Signal Scanner · WORKFORCE, SKILLS & ORGANISATIONAL CHANGE · 12 September 2026

Certificates Without a Premium: More Than 40% of Certified Training Sits in Low-Return Topics

New OECD measurement puts most employer training in the subjects with the weakest wage association, while the largest premium attaches to teamwork, leadership and project management. Exposed: HR and learning functions, compliance, finance, training providers and certification bodies.

The reskilling conversation is almost entirely about volume: not enough training, not enough hours, not enough budget. On 7 July the OECD published something more awkward. Using the 2022-2023 adult skills survey, it measured which training topics actually associate with higher wages, and then measured which topics employers buy. The two lists barely overlap. The subjects carrying the largest premium are the ones firms fund least deliberately; the subjects firms fund most are the ones with the weakest measured return. Nothing in that is an argument against training. It is an argument that the composition of the budget, not its size, is the decision nobody is currently making.

Signal Identification

A measurement shift with a budget consequence. Corporate learning has never had a topic-level return estimate to allocate against, so allocation has defaulted to obligation: what law, insurer or auditor requires. The OECD has now put numbers on both sides of that default, at the same moment the EU has recast its newest workforce training duty as documented effort rather than demonstrated capability.

Time horizon: 1-5 years (OECD evidence published 7 July 2026; EU AI literacy duty recast 8 July 2026 and in force from 27 July; training budgets re-cut through the 2027 and 2028 planning rounds)
budgets re-cut2026202720292030to 2035
Plausibility band: Medium–High
LowMediumHigh
Geographic / Jurisdictional Scope: Primary: the 31 countries in the OECD adult skills survey, with the EU-27 carrying the added AI literacy duty. Spillover: the United States, United Kingdom, Singapore and Korea.
PrimaryEU-27OECD PIAAC 31
SpilloverUnited StatesUnited KingdomSingaporeKorea
Sectors exposed:
Learning and development functionsCompliance and riskCorporate finance and budget ownersVocational education providersCertification and awarding bodiesManufacturing and frontline employersHealthcareProfessional services

What's Changing

The OECD ranked training by wage association and found the top of the list is social, not technical. Courses in teamwork, leadership and project management show the largest association, with “the training wage premium exceeding 8.5%”, while reading, writing, languages, computer and software and communication training sit at “about 5.5 to 8.5%” (OECD, 07/07/2026). Non-formal learning overall carries roughly 6.2%. Duration and format do not much matter: very short courses perform like long ones, and online or hybrid delivery performs at least as well as in-person.

Then it measured the purchase. Health and safety is the single most common topic adults complete, over one-quarter of all training undertaken, and the low-return group of health and safety, operating machinery and handling customers accounts for about 40% of in-person training. More than 40% of workers completing certificate-granting training “did so in topics with relatively low associations with wages” (OECD, 07/07/2026). Participation is also falling: formal job-related education down to 8% of adults, non-formal down 3 p.p. to 37%, and the education-to-wage association itself down 2.2 p.p. between survey cycles.

The next mandate points the same way. On 8 July the EU replaced the AI Act's literacy duty: providers and deployers must “take measures to support the development of AI literacy” of staff, and the article now says expressly that the obligation “does not require providers or deployers to guarantee any specific level of AI literacy of any individual” (EUR-Lex, 08/07/2026). A duty satisfied by documented effort is a duty that will be met by a course and a completion record.

What the OECD measured: the premium ladder, and where the training actually is

TRAINING TOPICS, ORDERED BY MEASURED ASSOCIATION WITH WAGES Teamwork, leadership, project management above 8.5% Reading, writing, languages, software, communication 5.5% to 8.5% Non-formal learning, all topics about 6.2% Health and safety, low association with wages machinery, handling customers AND THIS IS WHERE THE TRAINING ACTUALLY SITS over 25% of all training adults complete is health and safety about 40% of in-person training is in low-return topics over 40% of certificate-granting training is in low-return topics

Source: OECD Employment Outlook 2026, Chapter 4, Skills at work (7 July 2026), drawing on Cycle 2 of the Survey of Adult Skills. Associations, not causal estimates.

Disruption Pathway

Stage one is the 2027 planning round, where the finding meets a budget that is already committed. Compliance training is contracted, audited and defended by the risk function; the high-return topics have no such sponsor. Stage two runs through 2027 and 2028 as the AI literacy duty is implemented across the EU-27, adding a new mandated line that, by design, can be discharged with a course record (EUR-Lex, 08/07/2026). Stage three is measurement: once one employer publishes topic-level outcomes, the rest are asked why theirs are unmeasured.

