Over the past 25 years, business has sunk billions into the creation of the sector education & training authority (Seta) system. Intended to catalyse a skills revolution, this massively inefficient edifice consistently fails to meet its own skills targets, is plagued by allegations of malfeasance, and facilitates training more costly on some metrics than obtaining a university degree.
Barely a week goes by without the Setas making headlines — always for the wrong reasons. In July, President Cyril Ramaphosa fired higher education minister Nobuhle Nkabane after she made several overtly political appointments to Seta boards and then lied about it before a parliamentary committee.
This has highlighted that the Setas, like so many other public entities, have become sinkholes of political patronage and a drain on the economy. It has heightened calls for the system to be overhauled, if not scrapped entirely.
The Seta system was a state-led intervention to remedy the skills deficit inherited from the apartheid era. Its main rationale was to facilitate the kind of generalised training that most businesses would not provide themselves.
It is, in theory, designed to solve this perceived market failure — what economists refer to as a “collective good” problem. It does so by compelling business to contribute to the collective cost of training through a mandatory 1% skills levy (a payroll tax) on firms whose annual payrolls exceed R500,000.
It is coupled with a grant mechanism which refunds companies up to 20% of their levy as long as they submit an annual report on certified training undertaken and a workplace skills plan — something fewer and fewer of them are bothering to do.
However, despite the Seta system absorbing about R20bn in levies a year, and supporting a staff of close to 3,000 people, not even 100,000 individuals were certificated in 2023/2024. And as ever, firms continue to cite skills shortages as a major business constraint.
Setas are unperforming because they are entirely the wrong way of thinking about how to raise skill levels. They are premised on the idea that the best way to do so is to create a central planner that can decide what skills are needed
— Roy Havemann
In a new Bureau for Economic Research (BER) research note, which analyses the performance of the 21 Setas from 2011/2012 to 2023/2024, independent consulting economist Robert Botha and BER senior economist Roy Havemann lay bare just how costly, wasteful and deeply inefficient the system has become.
The 82-page report, which was funded by Economic Research Southern Africa, finds that the Seta system is not an efficient way to improve skills. Indeed, given its significant cost and the fact that it’s funded by way of a payroll tax, it is likely damaging both growth and employment, the BER argues.
It recommends that the system be overhauled and favours replacing it with a private sector-driven scheme funded through a revenue-neutral tax incentive.
A costly, wasteful and deeply inefficient system
The BER’s first key finding is that the Seta system is extremely expensive. Its levy allocation rose from R8bn in 2011/2012 to R17.9bn in 2023/2024 — draining a cumulative R164bn from firms’ coffers over 13 years. Of this, 10% went on administration costs.
The Seta levy is also significant relative to other tax and budget sources.
According to the 2025/2026 budget, the Seta system will get R20.8bn in the coming year, more than the net revenue from the withdrawn one percentage point VAT increase (R11.5bn), or R13.7bn in transfers to technical & vocational education and training (TVET) colleges.
Second, the Seta system is also excessively costly per beneficiary compared to other forms of education and training.
For instance, in 2023/2024 the cost per Seta enrolment (R162,879) was more than twice the cost of a university enrolment (R76,405), when low-cost, high-volume Seta skills programmes are excluded.
(Skills programmes are mainly short programmes, and in some cases single-unit courses, so are not comparable to three-year university degrees. To include them inflates Setas’ enrolment and certification figures and lowers their unit costs.)
The cost per Seta enrolment was also significantly higher than the cost of National Student Financial Aid Scheme (NSFAS) funding per university student (R73,829) and TVET college funding per student (R34,230).
The cost per Seta certification rockets to R388,052, again if skills programmes are excluded. It also exceeds that of a university graduate certification (R370,923), despite the fact that universities also have costly research mandates.
(The cost of enrolment is the system’s annual levy allocation divided by the number of individuals enrolled in that year. The cost of certification is the same budget divided by the number of people who successfully completed their studies. The reason the cost of certification vastly exceeds that of enrolment is due to the high number of dropouts.)
