SIL commissioning is coming.

The NDIA buys $16 billion of SIL support a year for nearly 37,000 participants. The largest provider holds under 4% of the market and the top ten deliver around 10%. That fragmentation, layered onto a market for lemons and a common-goods funding problem, has produced steady quality decay and cost escalation. This article sets out the structural faults that make SIL ungovernable, shows that aggressive commissioning is the only rational response, and explains how the enacted legislation works inside a commissioning framework to cut cost, provider numbers and participant choice at once.

The structural faults

SIL combines two perverse incentive structures that few markets manage at once.

The first is that SIL is a market for lemons. Quality is opaque, because the things that matter, a participant's safety, dignity and progress, are slow to appear, hard to see from outside the house, and hard to attribute to the provider rather than the participant. Because the price is fixed, providers exploiting this opacity enjoy substantially higher margins. Over time that pushes good providers out as the freed margin is recycled into marketing and conflicted referral.

Correcting a lemons market takes a strong, active regulator. SIL does not have one, because it is too fragmented to police. With over 3,000 providers, the expected cost of being caught and punished sits far below the private gain from defecting, and the providers caught anyway buy their way out through asset sales, phoenixing and related-party transfers. This is a one-shot game: detection is unlikely, the Agency cannot enforce conduct, and a small provider risks almost nothing each time it under-services, unlike a major provider staking thousands of participants.

The second is a common-goods problem on the cost side. Funded need is pressured by a small group whose incentives all point one way. The participant wants more support, reasonably. The provider wants a bigger package, because that is a bigger book. The support coordinator and clinicians who shape the plan are effectively engaged by the participant, advocate for them, and carry none of the cost they recommend. Everyone at the table has a reason to push the package up. The funding distribution suggests this: 20% of participants consume close to 50% of SIL funding, and while much of that is genuine need, a skew that sharp is the signature of an incentive structure.

The two faults compound in the business model. With needs assessment inaccurate and the capacity to under-service latent, the rational strategy is to select participants funded above the cost of serving them and to expand stated need wherever possible. The one-shot structure leaves the Agency unable to respond to that arbitrage, so whole business models are built on exactly this pattern. The market cannot reward moving a participant toward independence, because independence shrinks the package. It rewards selecting and escalating.

The most damning feature of the current structure is that it has made working against a participant's independence the commercially dominant strategy.

Moving the pieces into play

This escalation is already being shut down, and unfortunately it will catch participants and providers with genuine unmet need. While most sensible readers will notice that the validity literature on the I-CAN is laughable, the point is not accuracy, it is standardisation. Packages will be reference-normed, and the escalation pathways locked, by new legislation:

• Needs assessments are now conducted by the Agency, not the participant's own clinician s32L(4A).
• Support Coordination is being nationalised, with major changes to Plan Management.
• Unscheduled reassessment is gated behind genuine, significant, ongoing change, with the decision window stretched from 21 days to 90 ss48A, 48(3).
• Plans auto-renew at existing funding with no reviewable decision on the renewal, while merits review of the plan itself survives ss50A, 103A.
• Funding for a class of supports can be cut by a set percentage across plans, expressly even below the cost of the support, subject to participant safety s34A, from 1 Oct 2026.
• The Minister can cap the funding amount, the intensity, or the worker-to-participant ratio for a class of supports s33(2EA), from 1 Feb 2027, pointed straight at the one-to-one-and-above ratios that consume half of core daily activities spend.

Taken together, the rules lock each plan inside a fixed reference range: the package is set by a standard assessment, cannot be quietly re-escalated, and renews at the same number. Each of those provisions is in the bill before the Parliament. You can read the primary text yourself, linked in the evidence panel below.

Why it backfires in SIL

This approach will not work in SIL, and will likely make the problem worse. That is precisely why commissioning is not optional.

Standardising the package compresses funding toward a consistent level, which pushes a large cohort below the point at which they are commercially worth serving. In a market with provider choice and no obligation to serve, the rational provider simply refuses them. Those participants are then stranded, and the Agency's only release valve in an open market is to escalate the package back up until someone takes them. So the barricade dropped into the current market does not cut spend. It strands the unprofitable, forces re-escalation, and defeats its own purpose, while the lemons dynamic ensures the participants who are taken on are quietly under-serviced or warehoused in the vulnerable not-for-profits least able to refuse them. Anyone watching the rollout without the second half of the reform will read this as proof the reform has failed, when it is proof the reform is only half-built.

