She has the job, the pay rise and the lanyard, and the qualification is still unfinished. In early 2024 a woman in her early forties, living in a regional Victorian town, enrolled in a Fee-Free TAFE place in a Certificate III in Individual Support — the entry-level qualification for aged care and disability support work. She had thought about it twice before and stopped at the same point both times: the tuition fee, payable at the front end, in a household budget that had no room at the front end of anything. The fee-free place removed that obstacle in a single stroke. She enrolled within a fortnight of hearing the course was free.
Ten months later she had done the units that an employer cares about most: manual handling, personal care, infection control, and her supervised placement hours. A residential aged care provider forty minutes up the highway offered her shift work at a rate well above what she had been earning in retail. She took it. The remaining units — the ones her employer does not require and her state's regulator does not mandate for the role she now holds — sit unfinished. She has not formally withdrawn. She has not been back to a classroom in a year.
In the national dataset, she is not a completion.
Consider her alongside a nineteen-year-old who took a fee-free pre-apprenticeship course in construction and left it halfway through because a builder on a nearby estate offered him a start on the tools. He is not a completion either. Both are better off than they were. Both are counted, in the headline arithmetic, as a place filled and a qualification not obtained.
This article is not an argument that Fee-Free TAFE failed. On the evidence available in September 2026, that claim cannot honestly be made, and this site will not make claims it cannot support. The argument here is narrower, and because it is narrower it is harder to dismiss. It is about what the country measures, what it publicises, and when. Enrolments are an input. They are counted quickly, reported quarterly, and promoted heavily. Completions are the output that carries the economic value, and they are reported far less prominently and years later. And completions in the specific occupations that Australia has formally identified as being in shortage — the actual point of the exercise — are barely reported to the public at all.
What the government did
Fee-Free TAFE began at the Jobs and Skills Summit in Canberra in September 2022. The Commonwealth and the states and territories agreed a twelve-month National Skills Agreement providing 180,000 fee-free places in vocational education and training during 2023, jointly funded, with the states responsible for delivery through TAFEs and, in some jurisdictions, other approved public and community providers (Department of Employment and Workplace Relations; National Skills Agreement documentation, 2022–23).
"Fee-free" is a precise term and worth defining at first use, because it is narrower than the phrase sounds. It means the tuition fee normally charged for the course is waived. Depending on the jurisdiction, the provider and the qualification, a student may still face costs for materials, tools, uniforms, textbooks, amenities or, in some trades, safety equipment. The barrier removed is the largest and most visible one, not every one.
The first year ran ahead of its target. Enrolments in 2023 exceeded the 180,000 places the agreement had funded, which the government reported as evidence of demand and which critics read as evidence the target had been set low (DEWR reporting, 2023–24). The program was then extended. From 1 January 2024, a new five-year National Skills Agreement between the Commonwealth and the states and territories took effect, worth in the order of $12.6 billion (approx.) in Commonwealth funding to the jurisdictions across its term, and including a further tranche of around 300,000 fee-free places over 2024 to 2026 (National Skills Agreement, 2024–2029; Budget papers, 2023–24 and 2024–25).
In 2025 the Parliament legislated the arrangement into ongoing form. The Free TAFE legislation established a continuing minimum of 100,000 fee-free places a year from 1 January 2027 (Parliament of Australia; Free TAFE Act, passed 2025). That is a significant change in the character of the policy. A one-year agreement is a stimulus measure. A legislated annual floor is a permanent line item, and permanent line items deserve permanent accountability mechanisms. Whether one was built alongside it is the question this article ends on.
Eligibility has been targeted rather than universal. Priority groups across the agreements have included First Nations Australians, young people broadly in the 17-to-24 bracket, people out of work and on income support, unpaid carers, women moving into fields where they have been under-represented, people with disability, and certain visa holders. Priority course areas have tracked the government's stated shortage list: care and support work including aged care, disability and early childhood; technology and digital; construction; agriculture; hospitality; sovereign capability; and the clean energy transition.
