions, losing rehabilitation gains, condition, skills, confidence and social connections. (link)
Disability Royal Commission commissioned research: restrictive practices inflict trauma and pain, and adequately resourced individual support is part of the pathway to their elimination. (link)
Every major review points the other way. Tune review and the Government response: parts of the Act are ‘unnecessarily rigid’, grounding the Participant Service Guarantee. (link)
NDIS Review (2023): the reform blueprint is needs-based budgets with check-ins and navigators who ‘help to quickly respond to change in circumstances’. (link)
Productivity Commission (2017): support should ‘not be withdrawn too early’, and plan quality, not throughput, determines outcomes. (link)

Evidence strength Strong. Australian statutory oversight (Ombudsman), a statutory inquiry (Graham), national statistics (AIHW), three successive independent reviews (PC 2017, Tune 2019, NDIS Review 2023), plus peer-reviewed burnout and episodic disability literature.

Submission 429 - Supplementary Submission

Evidence direction Contrary to established evidence. Deterioration is typically foreseeable and cumulative; requiring it to be unanticipated inverts how need actually changes, and delay is documented to convert manageable need into crisis.

Risk if assumption is * Likelihood: high, fluctuation is the norm in this cohort. Severity: severe, crisis escalation, hospitalisation, restrictive wrong practices, school exclusion, family breakdown. Reversibility: partly reversible once reassessment occurs, but interim harms, especially for children, persist. Overall: very high.

Policy implication Supports Attachment A issue 12: remove the ‘unanticipated’ requirement; recognise masking, delayed presentation of distress, cumulative deterioration and Autistic burnout as grounds; provide urgent and crisis reassessment pathways, deemed refusal provisions and temporary stabilisation funding.

Submission 429 - Supplementary Submission

Matrix 5: Class-based funding reductions and caps (s34A, s33(2EA), s33(2EB)) Bill provision Item 34 inserts s34A allowing the Minister to determine, by legislative instrument, reduced funding for groups of supports, including a percentage lower than 100 per cent by which a funding component for a specified group of supports is reduced; these changes are not subject to merits review. Related powers allow caps on supports (proposed s33(2EA) and (2EB)). Verified against the interim report, chapter 1, paragraphs 1.70 to 1.72 and 1.96 (link) and the National Legal Aid submission (link)

Proposed change Ministerial power to reduce or cap funding for whole classes of supports across the scheme, by instrument, without merits review of the resulting individual impacts.

Legislative That population-level funding rules can fairly allocate support for heterogeneous individuals, and that harms will be assumption identified and corrected without individual review rights.

Population affected Scheme-wide. Autistic participants are particularly exposed because heterogeneity of presentation and co-occurrence is the population’s defining feature.

Evidence question What does evidence show about the risks of applying class-based funding reductions or standardised assumptions to heterogeneous populations such as Autistic people? What safeguards are required where population-level rules affect individualised support allocation?

Evidence found • Heterogeneity is the defining feature. Lombardo, Lai and Baron-Cohen (2019): the diagnostic label ‘masks a wide degree of heterogeneity between and within individuals at multiple levels of analysis’. (link)

Lai and colleagues (2019), meta-analysis: co-occurring conditions are common and highly variable, requiring individualised assessment and care. (link)
The Lancet Commission on autism (2022): autism ‘is heterogeneous and requires personalised, evidence-based assessments and interventions’. (link)
Australian lessons on class-based and automated decisions. Robodebt Royal Commission (2023): the scheme reversed the onus of proof onto recipients and calculated debts without human involvement; the Commissioner described it as ‘a crude and cruel mechanism, neither fair nor legal’. (link)
Joint Standing Committee on the NDIS (2021): standardised assessment tools ‘may not be appropriate for certain cohorts of people with disability’; the approach was abandoned. (link)

Submission 429 - Supplementary Submission

Required safeguards under good regulatory practice. Office of Impact Analysis: costs and benefits of new policy must be ‘understood from all angles’, and post-implementation review must recognise ‘the cumulative burden of the policy on individuals’. (link)
NDIS Review (2023), Action 2.4: all Australian governments should incorporate disability impact assessments into new policy proposal processes; Recommendation 25: coordinate and consult on amendments to relevant legislation. (link)
Disability Royal Commission: a human rights approach underpins its 222 recommendations. (link)
Parliamentary Joint Committee on Human Rights: all Commonwealth legislation is examined for human rights compatibility, supported by statements of compatibility. (link)

Evidence strength Strong. Peer-reviewed heterogeneity literature including a meta-analysis and a Lancet Commission, plus two authoritative Australian inquiry records directly on the failure mode, plus Commonwealth regulatory practice standards for the safeguards.

