NSFAS's own lawyers say it cannot set your rate
Retha van Rooyen
Issue 02 ยท Tuesday 1 September 2026
2,706 words, about 11 minutes
IN THIS ISSUE
- What eleven presenters told Parliament about your cap on Friday, and the one thing they all agreed on without noticing
- The legal opinion that means the cap can never again be a number announced from a podium
- What the cap should actually be: a defensible corridor, computed from measured numbers, not a wish, and the test that showed my arithmetic and the developers' arithmetic agree to within 0.86%
- The deduction hiding inside the rate ladder that looks exactly like a downgrade, and how to check your own payments for it
The bottom line. Parliament's student accommodation colloquium on Friday produced a rare thing: the department, the universities, the students, the providers and the property regulator all asked for versions of the same cap, evidence-based, differentiated by what a bed is and where it stands, adjusted annually. NSFAS's own legal opinion then made that the only lawful road: it probably has no power to set rates, only to build a defensible allowance. I have spent the past month building exactly that arithmetic from measured numbers, and this issue walks you through it. The short answer: the level of the metro cap is roughly right, the plumbing that delivers it is not, and the new-build crisis needs a second instrument, not a bigger cap.
1. WHAT MOVED
Parliament held a colloquium on your cap, and convergence broke out
On Friday 28 August the Portfolio Committee on Higher Education gave a full day to student accommodation. I watched the presentations. The department's DDG said the accommodation cap "should be reconceptualised, taking into account the realities of provision and costs, and providing graded costs for different accommodation provision." Universities South Africa said a single national cap is "neither suitable nor realistic" and asked for regional differentiation, a catered and self-catering split, annual inflation adjustment and a local exception mechanism. The students' union, SAUS, did not ask for the cap to be scrapped: it proposed a differentiated tariff with bands by location, quality and services, and a hard rule that funded students never pay the gap. The PPRA, appearing in this sector for the first time I can find, asked for an evidence-based review recognising regional cost differences. Read those four positions next to each other and they are one position, in four vocabularies.
NSFAS told Parliament it probably cannot set rates at all
The quiet bombshell came from NSFAS's own legal review. Counsel's opinion: the NSFAS Act empowers the scheme to set conditions for accreditation, to decide where funded students may live, and to determine the allowance it pays, but it probably has no power to prescribe what a provider charges. Rate-setting by a public body needs an express statutory power, the way NERSA has one for electricity tariffs, and the NSFAS Act contains none. The consequence is bigger than it sounds: the cap cannot lawfully be a price control. It can only be a funding formula, a published method that produces a defensible allowance. The rest of this issue is about what that formula should contain.
The providers put numbers and a demand on the table
The Private Student Housing Association showed Parliament a costed viability gap: a compliant new-build bed costs about R326,300 to create and needs roughly R65,000 a year against the R52,000 annual cap, a shortfall of about 25%, and new-bed delivery has collapsed from a 17,489-bed peak in 2023 to under 4,000 a year. SANSAA, a federation whose affiliates house over 300,000 students, showed a pilot in Mthatha where stacked deductions took 40% of revenue, said plainly that "the cap is never reliably paid", and supported a revised cap of R66,000. Handle with care: both presentations are the providers' own models and I have not audited them, but the deduction mechanics they describe match what I measure in my own payment records, and I will show you that measurement below.
The state put its accommodation bill on the record, and the police on notice
Two more things Friday gave us. First, NSFAS disclosed its full accommodation spend for the first time in one place: R18.83 billion for university students and R3.52 billion for TVET students in 2026, R22.4 billion in total, up from R15.6 billion in 2023. Second, the department called for an urgent investigation into accreditation and costing practices "with the help of law enforcement agencies", and the Competition Commission confirmed it is investigating a DHET complaint alleging providers charge suspiciously similar rates.
Note the irony for later: when everyone prices at an administered cap, identical pricing is what the system was designed to produce.
2. WHAT SHOULD THE CAP BE
Three terms, before we go further. The cap is the ceiling NSFAS pays for a bed each month, R5,200 for a single room on the 2025 card, which is still the operative card because the 2026 rates remain unannounced in September. The net is what actually lands in the landlord's account after the accreditation partner's 5% and any institutional deductions. The cost to serve is what it takes to run an occupied bed for a month, and it runs for twelve months while the cap pays for ten.
Everyone at Friday's colloquium argued about the cap with a position. Almost nobody argued with a method. Here is the method, and it is short enough to say in one sentence: a defensible cap covers twelve months of measured operating cost plus a fair return on the capital actually invested in the bed, delivered across the ten months the funding year really pays, with an allowance for vacancy and collection loss.