Commissioning is the half that makes the rest work

Commissioning is the mechanism that lets the standardised assessment actually deliver, by closing the arbitrage that would otherwise strand the newly-unprofitable cohort and force their packages back up.

The instrument is the bundled slot. Under commissioning, a provider does not pick up participants one at a time. It wins a slot through a competitive tender, and the slot is defined as servicing every participant in a region at the standardised allocation. The profitable and the unprofitable participants are bundled into a single indivisible obligation, and the provider accepts the unprofitable ones as the price of winning the profitable ones. This is precisely the universal-service obligation that regulators have used for a century to solve exactly this stranding problem in utilities and telecommunications: you may have the profitable market only on condition that you also serve the unprofitable customers, with the tender as the lever that binds the two together.

This re-prices defection completely. In the one-shot market, refusing a marginal unprofitable participant costs the provider nothing and gains it the avoided loss. Under commissioning, refusing or under-servicing that same participant risks the slot, which is to say the entire regional book, because the obligation to serve all participants is the condition on which the slot is held. The marginal participant is no longer a discrete loss to be avoided. It is the price of holding the whole region, and no rational provider gives up the region to shed one participant.

The single-slot framing still understates it, because the buyer is a monopsony and the provider holds a portfolio of slots, every one of them against that same buyer, who sees everything it does in every region and remembers all of it. This is a repeated game played across the provider's entire national footprint against the only customer it will ever have. Bernheim and Whinston showed that when the same two parties meet across many markets at once, cooperation becomes far easier to enforce, because defection in any one market can be punished across all of them. So a provider that under-services or arbitrages in one region is not risking that region. It is revealing its type to the buyer that controls its revenue everywhere else it operates and every future tender it hopes to win. The stake on a single act of defection is the whole enterprise, present and future, which is why arbitrage stops being merely unprofitable and becomes unthinkable: no rational provider risks its national existence to capture an allocation error on one participant. The same mechanism closes the quality defection, because under-delivery is now detected by the one buyer and punished across the whole book, and the cheapest way to protect the portfolio is to actually perform. Both defections close with the same move, because both grew from the same one-shot fragmentation, and consolidation turns the one-shot player into a portfolio player who cannot afford either.

This is also where participant choice quietly goes. In the open market a participant can in principle approach any of three thousand providers. Under commissioning they choose among the one to ten that hold slots in their region, and because allocation will almost certainly run on vacancy ordering, the real choice on any given day is whoever has a bed, which is one to three. Choice does not disappear on paper. It collapses to whatever the slot structure leaves available, which is the price of making the slot mean anything at all.

The enforcement the Agency now has

The repeated game only disciplines a provider if the buyer can actually see, compel, and punish. For most of the scheme's life it could not, which is why even the head of the market was only loosely governed. The bill changes that, and this is the part that justifies describing the shift as a revolution rather than a tightening.

The Agency, the buyer itself, now holds its own monitoring, investigation, and search-and-seizure apparatus over its own functions, with NDIA inspectors and investigators appointed to use it new Part 3C of Ch 4. It can compel a person to appear and answer questions, with a civil penalty for refusal (sections 54 and 53(3)). It can require seven years of records to be kept and produced (section 45B). It has cut the claim window from two years to 90 days, which is a direct cash-flow and reconciliation discipline on thin-margin operators (section 45A). And it can run allocation decisions through automated decision-making at scale, including evaluative determinations governed by standard operating procedure instruments (sections 59B to 59E), with pricing decisions moving to the Minister (section 45C).

Put those together and the change is not incremental. The Agency moves from policing three thousand one-shot strangers it can barely see, to managing a panel of twenty to thirty portfolio players plus a recompetable regional tail, with automated allocation, its own investigatory teeth, compulsion powers, unilateral compression, and ratio caps. The difference between those two worlds is the difference between a quality-assurance task that is structurally impossible and one that is routine. High-touch, case-by-case quality management, the kind the hard cases actually need, becomes feasible for the first time, because there are few enough counterparties to manage closely and enough leverage to hold each of them to account.

The wall: why it cannot consolidate all the way

If consolidation is what restores the Agency's control and makes the rest of the reform work, the naive conclusion is that the Agency wants as few providers as possible. This is the logic behind the periodically floated idea of rolling the sector into five to twenty mega-providers backed by institutional capital. It is wrong, and it is wrong for the same hold-up reason that created the original problem, now running in reverse.