Evidence direction Contrary to established evidence. Applying population-level averages to a population defined by its variance is the design error found in both independent assessments and Robodebt, and the absence of merits review removes the mechanism by which such errors surface.

Risk if assumption is * Likelihood: medium to high, contingent on how the power is exercised. Severity: severe and scheme-wide, individual wrong loss of essential supports with no merits review. Reversibility: partly irreversible, harms accrue before any correction and instruments can operate for extended periods. Overall: extreme.

Policy implication Supports Attachment A issue 11: no broad or scheme-wide application without Category A rules process, human rights assessment, safeguarding assessment, independent disability impact assessment, public transparency, participant notification, written reasons, merits review rights, mandatory post-implementation review and parliamentary scrutiny.

Submission 429 - Supplementary Submission

Matrix 6: Alternative support systems rule-making power (s25B) Bill provision Schedule 1, Part 9. Proposed s25B provides that a person meets the ‘alternative support requirements’ in prescribed circumstances, including where the Minister, with state and territory agreement, determines via delegated legislation that a support is an alternative support. Rule-making powers cover prescribed classes, prescribed circumstances and excluded impairments. Verified against the Bills Digest (link) and the National Legal Aid submission (link)

Proposed change Access can be declined on the basis that another service system is deemed responsible for the person’s support.

Legislative That the alternative systems are actually available, accessible, affordable and autism-capable now, and that disability assumption support needs can be met sequentially through other systems before the NDIS responds.

Population affected Prospective participants directed to health, education, mental health or future foundational supports, with children needing early intervention most exposed to delay.

Evidence question What does evidence show about directing people to theoretical alternative supports, and about the readiness of the systems intended to receive them?

Evidence found • Needs are integrated, not sequential. Lai and colleagues (2019): co-occurrence is the norm in autism, and assessment and care must be integrated into practice rather than sequenced across systems. (link)

WHO: Autistic people’s health care needs ‘are complex and require a range of integrated services’. (link)
Delay converts into harm. The early intervention evidence in matrix 2 (PACT trial, Project AIM meta-analysis and the Australian Whitehouse infant trial) applies with full force where access is deferred pending other systems. (link)
Receiving systems are not demonstrated ready. The NDIS Review frames the NDIS as one part of a larger ecosystem that must be built; foundational supports remain in development. (link)
Attachment A issue 15 records that Government has not demonstrated workforce readiness, service availability, waitlist capacity or autism capability in the receiving systems. * the same readiness gap applies to s25B deeming.
Formal support withdrawal does not conjure alternatives. Verbakel (2018): less generous formal provision is associated with less, not more, informal and substitute care. (link)

Evidence strength Strong on integration and delay harms; the readiness gap is a documented absence of evidence from government rather than a contested empirical question.

Submission 429 - Supplementary Submission

Evidence direction Evidence gap on receiving-system readiness and potentially inconsistent with evidence-informed practice on integrated, concurrent support.

Risk if assumption is * Likelihood: high if rules are made ahead of demonstrated readiness. Severity: severe, support exclusion based on wrong theoretical availability, with short-term savings converting to higher long-term cost. Reversibility: partly reversible, but developmental delay losses are not. Overall: very high.

Policy implication Supports Attachment A issue 2: remove Schedule 1, Part 9 including s25B, or at minimum a no-gap transition guarantee: no participant loses access based on theoretical system responsibility, and alternative systems must be operational, accessible, affordable, autism-capable and independently evaluated before any deeming.

Matrix 7: Comparable supports and the evidence hierarchy (s34(1A), (1E), (1F)) Bill provision Proposed s34(1A) introduces value for money and ‘comparable support’ considerations. s34(1E) sets an order of importance for evidence the CEO may consider, starting with published, peer-reviewed, generalisable research; s34(1F) permits refusal where general evidence is ‘limited’. Verified against the interim report, chapter 1, paragraphs 1.100 to 1.101 (link) and the National Legal Aid submission (link).

Proposed change Support decisions weighted toward generalisable population-level evidence and cheaper comparable supports, with refusal available where the general evidence base is limited.

Legislative That effectiveness for a heterogeneous individual can be judged from generalisable population evidence, and that limited assumption published evidence about a support type justifies refusing it for a particular person.