The stated rationale was never mainly about cost of living, although the saving to a student is real and in many cases runs into the thousands of dollars. The rationale was skills supply. Australia has a formal instrument for identifying which occupations are short: the Skills Priority List, produced annually by Jobs and Skills Australia, which rates occupations at the national and state level as in shortage or not, with a supporting assessment of future demand. Care occupations, construction trades, and a long list of technical and engineering roles have appeared on that list repeatedly. Fee-Free TAFE was presented, correctly, as a supply-side response to a demand problem that had been measured and published by a Commonwealth statutory body.
That framing is what makes the completion question a legitimate one rather than a pedantic one. If the policy had been sold purely as cost-of-living relief for students, the number of people helped would be a reasonable measure of success. It was sold as a fix for a labour shortage. A labour shortage is not fixed by an enrolment. It is fixed by a qualified person turning up to work in the occupation that was short.
The strongest case for the policy
The completion clock has not run out, and the participation gain is real on its own terms
Put the government's defence at its best and it is strong enough that a fair critic has to concede most of it before saying anything else — and the strongest single point is a measurement point, not a political one. The National Centre for Vocational Education Research, the Commonwealth's own statistical authority for vocational education, reports actual qualification completions four years after commencement. That convention exists because vocational students routinely take longer than the nominal course duration: they study part-time around work and caring responsibilities, they defer, they suspend and resume, they change providers, and they finish on their own timetable rather than an academic calendar. Measuring completions against enrolments before a cohort has had its full four years does not produce a low completion rate. It produces an incomplete count, and an incomplete count read as a rate systematically understates the truth. Judging students who commenced in 2023, 2024 and 2025 by what they had finished by early 2026 is measuring a race that most of the field is still running.
Second, the participation gain is not in dispute and it is not trivial. Departmental reporting shows enrolment increases concentrated among precisely the cohorts the policy set out to reach: younger students, First Nations learners, people with disability, and people who were out of work when they enrolled (DEWR, Fee-Free TAFE reporting, to the March quarter, 2026). Those are the groups for whom an up-front tuition fee is not an inconvenience but a decisive obstacle. Widening access to vocational education for people who would otherwise not have enrolled is a legitimate objective of government in its own right, defensible independently of what any completion statistic eventually says. A country that only funds education for the people already most likely to finish it is not running a skills policy. It is running a filter.
Third, the arithmetic of who was attracted works against the headline number in a way that is a feature, not a fault. Students who are younger, who have a disability, who are managing income support and unstable work, or who are the first in their family to undertake post-school study, complete at lower rates and take longer to complete than the general student population. That pattern is long-established in the vocational education literature and it is not a judgment on those students; it is a description of the additional obstacles they carry. A policy deliberately designed to bring more of those students into the system will, by construction, report a lower average completion rate than a policy that did not. A lower headline rate is therefore partly evidence that the policy reached its target population — which was the point.
Fourth, partial completion is frequently a rational and economically valuable outcome rather than a failure. A large share of vocational students enrol intending to acquire specific competencies, licences or units required for a particular job, not to obtain a full qualification. A student who gains a high-risk work licence, a food safety competency, an infection control skill set or a Certificate III's core placement hours and then walks into paid employment has not dropped out in any meaningful sense. They obtained the thing they came for. The system's headline metric records that as a non-completion because the metric was designed around qualifications, not around the reasons people actually enrol.
Finally, the policy answered a real and measured problem. Jobs and Skills Australia's Skills Priority List documented persistent shortages across care, construction and technical occupations. Australia's vocational sector had been through a decade of contested funding arrangements and a Productivity Commission review that found the previous national skills agreement was not doing its job. Acting on a documented shortage, quickly, with a mechanism that demonstrably moved enrolment volumes, is a defensible use of public money. The reasonable position is not that the policy was wrong. It is that we do not yet know how well it worked, and we will not know for some years.
The evidence
Two numbers sit at the centre of this topic, and almost every public argument about Fee-Free TAFE misuses at least one of them.