The Agency's recovered control depends on the cross-portfolio threat being credible, and that threat is only credible if each provider remains removable. Removability has a ceiling that scales with size. A provider holding a few hundred participants can be removed, because the Agency can rehome them. A provider holding several thousand cannot, because removing it would strand an unabsorbable number of vulnerable people at once, and the Agency, which carries the safety obligation, will never pull that trigger, and the provider knows it. The moment a provider becomes too big to remove, the multimarket enforcement that the whole structure relies on stops being credible against it, and the hold-up power inverts straight back to the provider.

The optimal provider is big enough to have everything to lose and small enough that the Agency can still take it.

So the monopsony's enforcement power and the cartel wall are the same constraint seen from opposite ends. The Agency wants each provider holding enough slots that cross-portfolio retaliation gives it everything to lose, and few enough providers overall that it is worth disciplining, but no provider so large that retaliation becomes incredible because it cannot be dropped. The five-provider model sails straight through the wall: five national providers are each unremovable by construction, the enforcement evaporates, and the Agency would have handed away its leverage while routing public funding to investor returns, which cuts against both its fiscal logic and its plain preference for not-for-profit and government delivery. The overhead saving such a model offers is real but small against a thin margin, and the actual driver of scheme cost sits upstream in allocation and eligibility, untouched by how many corporate roofs the houses sit under. It solves the wrong problem and surrenders the prize doing it.

The defensible consolidation sits in the band between the two failures, and that band is not a matter of taste. It is set by the arithmetic of the market, and it can be read directly off the data the Agency already publishes.

Reading the answer off the data

We built this as a constrained problem: the fewest providers consistent with keeping each one removable, given how SIL is actually delivered. The full model is the interactive tool below, and the conclusion holds across almost any assumptions you feed it.

Start with the market we have. The top fifty providers hold roughly a third of the national market and the largest sits under four percent. That is a fat head on an enormous hollow tail, the signature of a market that grew by registration rather than delivery, which means the delivered market is already far smaller than the three-thousand headline implies. Commissioning is the deliberate widening of that head and clearing of that tail.

There is a second enforcement mode that the portfolio logic alone does not capture, and it matters because not every district can carry a portfolio player. In the thin regional districts, the participant base is too small to sustain providers at the scale that makes multimarket retaliation bite. The discipline there is not the portfolio threat; it is direct recompetition. Those districts are filled by packing in as many providers as the base sustains at minimum-viable scale, each individually small enough that the Agency can recompete its position outright if it fails. The two modes are the same principle, credible removal, seen from opposite ends. In the thick districts, removability is preserved by capping any one provider's share so it can always be dropped. In the thin districts, removability is preserved by smallness, so any one provider can always be replaced. Enforcement is universal across the map; only its instrument changes with the density of the district.

The model runs in two steps. The first turns geography into positions. SIL is delivered house by house, so the unit is the service district, of which the NDIA recognises around eighty. Each district needs a minimum number of provider positions, which we call slots, set by two competing limits. A viability floor, because a provider needs enough participants to run a stable roster and absorb a death or an exit without a house dropping below break-even, which puts the floor around a three-house cluster. And a removability cap, because no provider should hold more than a defensible share of one district, which we model at ten percent. In a large metropolitan district the cap binds and the district carries many contestable providers. In a thin regional district the floor binds, achievable concentration is forced up, and the district stays contestable only through direct recompetition. That cap-binds versus floor-binds split is those two enforcement modes, now drawn on a map.

The second step turns positions into organisations, governed by reach. A provider fills slots across many districts but may hold only one slot per district. The question is how many districts a provider can effectively cover, and the answer is governed by decay: the first region is cheap, each additional region costs more for less. We measure that decay directly in today's top fifty, allocate slots largest-first under it, cap any provider at ten percent of a district and five percent of the nation, and let the curve run until the slots are gone.

The tool below runs both steps live. Move the regional share cap and the reach dial, and watch the survivor set resolve.

The five percent national cap is the wall from the previous section, expressed as a number: the point past which the largest survivor grows into the un-removable monster, and where the minimisation stops.

The output has one feature worth dwelling on. The total survivor count is unstable, swinging from several hundred local operators to around forty national ones depending on how far providers extend their reach. But the number holding a genuinely substantial book barely moves. It sits at twenty to thirty across the entire plausible range. Everything below it is mop-up. The argument over whether commissioning leaves forty providers or four hundred is a distraction. There will be twenty to thirty organisations that matter, built on thick metropolitan slots, above a tail whose size depends only on how far the survivors choose to spread.

