Career economics · Diagnostic
A South African building agreement paid a Class 2 tradesman R73.84 an hour and an artisan R98.90 in 2022/23. What moved a worker from one to the other was a certification event. At those rates, an experienced worker would recover the reported cost in about three months, and the country still produces far fewer artisans than its development plan wants by 2030.
Most thinking about career progression assumes the step up is a new skill. On this frontline ladder in 2022/23, every rung had a written price and the largest step was a certification event. The Cape building industry's collective agreement describes each tradesman grade as a worker who "has not completed a trade test, but could be regarded as proficient at this level in the relevant trade". In 2022/23, passing the test, or completing the last of the credits, added R25.06 an hour, the biggest rand step in the schedule. For a worker with years of experience and no formal credits, the test is the only way across, and the route to it has three locks, none of them the worker's skill.
The collective agreement of the building industry's bargaining council for the Cape of Good Hope set six minimum hourly rates for the year from November 2022. A labourer earned R34.70 and a general worker R45.11. Three classes of tradesman sat above them, and at the top was the artisan at R98.90.1
The definitions are the interesting part. Every tradesman class is described in the same words: an employee who "has not completed a trade test, but could be regarded as proficient at this level in the relevant trade". What separates the classes is the share of credits completed for the trade, under 55 per cent for Class 4 and 75 to 99 per cent for Class 2. An artisan is a worker registered under clause 17, which provides for registration once the employee "has completed and passed the trade test or has completed the number of credits that qualify him to work as an artisan and his employer has requested registration in writing".1
Read the definitions against the rates. The agreement treated each tradesman as proficient at their level and priced the levels by certified progress. In 2022/23, the two largest steps both sat at certification thresholds: Class 3 to Class 2, triggered by further credits, was worth R19.67 an hour or 36 per cent, and Class 2 to artisan, triggered by the trade test or the last of the credits, was worth R25.06 an hour or 34 per cent. The parties who wrote the schedule, employers and unions together, put the largest rand step on the ladder at the point of certification.
Paper six argued that frontline ladders exist but are undrawn. This one is drawn, gazetted and priced, which makes it the cleanest place I know to measure what moving one rung is worth.
Why should the largest step on the schedule sit at a certification event? Because in South African law the trade test is the only door to the artisan qualification.
Section 26D of the Skills Development Act provides that no person may obtain an artisan qualification without passing a trade test, and that the Quality Council for Trades and Occupations must issue a trade certificate to a candidate who passes. Trade tests are moderated by the National Artisan Moderation Body, established under the same Act as a unit of the Department of Higher Education and Training.2 The certificate is national, issued by a statutory body, and checkable by an employer who has never met the worker.
The pay agreement was slightly wider than the statute, since it also admitted a worker who has completed all the credits for the trade. For the worker this paper is about, someone with years of practice and no formal credits, that alternative does not exist, and the test is the only way across.
That makes the step the verifiability term from paper nine in close to its purest form. Scarcity, demand and transferability are identical on the day before the test and the day after. The worker's hands are the same hands. What changes is that a stranger can now check what those hands can do, and the agreement priced that change at a third of the hourly rate in 2022/23.
The system also contains the route this worker needs. Artisan recognition of prior learning, set out in criteria gazetted in March 2017, lets a worker aged 19 or over with at least three years of relevant, verifiable experience reach the trade test without a formal apprenticeship. After orientation, the candidate is evaluated on knowledge, a practical task and workplace experience; the evaluation grades each part, and a score of 50 per cent or more marks the candidate ready for the test.3 On paper, the experienced but uncertified worker has a door built for them.
Start with a single worker's decision, because the arithmetic is short.
What is the trade test worth to a Class 2 tradesman?
Rates were the agreement's minimums paid in 2022/23, November 2022 to October 2023, the same year as the reported cost; later years added between 5 and 7 per cent a year. The cost is a single report for Pretoria and Cape Town rather than an official fee schedule, so read the payback as an order of magnitude: months, not years.
A payback of roughly three months is short for any investment, and remarkable for a step worth a third of the hourly rate for the rest of a working life. A return that fast ought to draw candidates to the test in numbers.