The first is enrolments: approximately 814,000 Fee-Free TAFE enrolments since the program began in 2023, as at the March quarter, 2026 (Department of Employment and Workplace Relations and NCVER reporting). The second is completions: approximately 258,000 as at the same period. Divide the second by the first and you get roughly 32 per cent. That single division is where the public debate usually goes wrong, and the next section explains why in detail, because it is the analytical heart of this topic.
Before that, some care about what the enrolment figure is. It counts enrolments, not people: a student who enrols in one course in 2023 and another in 2025 appears twice. That is not a trick, and the Commonwealth has generally been explicit about it — it is how vocational activity is conventionally counted. But it means the phrase "814,000 Australians have taken up a fee-free place" is not quite right. The number of distinct individuals is lower, by an amount not published prominently enough for a reader to work out unaided.
The four-year rule, and why it decides this argument
NCVER's standard practice is to report actual qualification completions four years after commencement. This is not an administrative quirk or a delay caused by slow data processing. It is a deliberate methodological choice, and understanding why it exists changes how every completion number in this field should be read.
Vocational education does not run on a fixed academic clock. A Certificate III might have a nominal duration of twelve months full-time, but a substantial share of the students undertaking it are not full-time. They are working, or caring, or on income support with mutual obligation requirements, or living somewhere with limited course availability that forces them into a mixed-mode or slower delivery pattern. They suspend enrolment when a shift roster changes. They resume the following year. They transfer providers and carry credit across. They finish a qualification in two years, or three, or four. NCVER's four-year window exists because that is roughly how long it takes for a cohort's completions to substantially stop arriving.
Apply that convention to the Fee-Free TAFE population. The program started in January 2023. The four-year window for the very first cohort therefore does not close until the end of 2026. For students who commenced in 2024, it closes at the end of 2027. For 2025 commencers, the end of 2028. For anyone enrolling under the legislated ongoing places from January 2027, it closes at the end of 2030.
This means that the approximately 814,000 enrolments in the denominator of that 32 per cent calculation include very large numbers of students whose completions cannot yet have been recorded, because their courses are not yet over and their window is still open. The 258,000 completions in the numerator are real. The denominator is not the matched denominator for them. Dividing a partial numerator by a full denominator produces a number that looks like a completion rate, behaves like a completion rate in a headline, and is not one.
The 32 per cent figure is a snapshot of an unfinished race. Comparing it to a mature completion rate is a category error, and anyone who does it is arguing badly.
Say the consequence plainly, because it constrains what this article is permitted to conclude. The approximately 32 per cent figure, as at the March quarter, 2026, is a partial-cohort snapshot. It is not the completion rate of Fee-Free TAFE. It cannot be compared to the completion rate of any mature cohort, in this program or any other, and it cannot be compared to a university completion rate, which is calculated over a different window on a different population under different rules. A commentator who sets 32 per cent beside a mature figure and calls the gap a scandal has made an error of arithmetic, not a discovery. Readers should be sceptical of anyone who does it, in either direction.
It also cuts the other way, and this is the part the government's defenders skip. The four-year convention is a reason to withhold judgment. It is not a reason to withhold data. There is a great deal that could be reported now, in a form that is methodologically clean, and is not being reported prominently: completions matched to their own commencement cohort rather than to the cumulative program total; the share of each cohort still active versus lapsed; qualification-level completions broken down by field of education; and, above all, completions mapped against the Skills Priority List. None of that requires waiting four years. It requires deciding to publish it.
The other reason the headline number is low: vocational completion has always been low
Even once every cohort's four years have elapsed, nobody should expect a vocational completion figure to look like a university one. Completion rates in Australian vocational education have historically been low across the board, for structural reasons that long predate this policy, this government, and this decade. This must be stated clearly before any critique is made, because a critic who treats a long-standing structural feature of the sector as though it were caused by a 2023 policy decision is not analysing; they are scoring points.
