OECD employment remained near historic highs in early 2026, yet structural labour shortages persisted while recent graduates in Australia, Canada, the European Union and the United States faced a widening unemployment gap. The apparent contradiction reflects several different problems involving skills, experience, recruitment and geography, rather than one economy-wide shortage of educated workers.
The OECD Employment Outlook 2026 put the average employment rate across its member countries at 72.1% in the first quarter of the year. Unemployment was 4.9% in May.
Those headline figures were strong, but momentum had weakened. The median national unemployment rate increased from 5.4% in May 2025 to 5.7% a year later, while employment growth flattened and labour shortages continued to ease from their post-pandemic peak.
Young people felt more of that deterioration. Among recent university graduates aged 23 to 29, the unemployment gap with the wider working-age population had been rising since before the pandemic in every economy examined by the OECD: Australia, Canada, the EU as a weighted average and the United States.
Different statistics describe different problems
It helps to separate four terms that are often grouped together as a “skills mismatch”.
A skills shortage occurs when employers have difficulty recruiting people with particular abilities. A skills gap exists when current employees lack some of the capabilities a business needs. A qualification mismatch occurs when somebody’s education is above, below or in a different field from the requirements of a job. A geographic mismatch places suitable workers and vacancies in different locations.
The International Labour Organization uses skills mismatch as a broad term covering several of these situations. It also warns that different forms can coexist. A person can be highly qualified and still lack a job-specific skill, while somebody with the required capability may live too far from the available work.
This helps explain part of the apparent paradox. Graduate unemployment measures whether people have work. An employer survey may measure difficulty filling particular vacancies. A company survey of existing staff may measure something else again. None of those statistics describes the whole labour market.
Employer surveys do not measure the same shortage
ManpowerGroup’s latest global survey found that 72% of employers reported difficulty finding the people they needed, down from 74% the previous year. The 2026 survey covered 39,063 employers in 41 countries.
That is evidence of widespread reported hiring difficulty, but it is not an official count of unfilled vacancies. It also cannot establish whether every problem reflects a genuine scarcity of workers. Pay, working hours, location, experience requirements and recruitment practices can all affect the number of suitable applicants an employer receives.
A separate OECD employer study examined firms in Hungary, Italy, the Netherlands, Portugal and the Slovak Republic using data collected in 2021 and 2022. More than one-third reported gaps between the skills they needed and those held by employees. Technical skills were most frequently mentioned, followed by problem-solving and teamwork.
However, the gaps were usually concentrated. Fewer than 5% of firms said most or all of their workforce lacked the necessary skills. Among businesses reporting gaps, 63% said existing employees faced heavier workloads, 50% reported higher operating costs and one-third said the problem limited technology adoption.
That is a different problem from a shortage of graduates. A business may have plenty of applicants but still need to train existing staff to work with new equipment, software or production methods.
Experience and credentials can narrow the talent pool
The difficult step for many graduates is moving from education into work. Employers value qualifications, but they also want evidence that candidates can apply knowledge in a workplace.
The World Economic Forum’s Future of Jobs Report 2025 found that 81% of surveyed employers expected to use work experience when assessing candidates between 2025 and 2030. Some 48% expected to use pre-employment skills tests, while 43% expected to continue using university degrees as a requirement.
These are employers’ stated plans rather than observed hiring decisions. Even so, they show why a degree may open fewer doors when it is not accompanied by work experience. An employer seeking somebody who can contribute immediately may regard a graduate programme or extensive on-the-job training as an additional cost.
That calculation can become self-defeating across the wider economy. If many businesses compete for experienced workers but few are willing to create that experience, recruitment difficulties can persist even when capable new entrants are available.
This has encouraged interest in skills-first hiring, which gives more weight to demonstrated abilities and less to degrees or previous job titles used as proxies. The OECD’s 2026 skills-first report says the approach can widen access, particularly for people whose skills were acquired through work, short courses or informal learning.
There is an important catch. Removing a degree requirement from an advertisement does not guarantee different hiring. Research cited by the OECD found that fewer than 4% of newly degree-free US job postings translated into additional hiring of people without degrees. Recruitment practices, assessment methods and managers’ decisions must change as well.
Some shortages begin with the job
Discussion of skills shortages often starts with what workers need to learn. Sometimes the job offer also needs attention.
A 2026 European Commission working paper found that persistent shortages can be both a constraint on productivity and a symptom of poor working conditions. Low pay and low-quality jobs were particularly relevant in occupations such as care and transport.
That does not mean every employer reporting difficulty should simply raise wages. Some occupations require years of training, licences or specialist experience that cannot be produced quickly. However, describing every recruitment problem as a shortage of skilled people can obscure the effect of pay, schedules, location or narrowly written job requirements.
Businesses can also develop the people they already employ. In the five-country OECD study, nine in ten firms with internal skills gaps used training and staff development. Recruitment was used by 49%, while 39% changed work practices, reassigned tasks or automated parts of production.
One 2025 National Bureau of Economic Research working paper shows why matching still matters after somebody is hired. Using quasi-random job assignments among new US Air Force enlistees, the researchers found that overqualified workers performed better on job-specific measures and were promoted more often, but were also more likely to leave. The military setting cannot be assumed to describe civilian workplaces, but the findings show that placing a capable person into any available job does not necessarily create a stable match.
Geography and timing keep workers and jobs apart
National figures can also hide large differences between local labour markets. According to the OECD’s regional analysis, the full range between the highest- and lowest-employment regions exceeded 20 percentage points in more than half of OECD economies. The average gap between regions in the top and bottom employment quintiles was 11.4 points.
Moving is not a frictionless solution. Housing costs, family responsibilities, a partner’s work and access to schools can prevent a suitable applicant from relocating. Remote work removes part of the barrier for some office jobs, but it cannot place nurses in hospitals, technicians beside machinery or construction workers on building sites.
Technology creates a timing problem as well. Courses take years to design and complete, while employers can introduce new software or production systems much faster. The OECD nevertheless cautions against treating generative artificial intelligence as the main explanation for the recent graduate employment problem. The deterioration began before companies started adopting large language models heavily, and the evidence on AI’s employment effects remains mixed.
Employer shortages and graduate unemployment can therefore coexist without representing the same failure. Producing more graduates will not fill every specialist vacancy, while asking unemployed graduates to collect more qualifications will not create suitable work. The harder task is improving the links between education, practical experience, employer training, recruitment and the places where jobs are being created.