Population affected Participants whose effective supports have thin population-level evidence bases precisely because the population is heterogeneous and research lags practice; Autistic people with co-occurring conditions are heavily represented.

Evidence question Is it evidence-informed to allocate individual supports by a hierarchy that privileges generalisable evidence and permits refusal where evidence is limited, in a population defined by heterogeneity?

Evidence found • Lombardo, Lai and Baron-Cohen (2019) and the Lancet Commission (2022): heterogeneity means population averages transfer poorly to individuals, and personalised assessment is required. (link)

Submission 429 - Supplementary Submission

Sandbank and colleagues (2020): even for well-studied early interventions, effects vary by approach and by measurement choices, cautioning against coarse evidence hierarchies deciding individual funding. (link)
Autism CRC synthesis: the Australian evidence base supports individualised, goal-directed intervention rather than uniform prescriptions. (link)
Ratti and colleagues (2016): person-centred planning has positive effects on participation and choice outcomes, supporting individualised rather than standardised decision logic. (link)
NDIS Review (2023), Action 3.4: recommended new needs assessment processes that determine the level of need for each participant and set budgets on this basis, an individual rather than class-level decision logic. (link)
- An ‘absence of evidence equals refusal’ rule (s34(1F)) inverts the direction of clinical reasoning for heterogeneous cohorts, where individual response to support is the relevant evidence and published generalisable studies may not exist for valid structural reasons.

Evidence strength Moderate to strong. The heterogeneity literature is strong; the specific application to funding hierarchies is a reasoned inference from that literature rather than a directly studied question.

Evidence direction Potentially inconsistent. The provisions are not directly contradicted by a trial literature, but they sit poorly with the settled heterogeneity evidence and with person-centred practice.

Risk if assumption is * Likelihood: medium to high. Severity: moderate to severe, narrowing of whole-of-person supports and disadvantage to wrong preventative and participation-focused supports. Reversibility: largely reversible through decision review if review rights are preserved. Overall: high.

Policy implication Supports Attachment A issues 7 to 9: remove s34(1A) or qualify it substantially; require public guidance, transparent decision-making, preserved review rights and monitoring of cumulative narrowing; broaden the effectiveness consideration rather than legislating a rigid hierarchy.

Submission 429 - Supplementary Submission

Matrix 8: Repeal of section 31 planning principles Bill provision Item 66 repeals s31 of the NDIS Act (principles relating to plans, including individualisation, participant direction, choice and control). The Explanatory Memorandum asserts the principles will be included in amendments to s4 and new s17B, but the Bills Digest observes that ‘it appears that the rights-based principles, such as the right of the participant to exercise control over their life, will not be replicated elsewhere in the Act.’ (link)

Proposed change Removal of the statutory planning principles that currently anchor individualised, participant-directed planning.

Legislative That the principles are redundant or adequately preserved elsewhere, and that their removal will not change planning assumption practice.

Population affected All participants; most sharply those who depend on the principles to contest standardised planning decisions.

Evidence question What is the evidence base for person-centred, individualised planning and participant direction, and what is lost if the anchoring principles are repealed?

Evidence found              •  Ratti and colleagues (2016), systematic review of 16 studies: person-centred planning has positive, moderate effects

on community participation, activity participation and daily choice-making; the authors note the primary evidence quality is low. (link)

Lakhani and colleagues (2018), systematic review: self-directed models can increase service users’ control, realised when genuine decision-making support is provided. (link)
Lai (2020): autism comprises multiple spectra with multilevel heterogeneity, which is why individualised planning matters clinically. (link)
NDIS Review (2023): the reform blueprint is built around a person-centred scheme, not the dilution of individualisation. (link)
Disability Royal Commission, volume 6: ‘autonomy is a person’s right and freedom to make decisions, control their life and exercise choice’, with denial of autonomy treated as systemic neglect. (link)

Evidence strength Moderate on measured effects (the review literature is honest that effects are moderate and trial quality is mixed) and very strong on the rights and policy architecture (Royal Commission, NDIS Review, CRPD).

Evidence direction Potentially inconsistent. Repeal moves the law away from the direction of every recent authoritative Australian review; the empirical case for removal is absent.

Submission 429 - Supplementary Submission

Risk if assumption is * Likelihood: medium. Severity: moderate to severe, diffuse erosion of individualisation across all planning decisions, wrong hard to litigate once the anchor is gone. Reversibility: reversible in law, but decisions made in the interim are not. Overall: high.

Policy implication Supports Attachment A issue 14: retain s31 or explicitly preserve equivalent rights-based protections in the Act, verified on the face of the amended text rather than asserted in the Explanatory Memorandum.