The largest structural reason is that a great many people enrol in vocational education without intending to complete a full qualification. The system is built to permit this. Vocational training is organised into units of competency, and units can be undertaken, assessed and recognised individually. Employers and regulators frequently require specific units rather than whole certificates: a licence, a ticket, a safety competency, a compliance refresher. Someone who enrols to obtain exactly those units, obtains them, and leaves has used the system precisely as designed. The completion statistic records them as a loss.
Beyond that, the vocational student population differs from the university population in ways that make continuous study harder. Vocational students are older on average, far more likely to be studying part-time while working, and more likely to be juggling caring responsibilities, shift work, insecure income and, in regional areas, long travel. When an employer offers a full-time job midway through a course, the rational choice for a person on a low income is often to take the job. The economy gains a worker. The completion table loses a student.
There is also a measurement wrinkle running the other way. Students who leave with employable competencies frequently do not formally withdraw; they simply stop attending, and the timing at which administrative systems lapse them varies by jurisdiction and provider. That noise is another reason NCVER's four-year window is the sensible convention and short-run divisions are not.
So the honest position is that a mature Fee-Free TAFE completion rate, whenever it is published, should be expected to be well below what an unfamiliar reader would guess, for reasons that have nothing to do with whether this policy was a good idea. That is the baseline against which it must be judged. Anything else is comparing the policy to a standard the sector has never met and was never designed to meet.
What is not published: the composition question
Here the argument turns, because everything above is a defence of the government's numbers and the next part is not.
The policy's stated purpose was to relieve measured skills shortages. Australia has an official instrument for identifying those shortages — the Skills Priority List, published annually by Jobs and Skills Australia, rating occupations as in shortage nationally and by state, with supporting analysis of future demand. The list is detailed, occupation-level, publicly available, and updated. It is exactly the reference point a completion statistic would need to be mapped against for the public to know whether the money worked.
That mapping is not routinely published in a form a citizen can use. The public record reports enrolments in broad priority areas — care, construction, technology, agriculture, hospitality, clean energy — and reports completions as an aggregate national count. What it does not put in front of the public, on a schedule, is the thing that actually answers the question: how many completed qualifications, in which occupations, against which entries on the Skills Priority List, and what happened to those graduates in the labour market afterwards.
The distinction matters because broad priority areas are wide. "Care and support" spans aged care, disability support, early childhood and allied roles with materially different shortage severity, wage levels and retention profiles. "Technology" spans qualifications with strong labour-market returns and qualifications with weak ones. A completion inside a priority area is not automatically a completion in an occupation that was short.
NCVER does publish the raw material for much of this. Its Total VET students and courses collection reports enrolments and completions by field of education and qualification level. Its Student Outcomes survey reports what happened to graduates afterwards — employment status, whether the job is related to the training, and reported wage effects. The data infrastructure exists and is of good quality. What does not exist, in any prominent public form, is a regular report that joins those datasets to the Skills Priority List and to Fee-Free TAFE funding, and tells the public how many funded places became qualified workers in occupations Australia had formally declared short.
The reporting cadence, which is the measurable part of the complaint
The clearest evidence for the thesis of this article is not a statistic. It is a comparison of two publication schedules.
Enrolment figures are generated by administrative systems as students enrol. They are available quickly, reported quarterly, and they appear in ministerial media releases, budget documents and speeches within months of the activity occurring. They are a good news number and they are treated as one.
Completion figures move through NCVER's annual collections with a substantial lag, and the methodologically sound completion rate arrives four years after commencement. Outcomes data — whether the graduate got a job in the field — arrives later still. Occupational matching against the Skills Priority List is not, on the public record, produced as a standing report at all.
None of that is fraud and none of it is even unusual. It is the ordinary consequence of measuring inputs with administrative data and outputs with statistical collections. But the consequence is systematic and it runs in one direction. The metric that flatters the policy arrives fast and gets promoted. The metric that would test the policy arrives slowly, in a technical publication, with a four-year lag, and gets promoted by nobody. Over a five-year agreement and a legislated ongoing program, that asymmetry does real work on public understanding.
