A skills shortage is a signal stack, not a training order

The composite is an intake signal

The OECD's Mapping Skill Shortages in OECD Countries, published 14 October, documents the 2026 Skills for Jobs database for 38 countries. Its occupational-imbalance indicator uses six labour-market signals: hourly-wage growth, growth in hours worked, employment growth, the inverse unemployment rate, the annual change in the underqualification rate and growth in online job postings. Most country data extend to 2024. This is not a real-time 2026 vacancy forecast.

The methodology chapter filters trends, standardises every sub-indicator within country and year, and combines them with equal weights except half-weights for employment and posting growth. An occupation is therefore ranked relative to other occupations in the same country and year. A positive value does not mean the absolute shortage is larger than a positive value in another country. Cross-country differences can reflect different relative positions within national labour markets.

Each signal can move for more than one reason. More postings may reflect demand, churn, duplication or employers keeping ads open. Wage growth may reflect bargaining institutions or hazardous conditions. Hours can rise because firms cannot hire, demand is temporary or existing staff absorb work. Employment growth is ambiguous. Low occupational unemployment and rising underqualification may accompany tightness, yet occupation coding, education categories and survey coverage still matter. Some countries lack a posting sub-indicator, and adding it makes earlier and later series not directly comparable.

Preserve the six-signal stack

For every occupation flagged as tight, retain each signal's direction, observation window, source, weight, missingness and revision date. Add local evidence: accepted offers, withdrawal reasons, vacancy duration, internal mobility, overtime, attrition and proficiency assessment for the exact task family. Do not convert a national occupation code directly into an internal skill gap without mapping actual work.

Then classify the likely friction. Supply means too few people currently demonstrate a named capability. Matching means capable people and openings fail to connect. Retention means the organisation cannot keep the capability. Job quality means pay, schedule, safety or design deters entry. Demand uncertainty means the apparent gap may be temporary or measurement-driven. These can coexist, so record evidence and confidence for each rather than selecting one label by intuition.

Training is appropriate when the capability is observable, the target group can plausibly acquire it, the role will use it and another constraint will not dominate. If the bottleneck is licensing, geography or management, a course can improve a learning metric without improving fill or performance. The ILO's May synthesis also cautions against a narrow technical-skills lens: rounded profiles and complementary policy matter, but that report does not independently reproduce the OECD indicator.

Set a review date aligned with the data lag. A 2024 indicator can inform structural investigation but should not be presented as a current vacancy count. Recheck local flow measures monthly and the external composite when revised. Preserve changes to classification, components and weights so an apparent trend is not a method change.

For cross-country use, compare method coverage and missing signals before interpreting a rank. Publish missingness and comparability notes beside the result. The country indicator can help select a question; it cannot establish that one employer faces the same tasks, pay and mobility.

This extends the newsroom's skills-visibility evidence note. The immediate decision is to require a one-page six-signal stack and friction diagnosis before approving shortage-led training spend.