Matrix 9: Review rights, procedural fairness and automated decision making Bill provision s34A determinations are not subject to merits review (interim report, chapter 1, paragraph 1.70) (link). Attachment A issues 13 and 19 also record Alliance concerns about reduced review rights, procedural fairness and automated decision making across the reform package.

Proposed change Key instruments and decisions sit outside merits review, and the package reduces procedural protections while increasing standardisation and automation of decision inputs.

Legislative That decisions will be sufficiently accurate without independent correction, and that people affected can protect their own assumption interests without procedural accommodations.

Population affected Participants facing communication or advocacy barriers are disproportionately affected: exactly the cohort least able to identify and contest error.

Evidence question What are the consequences of weakening procedural fairness, merits review and human oversight in social security and disability contexts?

Evidence found • The Australian case study. Robodebt Royal Commission (2023): removing human oversight and reversing the onus of proof produced systemic, unlawful harm concentrated on vulnerable people; debts were calculated ‘without human involvement’. (link)

The normative standard. CRPD article 13: effective access to justice requires ‘procedural and age-appropriate accommodations’; article 12: equal recognition before the law with support to exercise legal capacity. (link)
ALRC Report 124: the will, preferences and rights of persons requiring decision-making support ‘must direct decisions that affect their lives’, with recommendations to recognise supporters in the NDIS Act. (link)

Submission 429 - Supplementary Submission

Supported decision-making works. Douglas and Bigby (2020): an evidence-based Australian practice framework for decision support exists; Bigby and colleagues (2022) demonstrated its efficacy in building supporter capability. (link)
Hidden disability, masking and communication differences may be misinterpreted by automated systems (see matrix 3 masking evidence), creating algorithmic bias risk recorded in Attachment A issue 19. (link)

Evidence strength Strong. A Royal Commission finding, a ratified treaty standard, a national law reform report and Australian trial evidence for the alternative.

Evidence direction Contrary to established evidence. The most recent and most authoritative Australian evidence shows what happens when review and human oversight are removed from high-volume decisions affecting vulnerable people.

Risk if assumption is * Likelihood: high in a scheme of this decision volume. Severity: severe for those least able to self-advocate. wrong Reversibility: partly reversible where review is eventually restored, but Robodebt shows harms run for years first. Overall: very high.

Policy implication Supports Attachment A issues 13 and 19: independent review rights for all substantive decisions affecting eligibility, supports and funding; access to reports and reasons; human oversight, bias auditing, transparency and explainability for any automated processing; supported decision-making embedded in the Act.

Matrix 10: Cumulative impact of simultaneous reforms Bill provision Cross-cutting: the combined operation of the access changes (s24(5), s25A, s25B), the assessment framework (s9B and rules from 2028), the reasonable and necessary changes (s34), reassessment restrictions (s48A), class-based reduction powers (s34A, s33(2EA), (2EB)), the repeal of s31, and the foundational supports transition. The Bill passed the House on 2 July 2026 with 30 amendments agreed on 1 July 2026 (link)

Proposed change Multiple interacting reforms land on the same cohort within a compressed implementation window, with key elements commencing at different times to 2028.

Submission 429 - Supplementary Submission

Legislative That each reform can be assessed in isolation and that the combined burden on participants and families is negligible or assumption self-correcting.

Population affected The whole scheme population, concentrated on those touched by several reforms at once: Autistic children (access, assessment, family capacity and reassessment changes simultaneously) and families already carrying high unpaid care loads.

Evidence question What does research show about cumulative impacts of multiple simultaneous policy reforms on vulnerable populations, and how should cumulative burden be assessed?

Evidence found • EHRC evidence review (2018): some groups are affected by a wide range of reforms at once, ‘and this can only be measured by a cumulative impact assessment’; single-reform equality assessments systematically miss combined effects. (link)

Portes and Reed for the EHRC (2018): cumulative modelling of UK tax and welfare reforms found households with a disabled adult and a disabled child lost just over 6,500 pounds a year, more than 13 per cent of net income, the largest losses of any group. (link)
Nilsen and colleagues (2019), implementation science: concurrent organisational change produces disengagement and change fatigue among frontline professionals, degrading implementation quality. (link)
Office of Impact Analysis, post-implementation review guidance: reviews must recognise ‘the cumulative burden of the policy on individuals’, and isolating the impact of a single change is sometimes not possible. (link)
The Alliance submission’s cumulative impact lens (section 7) already frames this; the matrix supplies the external evidence base for it.
Note: No published Australian whole-of-reform cumulative impact assessment for this package was able to be located for this submission; whether one exists internally to government is unknown. If it exists it is not easily found.

