For three years the Social, Community & Civic Participation line has been discussed in the language of policy intent, with plans said to be drifting from their purpose and funding said to be purchasing activity that mainstream community settings should provide, though none of that discussion attached a costed figure to the line. The consolidated response to Senate orders 503 to 508, tabled by the Minister for the NDIS in May 2026, attaches such a figure for the first time, and it is the largest of any measure in the package.
Ten measures are costed in the response. Reform item 2, the reset of Social, Community & Civic Participation and Capacity Building – Daily Activities budgets for all participants, reduces expenses by $13.2 billion over five years. It exceeds the Objective Test measure at $9.3 billion, and represents thirty-five percent of the $38.1 billion package.
The profile matters as much as the total. Nothing is booked in 2025-26, $1.1 billion in 2026-27 and $3.7 billion the year after, at which point the measure runs close to its mature annual value. The shape reflects the plan cycle: budgets reset as plans roll over, so the measure reaches roughly ninety percent of its mature value in its second full year. The practical consequence for a provider is that the adjustment window runs for roughly one year rather than the four the estimates span.
The same document contains a participant projection that has drawn less attention than the expenditure table. Against a pre-reform trajectory of 944,000 participants at 30 June 2031, the post-reform projection is 598,000, a gap of 346,000 people, or thirty-seven percent below trend.
The projection does not model slower intake. The relevant row is labelled "entered scheme before 1 Jan 28 and assumed to be exited," and counts 241,000 current participants modelled to leave. By 2029-30 the exited cohort is 205,000, roughly twenty-two percent of the participant base for that year. For a provider this is a volume reduction arriving in the same window as the reset, and unlike the reset it strips revenue off every line at once.
Unusually, the exposure does not need to be inferred. The NDIA's FOI release of the top-1,000 provider payments for 2025-26 itemises Social, Community & Civic Participation as its own column, so every provider's exposure to reform item 2 can be read directly. The analysis runs on the indirect list, which contains the organisations that actually deliver supports, covering 999 providers that represent $20.1 billion in payments.
Across that list, SCCP accounts for twenty percent of revenue and the median provider sits at nineteen percent. Once the assumed Capacity Building – Daily Activities share is added, the exposed base is twenty-three percent of sector revenue. The distribution has a long right tail: 283 providers earn more than thirty percent of revenue from the line being reset, and 98 earn more than half.
The composition runs against casual expectation. Sixty-three of those 98 are not-for-profits, and across the whole list not-for-profit providers draw 25.4 percent of revenue from the exposed line against 17.7 percent for for-profits. The community access and day-program estate is disproportionately the charitable estate, and it accordingly carries a disproportionate share of the reset.
A provider cannot decline the reset: the budget arrives smaller and the participant's need to access the community remains. The mitigation the reform documents plainly anticipate is to move one-to-one community access into group delivery.
The economics appear neutral by design. NDIS group pricing is constructed to be cost-neutral at ratio: one worker delivers to three participants, each billed at roughly a third of the one-to-one rate, so the provider's revenue per staff hour is unchanged while the participant's reduced budget still purchases supported hours. This is why conversion reads as the rational response, and why it will be attempted at scale.
It will not be attempted on everything, and the model does not assume it is. Behaviour support requirements, personal care embedded in the shift, participants for whom no compatible group exists within reach, and families who decline a group offer keep a material share one-to-one. The central case converts half of the exposed supports that remain funded after the reset. Conversion moves a provider's risk from the volume channel to the ratio channel, while declining to convert leaves it in the volume channel, and the model treats both channels as carrying cost.
The cost-neutrality of group pricing holds on one condition no provider fully controls: the group must run at its funded ratio. A one-to-three group delivered with two participants costs what a full group costs, because the worker is rostered, the vehicle booked and the venue paid regardless of attendance, while only two participants are billed. The session earns two-thirds of full-ratio revenue against an unchanged cost.
On the cost structure of section 06 the direct cost is eighty-two cents per dollar of full-ratio revenue, so a session at ratio contributes eighteen cents, and the same session one seat short contributes minus fifteen. The unfilled seat does not thin the session's margin; it converts the session from a contributor to overhead into a consumer of contribution earned elsewhere.
The mechanism will be familiar to readers of our SIL analysis. In supported independent living the roster is fixed against the property, and a vacant bed consumes the margin of the whole house; in community access the cost is fixed against the session, and an unfilled seat does the same to the run. The difference is observability: a vacant bed persists for months and appears in every occupancy report, while an unfilled seat lasts the length of a session, recurs on particular days, and is generally captured in no report at all.
The effect scales through a single factor. Where c is the share of exposed supports converted, u the share of sessions below ratio, s the seats short and x the funded ratio, staff hours required to earn a dollar of revenue rise by Λ = 1 + c × u × s ÷ (x − s). At central settings Λ is 1.038, an additional 3.1 cents of direct cost on every dollar of exposed revenue still delivered, which is a small figure until it is set against the margin available to absorb it.
Every prior treatment of the reset, including our own earlier drafts, ran on a single cost-stickiness figure. This paper replaces that with an explicit cost structure, because the composition of the stickiness matters more than its level. The modelled provider operates at a five percent EBIT margin on a ninety-five cent cost base: thirteen cents overhead, eighty-two cents direct delivery.