We have run this exact play before

If the argument feels like theory, it is worth knowing that Australia has already executed this precise monopsony-consolidation manoeuvre in an adjacent market, and the outcome is on the record. Federal employment services is the closest structural cousin SIL has: a demand-driven, income-linked human-services market purchased by a single government buyer from a fragmented supplier base. It went through exactly this transition.

The provider count fell from 300 organisations under Job Network, to 141 under Job Services Australia, to 44 under jobactive, to 43 under Workforce Australia. A contraction of around eighty-six percent, and once it reached the licensed-panel form, it stayed there. The mechanism the department used to force it is the detail that matters most, because it is the same instrument now arriving in SIL: it required providers to service entire geographic regions rather than cherry-picking within them, which is whole-of-region commissioning with mandatory acceptance. And the footprint barely moved while ownership concentrated: site numbers held roughly steady, from about 1,700 to 1,482, even as the organisations above them collapsed.

The obvious objection is that SIL is place-bound in a way employment services was not. You cannot deliver a supported home online, and you cannot reassign a participant to another caseload the way you can a jobseeker. That is true, and it is the basis of the hold-up problem set out earlier. But it does not break the precedent, because consolidation does not require the asset to move. It requires only the corporate roof to move, and the asset to stay exactly where it is. This is precisely what happens in the SIL consolidations the sector has already seen through merger, acquisition, and insolvency. The acquiring provider takes on the staff, the transfer is negotiated through the new organisation assuming the employment liabilities, leases novate to the new provider, owned property is sold to it. The participant stays in the same house, with the same team, under a new roof. The same pattern is visible in the reallocation of state out-of-home-care contracts.

The outgoing provider also cannot weaponise the participants on the way out. Commercially restricting clients to spite the buyer it depends on for its entire revenue is suicidal, and morally indefensible besides, which is why these transitions almost always settle rather than detonate. The one genuinely dangerous case is the disorderly one: insolvency with no acquirer stepping in. That case is the whole reason the structure insists every provider stay removable and rehomeable. The keep-everyone-removable design is not only the Agency's enforcement lever; it is the safety mechanism that ensures the disorderly case never grows beyond what the Agency can absorb.

That precedent does two things. It confirms the theory is not speculation, because the monopsony-consolidation logic predicted the move and the move happened, by the exact mechanism the theory specifies. And it tells you SIL will rhyme: it will floor at a somewhat higher absolute number, because SIL is place-bound, but it will cut a deeper fraction, because its three-thousand base is padded with hollow registrations the employment-services panel never carried.

Why there will be almost no resistance

The reflexive objection is that the Agency would never absorb the political cost of displacing thousands of community providers. The objection is already out of date, for three reasons that follow from the economics.

The first is that the Agency does not need to displace anyone. It controls allocation, which means it controls occupancy, which means it controls viability. A reallocation of exactly this kind is already in flight in the therapy space through Thriving Kids, which stands up a government intake-and-allocation function and pushes private operators to the margins. The Agency does not have to close a SIL provider. It can raise the regulatory and quality requirements on SIL while freezing new SIL allocations to the provider, which locks it out of growth and lets natural attrition do the rest. With no new participants flowing in to replace the ones who die, exit, or move, the houses bleed occupancy until the provider folds or sells. That is consolidation by attrition, achieved without a single headline-generating closure.

The second is that the constituency that protects incumbents in most markets is barely present here. The providers facing consolidation are disproportionately small operators with no electoral weight and no peak body of consequence, and after the Royal Commission the public reads provider closures as the system cleaning itself up. The participants have advocates. The providers, for the most part, do not.

The third reason is the one the sector finds hardest to say out loud. A provider contemplating resistance is contemplating it against the single buyer that will control its entire revenue, in every region, for the rest of its existence. You do not fight the monopsony you depend on across your whole book. The same portfolio logic that makes compliance the dominant strategy makes resistance suicidal. Finally, the commissioning process will almost certainly assess each organisation's absorption capacity and processes, so the NDIA will hold an exact map of who gets absorbed and who does the absorbing.