The national figures show output well short of what the country wants. South Africa produced 172,805 qualified artisans across 124 trades between 2011 and 2020, an average of 17,281 a year, with a peak of 24,050 in 2019 and a fall of 37.2 per cent to 15,106 in 2020.5 In November 2022 the Minister of Higher Education, Science and Innovation put average output at 20,000 a year against a National Development Plan target of 30,000 a year by 2030, and reported that 10,302 learners entered artisan programmes in 2020/21, a fall of 36.5 per cent on the year before.6
Two readings of the gap are possible. One is that demand for artisans is weak, so the high minimum rate is paid to few people; there is something in that, and section six returns to it. The other is that the route to the test is harder to travel than the payback suggests. The evidence on the second is thinner, but it points somewhere specific.
Follow an experienced but uncertified worker towards the test and the route has three points at which it can stall.
The first lock is evidence of experience, and the cheapest source of it is the employer. The recognition criteria are more generous than they might be: they accept experience "accumulated from various traceable employers or services provided to customers in the relevant trade", and they provide for an unemployed candidate to be evaluated to establish experience gained earlier.3 But a letter from the current employer is the quickest proof, and the alternatives are slower and less certain. In a question answered by the GroundUp newsroom in October 2022, a worker with three years of practical experience wrote that their employer would not allow them to get the documents required to take a trade test; the suggested recourse was the independent recognition route or an unfair labour practice claim at the CCMA.4 The building agreement added a second signature at the far end, since registration as an artisan also required that "his employer has requested registration in writing".1 One documented case proves nothing about frequency. It does show where the friction sits.
The second lock is the price. The same GroundUp answer reported the all-inclusive cost of the trade-test process at R14,500 in Cape Town and R12,500 in Pretoria.4 I could not find an official fee schedule, so this is one reported figure. Against the Class 2 weekly minimum paid in 2022/23 of R2,953.60 for 40 hours, it was four to five weeks of gross pay, paid up front, for a return that arrives afterwards.
The third lock is access to testing. The national artisan report for the Department of Higher Education and Training states that "accessibility to testing centres is an issue for TVET colleges and for the clients who are looking to get recognition of prior learning".5 A 2025 study of recognition of prior learning among accredited providers in Gauteng adds a shortage of skilled practitioners, weak management structures and financial strain, and found financial constraints and low stakeholder engagement "particularly detrimental".7 That study covers one province's providers, but it points the same way as the national report.
None of the three locks is the worker's capability. The test assesses that. The locks decide whether the worker reaches the test at all.
Put the two halves of the evidence together and the incidence of the gap becomes plain.
The worker pays first through the rate. In 2022/23, a tradesman the agreement regarded as proficient at their level was paid R25.06 an hour below the artisan rate while uncertified, which at 40 hours came to about R1,000 a week. The worker pays again through the fee, up front, to the test centre, for the right to be examined.
The first payment does not disappear. Where a Class 2 worker was doing work the employer knew to be of artisan standard, the gap between the two rates stayed with the employer. This is paper nine's point about informational monopsony made concrete: the current employer knows what the worker can do, outsiders cannot check it, and the difference between those two states of knowledge was worth R25.06 an hour in 2022/23 to whoever controlled it. One case cannot establish motive, and I am not suggesting the employer in the GroundUp question acted from it. The incentive, though, runs in one direction, and a system in which the employer is the cheapest verifier of experience asks the party that gains from the gap to help close it.
A sceptical reader has three objections, and each deserves a straight answer.
First, the Class 2 tradesman has not finished their credits, so the artisan rate paid for the remaining competence as well as for the test. That is true, and the agreement's own wording, proficient "at this level", concedes a gap. For the experienced worker, though, the recognition route exists precisely to assess what remains and to send the worker to the test once it is closed. The rung bundles an increment of skill with a large signal. The schedule priced them together, and the available data cannot separate them.
Second, negotiated minimums are not market wages. They are the floor in one region's bargaining council, many artisans earn above them, and the agreement cited here ended on 31 October 2025, with later years adding between 5 and 7 per cent a year. I could not find a national estimate of the artisan premium over semi-skilled work, from the Quarterly Labour Force Survey or anywhere else, and that absence is itself a finding: one of the country's most consequential frontline credentials has no published price outside collective agreements.