Third, the Seta system is marked by high dropout rates.
Over the past 13 years, 2.6-million individuals were registered in Seta learnerships, internships, artisanal programmes or skills programmes. Of these, roughly 2-million completed their courses and were certified. Though these figures suggest a sizeable operation, the authors describe it as a “leaky pipeline” in that 630,717 individuals dropped out before certification.
This systemic leakage is most pronounced in learnerships and internships — areas crucial for deep skills acquisition and facilitating the transition from school to the world of work.
For every 10 participants who begin an internship, six don’t complete it — a throughput rate of just 42%. This represents a “missed opportunity” to transition individuals into the workforce and points to a “critical area of weakness” in the Seta system, say the authors.
In addition to training those who already have jobs, the Setas are also expected to upskill unemployed people. So, providing learnerships for unemployed individuals is paramount.
Over the review period, 675,527 unemployed people were registered in Seta learnerships, of which 355,411 were certified. This is a 53% completion rate. However, it satisfied 99% of the Setas’ target for learnerships — something Setas set in conjunction with the minister of higher education & training. Clearly, Seta targets are woefully unambitious.
Artisanal programmes did much better with a throughput rate of 83%, and skills programmes did the best of all with 96%. This translates into a respectable throughput ratio of 76% across all programmes.
However, this is distorted by the fact that short-duration skills programmes account for about 60% of certifications. If they are excluded, the throughput rate for more substantive interventions drops sharply to 57%.
The BER suggests that the emphasis on achieving numerical targets may explain the proliferation of low-level (NQF 1 and 2), short-duration programmes as they allow the Setas to enrol large numbers of learners at a relatively low cost per head.
“This focus on mass programmes diverts financial and administrative resources away from the longer, more complex and more expensive interventions — such as apprenticeships and higher-level learnerships,” it notes.
Yet this is precisely the type of training required to address the critical shortages of artisans, engineers and technicians that constrain the country’s economic growth.
Fourth, the Seta wage bill has grown rapidly, from R745.8m in 2014/2015 to R1.9bn in 2023/2024 — an average annual increase of 12%. Over the same period, CPI averaged 5% a year.
This real growth is a product of a steep increase in Seta staff (from 1,752 to 2,748) and big annual pay hikes. In fact, the Seta wage bill has outpaced the growth of the public service wage bill since 2016.
In 2023/2024, the average Seta employee earned R715,000 a year, but 12 positions fell within the top salary band where the average individual annual salary is an eye-watering R3.2m. Heads of provincial education departments earn about R2m-R2.5m, though they shoulder far greater responsibilities.
Fifth, Setas are underperforming their own and the government’s targets.
The National Skills Development Plan stipulates that the Setas should aim to facilitate training for about 10% of the country’s 25-million-strong labour force (employed plus the unemployed) each year — about 2.5-million people. However, in 2023/2024, the system registered 165,125 people — only 1% of the employed and 0.7% of the labour force.
This is in sharp contrast to similar international schemes. About 50% of employees participate in the equivalent French scheme, and about 30% in the Canadian one.
The BER study’s input-output analysis shows clear signs of declining efficiency and productivity across the Seta system. For instance, while Seta revenue increased by 46% in real terms over the review period, the number of certifications decreased by 23%.
Similarly, while the number of Seta staff increased by 57% over the period, the ratio of certifications per staff member declined from 92:1 in 2014/2015 to 35:1 in 2023/2024.
Business Unity South Africa (Busa) CEO Khulekani Mathe agrees that the Seta system has failed to meet its goals.
“Key regulatory shifts have diverted a larger share of grant funding towards discretionary channels controlled by Seta bureaucracies, rather than supporting employer-driven training directly,” he says. “This has eroded the incentive for businesses to participate fully in the system.”
In addition, he says, rigid rules requiring funding to go only to accredited programmes exclude many agile, workplace-relevant offerings, especially in fast-changing industries, due to slow accreditation timelines.
“The outcome is a system disconnected from both economic demand and labour market realities.”