Evidence strength Moderate to strong. The UK cumulative assessment work is quantified and directly analogous; the Commonwealth’s own guidance requires cumulative burden assessment; the mechanism evidence is peer reviewed.

Evidence direction Evidence gap (no cumulative assessment published for this package) and contrary to established evidence (assessing interacting reforms one at a time predictably understates combined loss).

Submission 429 - Supplementary Submission

Risk if assumption is * Likelihood: high, interaction is structural, not hypothetical. Severity: severe, compounding losses concentrated on the wrong most affected families. Reversibility: partly irreversible, compounding effects on development, employment and family stability persist. Overall: extreme.

Policy implication A whole-of-reform cumulative impact assessment, disaggregated for disabled people and Autistic participants, before commencement of the major elements; aligns with Attachment A issue 18 (outcome measurement before implementation) and the Systems That Work Pre-Governance Gate.

Matrix 11: Foreseeable harm and irreversible consequences (cross-cutting) Bill provision Cross-cutting. The package commences major changes without legislated harm monitoring, readiness gates or outcome measurement (Attachment A issue 18), while several identified risks operate on developmental windows that do not reopen.

Proposed change Implementation proceeds on the assumption that harms, if they occur, will be detected and corrected afterwards.

Legislative That harm from wrong assumptions is observable, attributable and correctable after the fact. assumption

Population affected Most acutely children in early developmental windows, and participants whose deterioration (burnout, family collapse, institutionalisation) is practically irreversible once it occurs.

Evidence question What does evidence relating to foreseeable harm and potentially irreversible consequences require of reform design?

Evidence found • Developmental windows are real and time-limited. Harvard Center on the Developing Child: young children are more biologically sensitive to adverse exposures than adolescents, who are more sensitive than adults. (link)

Whitehouse and colleagues (2021), Australian randomised controlled trial reported in JAMA Pediatrics: pre-emptive intervention in infancy reduced clinical autism diagnoses at age three by two thirds, with the authors noting the first two years of life are when the brain is rapidly developing. The corollary is that missed windows are not recovered. (link)

Submission 429 - Supplementary Submission

Decision frameworks for irreversible risk. The precautionary principle (Rio Declaration principle 15): where there are threats of serious or irreversible damage, lack of full scientific certainty must not be a reason for postponing measures. (link)
ISO 31000 provides the accepted structure for risk assessment by likelihood and consequence, which the work program extends with reversibility, a defensible and conservative extension for disability policy. * (link)
Disability Royal Commission restrictive practices research: some harms (trauma from restraint and seclusion) are inflicted, not merely risked, when support fails. (link)

Evidence strength Strong on developmental windows (Australian RCT plus authoritative synthesis); the precaution and risk-standard sources are authoritative framing rather than empirical studies and are presented as such.

Evidence direction Contrary to established evidence for any implementation design that treats developmental-window harms as correctable later.

Risk if assumption is * Likelihood: high across a population of this size. Severity: severe. Reversibility: irreversible by definition for the harms wrong in question. Overall: extreme. This is the category that justifies gating commencement on demonstrated readiness rather than remediation after harm.

Policy implication Grounds the Systems That Work Pre-Governance Gate and Readiness Test in accepted risk practice: publish the risk assessment, gate commencement on readiness, monitor outcomes from day one (participation, education, employment, wellbeing, family sustainability, school exclusion, hospitalisation, restrictive practices) as sought in Attachment A issue 18.

Submission 429 - Supplementary Submission

Attachment C – The Pre-Governance Assessment: before Parliament authorises reform

Submission 429 - Supplementary Submission

Attachment D – The Safeguards Index: protection before harm

Submission 429 - Supplementary Submission

Attachment E – Illustrative Administrative Efficiency

Opportunities

Potential areas for reform include the following:

Potential efficiencies include:

eliminating repeated proof of permanent disability
extending the validity of trusted clinical evidence
reducing repeated specialist reports where disability has already been established
allowing long-term GP certification to verify that no material change has occurred for lifelong conditions where appropriate
reducing duplicated functional assessments
reducing repeated collection of information already held by Government
replacing full reassessments with targeted plan variations where appropriate. These reforms would reduce participant burden while releasing allied health capacity for direct support rather than report writing.