Feed measured numbers in. My cost to serve, measured across eight properties where the owner's and the manager's books are both visible, is R1,622 a bed a month, full economic basis. Management at the market rate adds R494. Capital at what existing student stock actually trades for is about R232,000 a bed, all in. A required return comes from the property market itself: Rode's grade-A band runs 7.3% to 8.2%. The arithmetic then gives you a floor of about R4,428 a month at the bottom of that band, a sustainable point of about R4,937, and a full 2026 metro corridor of roughly R4,900 to R5,400 once the parameters are allowed their honest ranges. The current R5,200 cap sits inside that corridor, and at the cap the implied unlevered return on an acquired bed is about 10.5%, which is comfortably investable.
You are entitled to ask why you should trust my arithmetic. Here is the test I ran after the colloquium. The providers gave Parliament their own costing of a new-build bed, built independently of anything I have done: R2,300 a month to operate including VAT, R326,300 of capital a bed, a 10.5% hurdle, ten months of funding against twelve of cost. I fed their four numbers into my formula instead of mine. It returns R64,709 a year. They told Parliament the requirement is R65,273. The two differ by 0.86%, and neither model had seen the other.
That matters more than any single number in this issue. The same arithmetic, applied by an operator to its own books and by a developers' association to its own build costs, lands within one percent. Which means the argument between operators and developers is not an argument about method at all. It is an argument about which bed is being priced.
So on the level, the uncomfortable finding for our own side: the metro cap is not too low for the stock that exists. The R66,000 put to Parliament is a number without a published method, and against measured costs on existing converted stock it prices a windfall. If we ask Treasury for it we will lose, and deserve to.
What is broken is everything between the cap and the bank account. The net ranges from R4,388 to R5,200 against one cap, depending on the institution. The clip and the carve-outs take the modal payment to about R4,080, which is 21% below the headline, and at that effective rate the implied return falls under 6%, below a bank deposit with none of the work. The funding year pays ten months against twelve months of cost, which is a hidden 17% discount no other administered price carries. And the payment itself arrives late and short often enough that both provider bodies told Parliament reliability, not level, is the crisis. My records agree: this year one university's payment runs have arrived 50% to 73% unpaid at run date. Unreliability has a price, and it is worth knowing: a counterparty that leaves half a billing run unpaid adds two to three hundred basis points to the return an investor demands, which through the same formula is worth R500 to R700 a month on the cap. It is materially cheaper for the state to pay reliably than to compensate us for the fact that it does not.
Then there is the one place the cap genuinely cannot reach: new build. The providers' R326,300-a-bed compliant tower and my R232,000-a-bed converted house cannot be priced by the same ceiling, and raising the cap until towers pencil hands roughly R127 million a year per R100 of increase to every existing bed that never needed it.
If you operate outside the metros, this gets worse. The non-metro rate sits 18% below the metro rate, but your costs do not. Electricity is a national price. So are insurance and building materials. Wages fall only partly. The non-metro discount is a revenue discount without a matching cost discount, and it works only where property is genuinely cheap enough to compensate. Run the numbers on a new non-metro build and it needs about R5,275 a month against an applicable rate of R4,050.80: short by R1,224 a bed a month, and it does not close even if you allow operating costs 10% lower than mine. A discount meant to reflect cheaper property is, in effect, a construction ban in exactly the TVET and rural catchments where the shortage is deepest and where there is the least existing stock to convert. The answer Parliament circled without landing is two instruments: an evidence-based operating price for stock in service, and a separate capital instrument, a grant or an availability payment, for verified new supply. The providers' own concession proposal on Friday is that second instrument wearing different clothes. There is a deal available here.
Last, the money. Any reform pitch that needs new billions in a flat budget is dead on arrival, so fund it from the leak: NSFAS confirmed to Parliament last month that 62,958 funded students sit in unaccredited accommodation, roughly R155 million a month for beds nobody inspected. Recover even half of that and the plumbing fixes fund themselves.
And there is a clock on this. The cost of servicing a bed has risen about 8.6% a year for five years while the ceiling has barely moved. Project that forward and the operating floor reaches roughly R3,700 by 2030. Against a frozen cap, the formula prices the capital component below what the market requires somewhere around 2028 to 2029, and at that point it is not new entrants staying away, it is existing owners leaving. The supply side is already showing it: purpose-built delivery peaked at 17,489 beds in 2023, the year the caps came in, and has run near 4,000 a year since. The standard was raised in October 2022 and the revenue was fixed in February 2023. That is the mechanism, and the delivery series is the result.