Stress concentrates in three places. Ownership is the first: compliance owns the mandated spend, learning owns the discretionary remainder, and nobody owns the ratio. Capacity is the second, because employers are committing to growth they do not plan to staff, with 61% expecting revenue growth by 2027 against 50% expecting headcount to rise (Gallagher, 09/09/2026). The third is demand-side mismatch, with Cedefop flagging shortage pressure in managerial and professional occupations “driven by strong demand, high replacement needs and qualification mismatches” (Cedefop, 21/07/2026). Two adaptations follow. Employers begin reporting training by topic rather than by hours or spend. And the cheap formats the OECD found perform just as well, short and online, get pointed at the expensive subjects instead of the compliance ones.

Why This Matters Now

This lands on CFOs signing the learning budget, on chief people officers who report it in hours, and on boards that treat training as a cost line with no return estimate attached. The decision architecture that needs revising is the one that books mandated and discretionary training to the same account and reports them as one number. Employers should split the two in reporting, so the obligation-driven share is visible, and should redirect the discretionary remainder toward the topics the evidence favours. Learning functions should stop defending volume and start defending mix. On the available evidence, the firms that gain over the next two planning rounds will not be the ones that spend more; they will be the ones that can say what they bought.

Decision-action posture for this signal: Prepare — the measurement is published and the next mandated training line is already law, but the estimates are associations rather than causal effects, so the sensible move is to instrument and re-report the budget now and commit the reallocation when internal outcome data confirms the pattern.

Counter-Argument

The strongest objection is that wages are the wrong yardstick, and the OECD's own summary is more careful than its chapter. It reports that numeracy and formal education “remain strong predictors of employment and wages, although their influence has weakened over the past decade” (OECD, 07/07/2026), and the chapter concedes its estimates are associations, not causal effects. Compliance training is also not bought for a wage premium; it is bought to avoid a loss. Where employers do measure, they measure something else: Tyson Foods reports 91% annualised retention among on-site learners, and CJK Group cut its path to master operator from six-plus years to two (Aspen Institute, 01/09/2026).

Both points are right and neither rescues the current allocation. If wages are the wrong measure, employers still have no topic-level measure of their own, which is the finding restated rather than refuted. And the market's belief runs the other way: 94% of employers told a Coursera survey they would pay more for a candidate with a microcredential, and six in ten would prefer a less experienced candidate holding a generative AI credential (Inside Higher Ed, 05/06/2026). Belief that certification pays is exactly what makes an unexamined training mix durable.

Implications

This is durable rather than cyclical, because the allocation rule behind it is institutional. Mandated training has an owner, a deadline and an auditor; discretionary training has a budget line. Nothing about the OECD's numbers changes that, but it does make the resulting mix a choice somebody can be asked to defend. The inflection window runs from the 2027 planning round to the first employer that reports training by topic and outcome. Winners are providers of short, online, social-skills content and the employers who buy it deliberately. Losers are compliance-led training vendors whose value proposition is the completion record, and awarding bodies certifying subjects with no measurable premium.

Early Indicators to Monitor

Disconfirming Signals

Strategic Questions

Keywords

Employer training; returns to training; OECD Employment Outlook 2026; Survey of Adult Skills; PIAAC; compliance training; AI literacy; EU AI Act Article 4; micro-credentials; skills-first hiring; learning and development budgets; workforce capability

Bibliography

Source tiers: Tier 1, governments, regulators and intergovernmental bodies. Tier 2, think-tanks, academic institutes, major consultancies and quality data providers. Tier 3, quality journalism and specialist trade press. Tier 4, vendor, company and practitioner sources, used only as directional corroboration.

This Signal Scan is one component of a Shaping Tomorrow intelligence cycle on Workforce, Skills & Organisational Change. Every monitored topic also includes monthly Change Trackers that assess what is materially shifting in momentum, and a quarterly Decision Intelligence briefing that converts the cycle into board-level strategic implications. Cycles may also close with Action Triggers, an end-of-cycle synthesis that reads across the three formats and surfaces patterns no single deliverable carries on its own. Continuous intelligence on this topic is available as a Single-Topic or Multi-Topic programme.

Discuss a Tailored Programme

Prepared by Shaping Tomorrow: 12 September 2026