Within the adjustment window the two behave differently. Direct cost is largely rosterable, since a casualised workforce, consumables and venue bookings flex with delivered hours, leaving a sticky residue of notice periods and lease commitments that we set at ten percent. Overhead does not flex: premises, corporate salaries, systems, insurance and audit run to their own timetable, and we treat seventy percent as fixed within the window.
The composite is that 18.2 percent of any revenue reduction is not matched by a cost reduction in the modelled year, which is gentler than the twenty-five percent stress assumption in earlier work and now rests on named components rather than a single assertion. Composition matters because of duration: the direct component clears as rosters catch up, but the fixed-overhead component persists until the provider restructures the overhead itself. At the central case the fixed-overhead stack, 9.1 percent of baseline revenue, becomes 11.5 percent of the base surviving to 2028-29 and 13.3 percent by 2030-31.
A provider can fill every group it runs and still lose money, through an unchanged overhead bill divided by a smaller denominator.
Taken singly, none of the three reductions is fatal. Participant attrition against a fully variable cost base would be margin-neutral; the damage is done entirely by the cost structure, and on that structure attrition takes the five percent margin to 2.1 percent, a level that is uncomfortable but survivable. The budget reset takes the sector to the edge at 0.5 percent. Group conversion at half converted and fifteen percent of sessions one seat short then removes what is left.
The modelled sector margin at 2028-29 is zero to within rounding, and sector EBIT moves from $988 million to minus $6 million, a swing of $994 million on a $19.8 billion revenue base. At settings chosen to be plausible rather than adverse, the reform consumes approximately one hundred percent of the operating profit of the community access sector by its second full year.
The year the snapshot is taken matters as much as the settings. The reset is close to fully phased by 2027-28, while the participant exits keep compounding to 2030-31, so the position deteriorates after the year most providers will conclude they have absorbed the change. A provider entering at the sector's five percent holds its head above water through 2028-29 with nothing to spare, cannot through 2029-30, and by 2030-31 needs a starting margin above eight percent to remain profitable without changing how it delivers.
The central case sits at the fifteen percent under-fill row and the thirty-five percent reduction column. The table's message is less any single cell than the gradient: every step of under-fill costs about a quarter-point of sector margin, and no plausible combination of the two levers returns the sector to its starting position.
A sector average conceals the distribution, and here the distribution is the operative result. At the 2028-29 central case, 436 of the 995 modelled providers trade below zero, holding $7.4 billion of revenue, or thirty-seven percent of the sector. 559 remain profitable, 375 record a marginal loss recoverable with hard operational work, and 61 land between minus five and minus fifteen percent, which requires structural change. At 2028-29 the sector is cut roughly down the middle rather than destroyed, and the variable that sorts providers is their modelled exposure to the line being reset.
The ownership split is starker than the margin gap suggests. Not-for-profits land at minus 0.5 percent against plus 0.1 percent for for-profits, and expressed as a headcount, sixty-one percent of modelled not-for-profits trade below zero against thirty-five percent of for-profits. The result inverts the standard assumption that the charitable estate is the resilient part of the sector.
Read structurally rather than fiscally, reform item 2 operates as a consolidation instrument, and it selects on two variables. The first is ratio-completeness, the share of sessions that run at their funded ratio, which is in turn a function of participant density within a catchment. A provider with forty participants inside a postcode can fill a one-to-three group on a Tuesday morning; a provider with twelve cannot, will run it at two, and loses money on every session. The reset therefore sorts providers by catchment density rather than by their cost base.
The second is overhead structure. The capacity to escape the fixed-overhead consumption is a function of scale: a large provider can restructure a corporate layer and spread what remains across a broader base, while a sub-scale provider carries a registrar, an audit, a payroll system and a lease regardless of how few participants remain. The participant attrition sharpens both effects rather than softening them, and the two measures work in the same direction, which is a reasonable indication that the direction is intended.
The endpoint is the one our SIL analysis reaches by a different route. In SIL, consolidation arrives through bundled regional slots and competitive tender; in community access it arrives through session arithmetic and overhead recovery. A provider that cannot fill its groups or carry its overhead is not removed by the Agency; it becomes unable to trade, and consolidation of that kind requires no policy instrument at all.
The measure lands lightly in 2026-27 and at roughly ninety percent of its full weight in 2027-28, so the window to build the capability that determines a provider's position is the first year, not the second. Four quantities determine that position, and most providers currently know none of them with precision: exposure by service line and site; ratio-completeness, the funded-versus-actual ratio of every group support measured as a distribution by day and site; density, participants per catchment and the compatibility matrix that determines who can be grouped with whom; and overhead tolerance, the revenue decline the current cost base can absorb before the result crosses zero.
On the sector settings modelled here that tolerance is roughly a third of revenue, and the central case consumes most of it by 2030-31. A provider that has measured its own number knows whether its overhead has to be restructured before the measure matures; a provider that has not will discover the answer in its accounts.
The providers that remain viable will be the ones that measured their ratio-completeness and their overhead tolerance before the measure reached full weight, rather than the ones that cut hardest after it did.
Occasional, considered analysis on SIL economics, NDIS reform, and provider viability. And you can unsubscribe anytime.