Busa believes the system must be fundamentally restructured to restore employer confidence, align incentives to job creation, broaden access to quality private providers, and embed transparency and accountability throughout the funding chain.
Manufacturing Circle executive director Philippa Rodseth agrees that a “fundamental rethink” is needed on how to develop the skills required by the manufacturing sector.
“The Seta system in its current form is inefficient and not fit for purpose,” she says. “From a practical perspective, we find that our member companies are not only contributing towards the skills development levy but also funding privately led initiatives tailored towards actual requirements.”
University of Cape Town (UCT) economics professor Haroon Bhorat also laments that South Africa has built a massive institutional structure that is just not delivering the goods. He says it’s difficult to extract even the most basic labour market information from many Setas, like vacancy rates, which lie at the heart of the economy’s skills gap.
“Too many Setas have been able to get away with massive wastage and poor governance,” he adds. “But perhaps often overlooked are those Setas that are seemingly well run and not corrupt but that do not actually deliver any meaningful returns in terms of skills enhancement and training.”
The conundrum of big surpluses
Given that the Seta system is plagued by allegations of corruption, the public would be forgiven for assuming that it is massively indebted. In fact, it is characterised by large and growing surpluses even as the auditor-general continues to report pervasive governance failures.
Between 2011/2012 and 2023/2024, the Setas’ combined irregular expenditure topped R9bn, or 5.5% of their total revenue from the levy. Over this period, 54% of the 273 Seta audits carried out were judged “unqualified with findings”, which, while it means the financial numbers were reliable, points to material noncompliance with legislation — often procurement rules.
The big build-up in reserves is because the system’s total revenue has consistently exceeded its total expenditure. This allowed its net surplus to rise from R3.9bn to R6.7bn over the review period.
The BER observes that while in a for-profit company, a large and growing surplus might be a sign of financial strength and prudent management, for a public entity with a public service mandate, the chronic accumulation of surpluses represents the unspent funds from all the previous years of operation. It is, therefore, a “profound indicator of systemic failure”, it says.
“The consequence is that billions of rand, specifically collected from employers to address South Africa’s critical skills shortages, are effectively taken out of circulation and left idle in Seta bank accounts,” the BER report explains. “This represents a massive opportunity cost.”
The report concludes that the Seta system, by raising the cost of employment with limited economy-wide benefits, diverts resources from other skills programmes and, by imposing a tax on employment, likely causes higher unemployment.
Should the Seta system be scrapped?
For Bhorat, a few clear solutions present themselves — such as caps on the salaries of Seta executives and board members, more active public sector financial oversight, and closer engagement with private sector human resource divisions in rebuilding dysfunctional Setas.
Ultimately, though, it’s not clear to him that South Africa will improve training outcomes by simply scrapping the entire skills system.
UCT senior research associate Andrew Donaldson, however, has no qualms about calling for the scrapping of the Seta system, including the 1% levy, believing there would be broad public support for such a decision.
He argues that the country’s tight fiscal position, coupled with the public consensus that the Seta system is not working and is a vehicle for political patronage, creates a “great opportunity” for change.
But scrapping the Setas shouldn’t be about saving the fiscus money, he says. It should be about changing the country’s skills training model to cut out the intermediary role that the Setas have assumed. Instead, firms and industry bodies should partner directly with TVET colleges and private training institutions that meet their needs.
He believes the “fundamental mistake” South Africa made in 1998 was to opt for sector-based training. He thinks training should be seen as a local benefit and, as such, should be funded by local businesses on a local or regional basis.
“The key thing is to reinforce a direct link and forge stronger ties between firms and their closest city-based or regional training colleges, whether public or private,” he explains. “We need to increase the revenue base of the underfunded and ineffective further education and training system and its governance by getting business more involved.”
Overall, Botha and Havemann argue that retaining some kind of skills incentive programme is important for long-run growth, but that it should not be the current Seta system.
They point out that since the Setas’ inception, the government has made multiple adjustments to legislation and policy to try to improve outcomes and accountability. However, none of these changes has significantly improved matters.
