The South Australian commencement anomaly

International Higher Degree by Research commencements in South Australia jumped roughly tenfold in February 2026 — from 143 to 1,466. On the surface it looks like an extraordinary recruitment win. It appears to be the opposite: most likely a statistical artifact of the Adelaide University merger, recorded at the very moment genuine new-student arrivals were falling.

The anomaly

Adelaide University — the merger of the University of Adelaide and the University of South Australia — began operating on 1 January 2026. A commencementin the Department of Education’s data is triggered whenever a student starts or changes a programme. Migrating every continuing student into a new Adelaide University programme code is, mechanically, a programme change. South Australia’s research cohort shows it most violently, because doctoral candidatures run for years, so a large stock of continuing candidates was re-commenced all at once.

What the official methodology says

The Department does not flag the merger anywhere in the release, but its own counting rules predict exactly this outcome. From the explanatory notes and the monthly data-tables page:

“A commencement is an enrolment that commenced at any time within the year it appears in.”

“Enrolment numbers is a count of a student’s enrolment within the reference period. If a student does multiple courses during the reference period, the enrolments will each be counted.”

The data is built from Confirmations of Enrolment (CoEs) in PRISMS, not from people. Transferring ~56,000 continuing students to a new CRICOS provider on 1 January issued each of them a new CoE: a new CoE starting in 2026 is, by definition, a commencement — and because the legacy CoE was also active at the start of the year, both enrolments would fall inside the year-to-date window and, on our reading of the counting rule, each be counted.

It isn’t new students

Splitting commencements by whether the student was already in Australia is decisive. Of the 1,466 research commencements in February 2026, 1,323 (90%) were students already onshore— continuing candidates, not arrivals. New research arrivals stayed flat (70 → 143). Continuing students don’t need a fresh visa, which is exactly why the visa data can’t — and doesn’t — support the rise.

It pervades every level — but hides in the totals

Here is the subtle part. The re-commencement of continuing students happened across every level — onshore commencements rose +133% for coursework masters and +39% for bachelors, not just in research. Yet the headline totals for those levels barely moved (Masters Coursework +4%, Bachelor −3%). The reason: for the large coursework levels, the surge in continuing re-commencements lands at the same time as a genuine ~45–50% collapse in newinternational arrivals, and the two roughly cancel. Doctoral is the only level where the artifact breaks the surface, because its continuing stock dwarfs a normal year’s ~110 commencements.

Continuing = already in Australia (new_to_australia = No); new arrivals = new_to_australia = Yes. South Australia, Higher Education, February year-to-date. Counts under 5 are perturbed per the de-identification rule.

The enrolments doubled too — in lockstep with the commencements

A fair objection to the programme-change explanation: if the surge were just continuing students being re-badged, the enrolment count should be stable. It is not. SA research enrolments jumped from 1,424 to 3,021 (doctoral alone: 1,312 → 2,713) — a clean doubling. That is not a second mystery; it appears to be the same effect seen from the enrolment side. On our reading of the counting rule above, a transferred student would carry twoenrolments through the early-2026 window — the legacy one and the new Adelaide University one — which, under the rule that “the enrolments will each be counted,” would both land in the total. We are confirming this treatment with the Department.

The field-of-education split is decisive. The new Adelaide University CoEs were registered under a generic “Mixed Field Programmes” classification — and that single label absorbs the entire increase, exploding from 12 to 1,403, while every genuine discipline is flat. The legacy enrolments (real fields, ~1,600) plus the new ones (mixed field, ~1,400) are the same cohort listed twice.

SA research (Doctoral + Masters by Research) enrolments, February YTD. The new Adelaide University CoEs sit in “Mixed Field Programmes”; legacy CoEs retain their real field.

Two corroborations. First, every major nationality doubled uniformly — China 2.28×, India 2.31×, Iran 2.09×, Bangladesh 2.41×, Sri Lanka 2.03×, Vietnam 2.00×. Real demand shifts never replicate the existing stock at a constant multiple; duplication does. Second, continuing-student enrolments went 1,354 → 2,878 (2.13×) while genuinely new arrivals went 70 → 143 — the doubling sits entirely on the cohort that was transferred.

The visa cross-check

If 1,466 research students had genuinely commenced, South Australia would have needed a comparable surge in postgraduate-research student visa grants. There was none. Grants have sat at roughly 450–1,000 a year for a decade, and the most recent full year was ~840 — nowhere near the commencement figure. The gap is the continuing students, who were already here.

How much does it skew the sector totals?

The artifact is one state. Every other state’s continuing-commencement count moved normally in 2026; South Australia’s leapt by ~2,400. That gap is the merger.

Taking the abnormal South Australian onshore gain as the artifact (~2,300–2,450 commencements; point estimate ~2,400), the effect on the headline sector figures is:

Other sectors (VET, Schools, ELICOS) are unaffected — the merger involves two universities, so the distortion is confined to Higher Education. Estimate range ~2,250–2,650; the conservative figure is used.

What the coverage says

The press has reported the real-world side of this — and, in one case, the data distortion itself:

• InDaily — “Adelaide Uni in red as international student numbers plummet” (2 June 2026)

  New international commencements ran ~40% below target with a projected $90m shortfall. The Vice-Chancellor noted the university “performed well” retaining existinginternational students — the problem was recruiting new ones. That retained-vs-new distinction is precisely the split that produces the statistical anomaly.

• The PIE News — “2026 NOSC allocations rise for Australian public universities” (14 October 2025)

  Nous Group’s Nicholas Dillon: the merger “accounts for most of the increase in Go8 NOSC” — strip it out and Group-of-Eight allocation growth falls from 9% to 4%, making the rising share “somewhat misleading.” The clearest acknowledgment that merger arithmetic is bending the published numbers.

• InDaily — “Premier officially opens merged uni” (29 January 2026)

  Confirms the go-live: continuing students of both legacy universities automatically became Adelaide University students in 2026 — the event that re-commenced the cohort.

What has not been reported is the specific mechanism here: that the official international commencementstatistic for South Australia surged at the same time, driven entirely by continuing students being re-commenced. The media has the “new students are crashing” story; this case study supplies the reconciling other half.

The takeaway

Any year-on-year read of South Australian international commencements or enrolments across the 2025→2026 boundary — at any level — is best treated as a structural break rather than a trend. Commencements appear inflated because re-issued CoEs register as new starts; enrolments appear inflated because a transferred student would carry two countable enrolments through the window. To recover the real new-student signal, filter to new arrivals (new_to_australia = Yes); on that basis South Australia’s new international students fell, in line with the national decline and tighter visa settings. Research is simply where the distortion is impossible to miss.

What to watch. The December 2026 YTD release will still carry this effect (both enrolments were active during 2026). Expect the series to snap back to roughly 1,500–1,700 research enrolments at February 2027, once only the Adelaide University CoE remains — and for “Mixed Field Programmes” to linger as SA’s dominant research field label until the CRICOS coding is corrected. The clean arbiter will be the federal student load (EFTSL) release for 2026: study load does not double when a student is re-enrolled, so flat overseas EFTSL will confirm the artifact beyond argument.