Why the providers who matter will want it

The resistance section is the negative case: the operators who might fight have no weight and no incentive to. The positive case is stronger, and it is the one the sector has not let itself say plainly. The providers who will still be standing actively want this, because commissioning kills the three things that are currently killing the honest, scaled provider. And on this point the Government and the surviving providers are aligned: the stated purpose of SIL commissioning, in the Government's own reform paper, is to address provider viability challenges. The reform is not being done to operators. On its central problem, it is being done for them.

Start with what destroys viability, because it is not price and it is not overhead. It is vacancy. Labour in a supported house is fixed to the house: the roster has to be staffed whether every bed is filled or not. Revenue, by contrast, is per participant. So a vacancy does not trim the margin, it detonates it.

Now see what the open market does with that fact today, and why it punishes the honest operator specifically. A compliant provider has to staff to the funded package; it cannot lawfully hold a participant on less support than the plan provides for. An operator willing to under-service is not bound by that. The margin that under-servicing frees up does not sit idle: it is recycled into marketing and into the deal-generation machinery that wins the next participant. So the compliant provider is not merely undercut on quality it cannot advertise. It is outspent on acquisition by a competitor funding that spend out of support it is not delivering. The field is not level. It is tilted by construction against the operator doing the right thing.

Commissioning and the enacted legislation hand the surviving provider three things it cannot get in the current market.

• Occupancy pooling. The bundled regional slot is not only the universal-service stick, it is the carrot. Vacancy risk that is lethal at the single-house level averages out across a regional book, and the buyer that controls allocation backfills the slot. The honest provider trades idiosyncratic, solvency-threatening per-house vacancy for pooled occupancy underwritten by the allocator. The single largest destroyer of viability in SIL is neutralised by the structure of the slot itself.
• A level cost field. Under-servicing stops being viable once it is detected by the one buyer and punished across a provider's whole portfolio, so the illicit margin that funded the cheat's acquisition advantage simply disappears. The compliant provider competes on equal terms for the first time.
• Clean demand. The conflicted referral channel is being closed by statute: a plan manager can no longer be, or be a related party of, a support provider, shared key personnel are barred, and a plan manager's related parties cannot provide other NDIS supports (sections 73E(2B), 73F(2)(j) and (k)). A new support coordination and connection function stands up from 2028. The demand-generation edge that rewarded conflicted relationships over delivery is being designed out.

On the genuinely difficult cases, the participants with real, profound, high-intensity need, the structure helps rather than hinders. It is far easier for the Agency to work through complex, safety-critical situations in detail with twenty to thirty organisations playing a repeated game it can hold to account, now backed by the investigatory and compulsion powers the bill provides, than with three thousand one-shot providers, many of whom have built their model around the very participants in question. Consolidation does not abandon the hard cases. It makes them governable, and it routes them to providers with the scale and the pooled occupancy to actually carry them.

What this means if you intend to survive it

If you operate in SIL and you plan to still be operating when this is done, the registration count was never your concern, and neither, in the end, was the price guide. The question that decides your future is whether you are a genuine delivering presence, with the scale, the geographic reach, and the claim integrity to sit inside the meaningful twenty to thirty, or whether you are one of the positions that consolidation will quietly clear away.

Three consequences follow, and none are optional. Scale has become a survival trait rather than a growth ambition, because the new normal sits several times above today's typical provider and the path to it runs through deliberate absorption and merger, not organic drift. Reach is the lever that determines who lives in the tail, so a provider's conscious extension of its footprint across districts is simultaneously how it secures its own position and how it pushes weaker operators out of theirs. And the business model itself has to change ahead of the structure, because the strategies that win in the current market, selective intake of well-funded participants, quiet under-servicing, and the steady inflation of need, are precisely the behaviours the new game is engineered to detect and punish across a provider's entire portfolio. The operator that builds its model on those behaviours is building toward the exact failure the reform is designed to produce.

It is arriving because a monopsony that cannot govern the fragmented tail of a commodity market has exactly one structural move available to it: to reshape that tail in the image of its own head.

The market that emerges from commissioning will be smaller, larger-bodied, and more concentrated than the one we have, and it is far closer than the sector wants to believe. It is not arriving because the Agency is cruel or capricious. It is arriving because a monopsony that cannot govern the fragmented tail of a commodity market, defected against on quality from one side and on cost from the other, has exactly one structural move available to it, until what remains is a panel of portfolio players the Agency can finally see, hold to standards, and trust to deliver the rest of the reform. The providers who understand that now, and who position against the constraint rather than against the politics, are the ones who will still be holding slots when the music stops.