“This suggests that the Setas are failing because their design is flawed,” says Botha. “Over time, Setas have also arguably become patronage vehicles. So, their dysfunction is likely a combination of a conceptual flaw compounded by poor implementation. This combination makes salvaging the system extremely difficult.”
The BER paper proposes four options for reform: phasing out the Seta system including the skills levy; halving the levy to rightsize the system; redirecting the levy to other skills programmes; or — the authors’ preferred option — scrapping the Seta system but retaining the skills levy to fund a revenue-neutral skills tax incentive.
Option 1: Phasing out the Seta system, including the levy.
The first proposal is to replace the centralised Seta system with a tax-incentivised, employee-led skills development programme. This would allow for a system of stronger incentives where businesses make their own decisions, in consultation with employees, about what skills development training their employees need.
If firms no longer paid the 1% payroll tax, this would increase corporate profitability, some portion of which would return to the fiscus via higher tax revenue. The tax incentive could be funded out of increased tax collection so as not to increase the burden on the fiscus.
Option 2: Halving the Seta levy from 1% to 0.5%.
A payroll tax, like the Seta levy, is a particularly inefficient type of tax because it is not levied on profits, but on one of the input costs for companies — that of labour. By raising the cost of hiring, any payroll tax creates a disincentive to hire more workers.
Halving the levy would reduce this disincentive. It would also represent a meaningful cost saving, especially in labour-intensive sectors. For example, on a payroll of R100m, halving the levy would save R500,000 annually.
This would favour sectors with high formal employment and relatively high wage bills. But even small firms, which are disproportionately burdened by compliance costs, would benefit.
In any event, the fact that the Setas are running large, persistent surpluses suggests that the 1% levy is not being fully utilised and so could be cut without initially undermining training.
However, given South Africa’s unemployment and skills-mismatch problems, critics may argue that the Seta surplus should be spent more effectively, rather than reduced, particularly to support youth employment and artisanal skills development.
Option 3: Redirecting a portion of the Seta levy to other skills development activities.
In recent years, a share of the Seta levy that is statutorily transferred to the National Skills Fund has been directed to supplement NSFAS funding shortfalls. More recently there have been calls to use Seta surpluses to plug large current gaps in the basic education budget too.
However, while it may be tempting to do so, prudent public financial management requires that recurring expenditure be matched with recurring revenues.
Option 4: Scrapping the Seta system and decentralising control to create a more flexible, growth-oriented model funded from a revenue-neutral skills tax incentive.
The skills levy would still be collected, but instead of being used to fund the Setas (which would be phased out), firms would be allowed to claim tax incentives for training expenditures using levy funds, channelled through the tax system.
In other words, companies would be able to claim their qualifying expenditure on skills programmes off their skills development levy contribution. This would essentially create a ring-fenced pool of money for each organisation to spend on skills.
The effect would be to shift the choice of skills provision from the Setas to the companies themselves on the assumption that they are better placed to tailor skills training to industry needs and to address skills shortages in their sectors.
In Havemann’s view, “Setas are not performing because they are entirely the wrong way of thinking about how to raise skill levels. They are premised on the idea that the best way to do so is to create a central planner that can decide what skills are needed.”
Option 4 argues that a better way would be the abolition of this central planner and a move towards a decentralised system to allow for greater flexibility and innovation. As such, there is likely to be improved uptake and effectiveness. It would also reduce the administrative burden that firms complain makes participating in the system onerous.
If combined with incentives to hire and train entry-level workers, it could even help to address the “common good problem” that arises because there is little incentive for profit-maximising firms to provide generalised training.
The authors concede that phasing out the Setas could result in an unco-ordinated skills development system. However, they feel the poor performance, governance and audit weaknesses of the current system call for nothing less than a radical overhaul.
Claire Bisseker is an economics writer and researcher at the BER
The full BER report can be found here
Would you like to comment on this article?
Sign up (it's quick and free) or sign in now.
Please read our Comment Policy before commenting.