Third, certification cannot create demand. Over the decade in which 172,805 artisans qualified, 30,844 artisans registered with the public employment service as unemployed work seekers, on a database that covers only 34 of the 124 trades.5 In a trade with weak demand the certificate multiplies a small number, which is paper nine's framework working as intended rather than failing. The case for opening the route is strongest in the trades where demand is real, and the national figures do not yet say which those are.
One further limit cuts the other way. A 2025 study of recognition assessments for candidate artisans found practitioners working with "very narrow understandings of competence", and an assessment toolkit that left attitudes and emotional competencies outside its scope.8 Even the strongest verification event in frontline work certifies the technical core and misses the social skill that paper nine argued demand has been shifting towards.
A system in which the employer is the cheapest verifier of experience asks the party that gains from the gap to help close it.
Three changes follow, in rising order of cost.
Publish the numbers. A national pass rate by trade, annual counts of candidates on the recognition route and a labour-survey estimate of the artisan premium would cost very little and would tell every worker, funder and provider what the rung is worth. I found no current figure for any of the three in the official record, which is the most surprising finding of this paper.
Make the alternative evidence route fast. The criteria already accept experience from other traceable employers and from customers, and they allow assessors to evaluate candidates who are out of work. What is missing is evidence of that kind a centre can accept quickly: verified records of work held by the worker, which do not depend on the goodwill of one employer. Paper twelve sets out what such a record has to contain to be trusted.
Fund the test against the gain. A step that repays in months is the kind of investment a levy-funded skills system exists to finance, and financing it up front removes the lock that falls hardest on the worker with the least cash.
| Lock | What the evidence shows | Strength | Fix |
|---|---|---|---|
| Evidence | Criteria accept other traceable evidence, but the employer is the quickest verifier; one documented refusal; registration needs the employer's request | Thin | Accept verified work records held by the worker as traceable evidence |
| Price | R12,500 to R14,500 reported in 2022; no official schedule found | Thin | Finance the test up front and recover it from a gain that repays in months |
| Access | Test-centre access flagged in the national artisan report; provider barriers in a 2025 study | Indirect | Publish centre capacity and waiting times by trade, then fund to them |
| Information | No current national pass rate, no recognition-route volumes, no national premium estimate | Absent | Publish all three annually, starting with the pass rate |
For readers in the Gulf the lesson is specific. Saudi Arabia already tests workers at scale: its Skill Verification Programme, the scheme paper five discussed and now reported as professional accreditation, had accredited more than 460,000 workers by the end of the first half of 2025, using theoretical and practical exams across more than a thousand professions.9 I have found no published wage schedule that prices the result. South Africa shows what happens when a certification event is written into a pay scale: it becomes the most valuable rung on the ladder. The Gulf has built the test without the price, and South Africa has the price behind a narrow door.
Disclosure
The case above stands on its own evidence; nothing in it depends on what follows. I am the founder of UpSkillMe, which builds assessment and a worker-owned capability record for household work in Saudi Arabia and the trades in South Africa. The second fix in section seven, verified work records held by the worker and accepted as evidence of experience, describes what my company builds for exactly the tradespeople this paper discusses, and the reader should discount that recommendation accordingly. The statute, the collective agreement, the gazetted criteria and the national output figures are public and stand independently of us.
Next in this series: why training without a signal can fail, and what the experiments say a certificate is worth on its own.
Written in British English; source titles and quoted text keep their original spelling. Wage figures are one bargaining council's negotiated minimums for the first year of an agreement that expired on 31 October 2025; the terms of any successor agreement could not be confirmed. The trade-test cost is a single reported figure. No current national trade-test pass rate, no published volumes for the recognition route and no national estimate of the artisan wage premium were located, and none is stated here. The unemployed-artisan registrations come from a database covering 34 of 124 trades and are a count, not a rate. The DHET annual report's small tested-and-qualified counts were excluded because they appear to cover a single programme rather than the national total.