What I cannot tell you yet. Whether the department's "reconceptualised, graded" cap will be a published formula or another announced number; what the 2026 rates will be, now a year overdue; how the Competition Commission's price-fixing investigation resolves the awkward fact that an administered cap manufactures identical prices; and where SANSAA's R66,000 comes from, which I have asked about. All four belong to the next issue or the one after.
My recommendation. If you hold accredited stock, underwrite at the net, never the cap: R4,869 student-weighted is the honest revenue line on a metro single, and R4,298 if your register grades like the Cape's. Model twelve months of cost against ten of income. And put your voice behind the formula, not behind a bigger number: the formula is the version of this fight the evidence lets us win.
The full formal treatment, the identity, the corridor tables, the four-economies matrix and the stakeholder-by-stakeholder case, is in the MSH cap position paper, that you can download below. It is written to be handed to an MP, a banker or a journalist as it stands.
3. ONE THING WORTH FIXING
Reconcile every NSFAS payment to the rate ladder, because the ladder hides a deduction in plain sight.
The published rates are an exact geometric series: each grade is precisely 95% of the one above, R4,940 to R4,693 to R4,458.35 and so on down, metro and non-metro alike. Which means when the standard 5% accreditation fee comes off a payment, the amount that lands is exactly the next grade's rate.
In our own August reconciliation, 77.5% of matched bed-months looked as if the properties had been quietly downgraded a full grade. They had not. The fee had. Sample-weighted it runs R181.76 a bed a month, which on a hundred beds is R218,112 a year that never appears as a line on any statement.
Until we reconciled payment-by-payment against the ladder, a material deduction was structurally invisible, and the same arithmetic will be hiding in your statements.
One afternoon with a spreadsheet: list each payment, divide by months claimed, add back any transport carve-out, and match it to a rung. Anything that lands between rungs is a question for the institution.
4. WHAT IS NEXT
The colloquium record publishes on the parliamentary website, date unconfirmed. Why it matters: the presentations become citable primary sources, and the committee's follow-up questions will show which proposals grew legs. I will read all of it so you do not have to.
The 2026 accommodation rates, overdue since February. Why it matters: every lease you sign for 2027 intake is priced against a card that is now two years old. The moment they land, the Insider runs the new arithmetic.
DHET's revised Minimum Norms and Standards, cleared for gazetting on 13 August. Why it matters: the standard defines the cost the cap must fund; if the standard rises and the cap does not, the viability gap the providers showed Parliament widens by decree.
September to November: the payment squeeze window. Why it matters: NSFAS's own mid-year figures imply the monthly run rate must more than double to meet the university allocation. Provision for slow cycles and partial payments in the fourth quarter, and do not confuse a timing gap with a bad tenant.
The MSH quarterly Pulse, November. Why it matters: first tracked update of the Monitor's indices, with the rate announcement, if it has come, priced in.
5. FILED AWAY
Presentations to the Portfolio Committee on Higher Education's Student Accommodation Colloquium, Parliament, 28 August 2026:
- DHET (T Lewin);
- Department of Human Settlements (N Buthelezi);
- NSFAS (H Valjee; legal review by ENS presented by W Carrim);
- Universities South Africa;
- SAPCO college submissions;
- SAUS (M Nkambako);
- SATVETSA (K Sonti);
- Competition Commission (A Gwabeni);
- PPRA;
- PSHA (K Mamabolo);
- SANSAA (D Monks, S Jennings, J King).
- MSH Student Accommodation Monitor 2026, published 27 August 2026, chapters 5 and 8 for the payment and cost measurements.
- MSH cap position paper, August 2026, attached.
- NSFAS Annexure B grading levels, 21 February 2025.
- NSFAS unaccredited-accommodation reply, NCOP, 13 August 2026. Parliamentary reply NW874, 4 May 2026.
- Rode Report on the SA Property Market, 2026:2.
That is a wrap for the month.
If something in here affects one of your properties and you want to talk it through, send it to [email protected]. What you send shapes what I go and find out next month.
Retha
ALSO FROM ME
- The MSH Student Accommodation Monitor 2026, free. Get the Monitor
- The book, new edition. Get a copy
- NSFAS Accreditation Masterclass. See the masterclass
- One hour on one property, your numbers, a direct answer. Book a consultation
- The property scorecard, free, seven questions, about two minutes. Grade a property
The Student Property Insider
Monthly intelligence for people who own student beds. What moved in student accommodation, what it took to absorb on the ground, and what is coming that belongs in your diary. Policy, funding and compliance read properly, then translated into what it means for a property.
Responses