The critique: a place filled is not a skill gained
The case against Fee-Free TAFE, stated as strongly as the evidence permits and no more strongly, is not that it failed. It is that the country has built a large, now-permanent spending program whose success is defined by an input, promoted by an input, and not yet accountable on any output.
Follow the logic in three steps. First, an enrolment is a place filled. It tells you that a student and a funded seat came together. It tells you nothing about whether the student finished, and by design it cannot. Second, a completion is a skill formally attained. It is a much better measure, but it is still not the objective, because a qualification in an occupation with no shortage does not relieve a shortage. Third, the objective is a qualified person working in an occupation the Skills Priority List identified as short. Only the third step is the policy's actual purpose. Only the first step is reported quickly and loudly.
A place filled is not a skill gained. A skill gained is not necessarily a shortage filled. The public is reliably told about the first, occasionally about the second, and almost never about the third.
The critique has to be stated with discipline, so here is what it does not claim. Not that the completion rate is bad; the mature rate is not yet knowable and will not be for years. Not that the money was wasted; a partial completion that produces a licensed, employed worker is a return on public money whether or not the certificate was issued. Not that students or teaching staff underperformed; nothing in the data supports that, and the vocational workforce has delivered a very large increase in volume under considerable strain. And not that the shortages were invented; they were measured by a Commonwealth statutory body and published.
What it does claim is a defect in accountability design. The program was legislated into permanence in 2025 with a floor of 100,000 places a year from 2027. A permanent program funded by the Commonwealth and delivered by the states, sitting inside a five-year agreement worth in the order of $12.6 billion (approx.) in Commonwealth funds, should have a standing, published, occupation-level outcome report. On the public record, it does not have one that a citizen can readily find and read.
Why the input metric is the one that gets used
There is no conspiracy here, and it would weaken the argument to suggest one. The input metric dominates because of ordinary incentives that would operate under any government of any party.
Enrolments are available inside the political cycle. A minister announcing a program in 2023 can report enrolments in 2023 and again in 2024. Completions for that cohort become methodologically sound in 2027, by which time the announcement is four years old, the portfolio may have changed hands, and the political value of the number has largely evaporated. Any government would prefer the number it can use.
The federal structure compounds it. The Commonwealth funds; the states and territories deliver through their own TAFE systems under their own arrangements. Enrolment counts aggregate cleanly across jurisdictions. Outcome measurement is harder to standardise, requires joining datasets held by different bodies, and produces results that are politically awkward for whichever jurisdiction comes last. The path of least resistance is to report the thing that aggregates cleanly.
And the underlying question is genuinely hard. Working out whether a completed aged care qualification relieved an aged care shortage means linking training data to employment data, allowing for people who would have trained anyway, and accounting for shortage being a function of wages and conditions as much as of supply. That is real analytical work, and it is not impossible — NCVER's outcomes survey and the ABS's data integration infrastructure provide much of what is needed. But it has to be commissioned, funded and published, and nothing in the current arrangements obliges anyone to commission it.
The question the data cannot yet answer, stated honestly
The central empirical question of this topic is whether Fee-Free TAFE completions are concentrated in occupations the Skills Priority List identifies as being in shortage, or dispersed across courses with weaker labour-market returns.
This article cannot answer it, and it will not pretend to. The publicly available reporting does not permit an answer, which is itself the finding. What can be said is that the question is answerable with data that already exists, that no published product currently answers it, and that a permanent program of this size ought to have one. A reader who finishes this article believing the composition of completions is poor has read it wrong. A reader who finishes it believing nobody has told them either way has read it correctly.
A design question sits underneath the measurement question, and it deserves stating even though it cannot be resolved here. Removing the price of a course changes who enrols; it does not change what people want to enrol in. If the courses students most want are not the courses the shortage list most needs, a fee-free place will faithfully deliver a student into the former. Where a shortage is driven by pay and conditions rather than by training supply, subsidising training will not fix it, and no volume of enrolments will substitute for the thing that would.
What they got right, and the honest caveats
Three things belong plainly on the government's side of the ledger, and the critique above is only credible if they are stated first and without hedging.