For example, for participants with lifelong, permanent disabilities where eligibility has already been established:

use trusted clinical evidence for longer;
allow a treating GP to certify there has been no material change;
avoid repeated specialist reports where there is no clinical reason for reassessment;
use targeted plan variations rather than full reassessments where appropriate. Potential benefits include:
reduced participant burden;
reduced report-writing demand on allied health professionals;
lower administrative costs;
faster decision-making; 61

Submission 429 - Supplementary Submission

reduced internal reviews and Tribunal matters;
more clinician time available for treatment rather than paperwork. 7.2 Improve decision quality to reduce downstream costs Poor initial decisions generate substantial avoidable expenditure through:
internal reviews
Administrative Review Tribunal proceedings
external legal costs
repeated expert reports
participant advocacy
administrative rework
implementation delays. Government should examine the cost of correcting decisions that should have been correct the first time and invest in workforce capability, decision support and quality assurance to reduce these downstream costs.

7.3 Reduce duplication across government Participants frequently provide similar information to multiple organisations, including:

NDIA
NDIS Quality and Safeguards Commission
Administrative Review Tribunal
State restrictive practice bodies
safeguarding bodies
Community Visitors
Ombudsman
advocacy organisations. Government should examine opportunities for:
one collection of information with multiple authorised uses
interoperable systems
shared reporting
reduced duplication of compliance activities 62

Submission 429 - Supplementary Submission

streamlined governance arrangements. 7.4 Reduce administrative reporting burden Evidence provided to the Alliance indicates significant administrative burden associated with:
Behaviour Support reporting
duplicated restrictive practice reporting
repeated provider compliance reporting
fragmented plan management processes
duplicated information requests. Government should examine whether current reporting arrangements improve participant outcomes or simply duplicate information already available elsewhere.

Administrative effort should be proportionate to risk.

7.5 Improve system design Better system design offers opportunities to reduce unnecessary expenditure, including:

reducing repeated participant retelling of their story
reducing administrative churn
reducing unnecessary correspondence
reducing fragmented navigation arrangements
creating a single accountable pathway for participants
improving digital integration across agencies
simplifying administrative processes. Participants should spend their time receiving supports—not navigating bureaucracy.

7.6 Invest in prevention rather than correction The Alliance considers there are significant opportunities to reduce long-term expenditure through earlier investment in:

evidence-based early intervention
workforce capability
implementation support
family capacity
market stewardship including positive behaviour support
navigation 63

Submission 429 - Supplementary Submission

community capability. Reducing crisis is generally more efficient than funding crisis responses.

Similarly, supporting implementation often delivers greater value than repeatedly funding assessments and reports without ensuring recommendations are successfully translated into practice.

7.7 Reduce avoidable reform implementation costs Where reforms are introduced before replacement systems are operational, governments incur avoidable expenditure associated with:

remediation
complaints
crisis intervention
transition failures
reviews
independent inquiries
emergency responses. 64

Australian Autism Alliance

Supplementary Submission

to the Senate Community Affairs Legislation

Committee Inquiry into the National Disability

Future Generations) Bill 2026

Ensuring Reform is Ready, Accountable and Safe

Submitted by: Australian Autism Alliance

Contact: Jenny Karavolos, Chair, Australian Autism Alliance

Acknowledgements

Contents

About the Australian Autism Alliance

Autism Advisory Group, the NDIA DRCO Co-Design Advisory and Reform groups, NDIS

Commission Disability Sector Consultative group, the National Autism Strategy Oversight

1 Executive Summary

Whole-of-Government Economic and Socioeconomic

Design-first

Table 1: Supplementary Recommendation Summary Table

5 NDIS Reform Dependence and Sequencing Problem

Table 3: NDIS Reform and Sequencing Map

Royal Assent safeguards

AUTHORISE → DESIGN → TEST → BUILD → TRANSITION

DESIGN → TEST → BUILD → DEMONSTRATE READINESS → AUTHORISE → TRANSITION

6.2 The Missing Ledger: Applying Real Case Studies

Daily Supports $1,209,600

Supported Independent Living $13,500,000

Estimated Total $16,515,600

6.2b Case Study Two – Young Autistic Entrepreneur Self Employed

7 Identifying Administrative Efficiency and Sustainable

Savings

of Reform Failure

Cost category Examples

Readiness

Accountability

Safeguards

Savings

Evidence

Outcomes

Attachment A: Systems That Work Framework

Table1: Summary of Legislation to Evidence Matrix

Attachment E – Illustrative Administrative Efficiency

Opportunities