The fee barrier was real, documented, and removed. Up-front tuition costs are a well-established obstacle to vocational participation, and they bind hardest on exactly the people the training system most needs to reach: people on low or interrupted incomes, people re-entering work after caring, people in regions with fewer options, and young people without family capacity to front the money. Removing that barrier was not a gesture. It was a targeted intervention on a known constraint, and the response demonstrated the constraint was binding.
The participation shift among target cohorts is measurable and it happened. Departmental reporting shows increases concentrated among younger students, First Nations learners, people with disability, and people who were unemployed at enrolment (DEWR, Fee-Free TAFE reporting, to the March quarter, 2026). That is a genuine equity outcome. It has value regardless of what the completion figures ultimately say, and a critique that treats access gains as worthless unless they convert to certificates is applying a standard no education policy has ever met.
And the policy is being judged early, by a convention that guarantees early numbers look weak. This deserves emphasis because it is the point most often abused in public commentary. NCVER's four-year rule means that any assessment of Fee-Free TAFE completions made in 2026 is looking at a partial count of an unfinished cohort. The government is entitled to say so, and it is right when it does.
The attribution also has to be drawn carefully. Low vocational completion rates were not caused by this government or this policy; they are a long-standing structural feature of the Australian training system, inherited by every administration of both parties for decades. This government did not cause them. It has presided over a period in which they persist, in a program of its own design. What it can fairly be held responsible for is the accountability architecture around a program it chose to make permanent — a live decision, made recently, and entirely within its control.
The caveats on the credit are equally honest. The enrolment figure counts enrolments rather than people, which inflates the intuitive impression the headline creates. The over-delivery against the original 180,000 target in 2023 is a weaker achievement than it sounds if the target was conservatively set. Extending and then legislating a program before the first cohort's completion data was methodologically available is a decision made in the absence of the evidence that would have tested it — defensible on the ground that skills shortages would not wait, but a decision taken without the number, nonetheless. And the strongest defence of the policy, the four-year convention, has a use-by date. It excuses judgment now. It will not excuse it in 2028.
The risk to the future
A permanent program measured by what goes in, with no standing report on what comes out
The forward risk is not waste, and framing it as waste would be both unfair and analytically lazy. The risk is that Australia has now legislated a permanent annual commitment of at least 100,000 subsidised training places — a program that will run through multiple governments and multiple economic cycles — without building a public accountability mechanism that ties places funded to shortages actually filled. That is a design gap, and design gaps compound quietly. A program judged on enrolments will, over time, be optimised for enrolments, because that is the number that determines whether it is reported as succeeding. Nobody has to intend that for it to happen.
The specific danger is that the four-year convention, which is a legitimate methodological safeguard, becomes a permanent shield. It is a good reason not to draw conclusions in 2026. It is a bad reason not to publish anything. If no standing outcome report is established now, the sequence writes itself: early criticism is deflected as premature, the data eventually matures inside a technical publication that attracts no attention, the political moment has passed, and the question of whether the money relieved the shortages it was appropriated to relieve is never put to the public at all. That is not a failure of the policy. It is a failure to ever find out.
What would close the gap is not complicated and does not require waiting. An annual published report — from the Commonwealth or from NCVER under a standing commission — tracking each commencing cohort to its four-year completion point, mapping completions by occupation against that year's Skills Priority List, and reporting employment outcomes using the existing outcomes survey. Every component already exists. What is missing is the obligation to join them and publish the result on a schedule that does not depend on any minister finding the answer convenient.
The generational stake is straightforward. The people this policy was built for — a woman in her forties changing careers into aged care, a nineteen-year-old choosing between a certificate and a start on the tools — deserve a training system that can demonstrate, with evidence, that the qualification it offered them leads somewhere. And the country needs to know whether the money it is now permanently committing is producing the workers it is short of, or simply producing enrolments. On the current arrangements it will find out slowly, quietly, and possibly not at all. That is a fixable problem, and it is easier to fix in the first year of a permanent program than in the tenth.