A new ESJS–Lightcast benchmarking method finds about 2% of EU vacancies likely suffer skill shortages, with substantial cross-occupation and cross-country variation; shortages cluster in occupations where skill content is evolving fastest, offering a replicable tool for targeted policy and AI-era monitoring.

Main Finding

The authors develop a replicable, objective measure of potential skill shortages by combining the 2021 European Skills and Jobs Survey (ESJS) with Lightcast job-advertisement data. Using this approach, they estimate that roughly 2% of vacancies in the EU are likely to experience skill shortages, with substantial cross-occupation and cross-country variation. Potential skill shortages concentrate in occupations where the skill content is changing more rapidly.

Key Points

• Skill shortages are defined here as vacancies that cannot be filled because suitably qualified/skilled candidates are lacking.
• The literature lacks consistent, replicable measures; this study addresses that methodological gap.
• Data sources: ESJS (skills and competencies currently used by workers) + Lightcast job ads (vacancy requirements).
• Novelty: benchmarking an objective measurement against an empirical distribution of skills actually used in the labour market.
• Estimated magnitude: ≈2% of EU vacancies likely face skill shortages.
• Heterogeneity: notable variation across occupations and across countries.
• Dynamic aspect: occupations with faster-changing skill content are more likely to show potential shortages.

Data & Methods

• Data:
  • 2021 European Skills and Jobs Survey (ESJS) — provides information on the current distribution of skills and competencies in the workforce.
  • Lightcast job-advertisement data — contains information on vacancy requirements and job postings across countries.
• Methodological approach:
  • Construct an objective, replicable measure of potential skill shortages by comparing skill requirements signalled in job ads to the observed distribution of skills and competencies in the labour force (from ESJS).
  • Benchmarking the measurement against an empirical skills distribution is presented as a methodological innovation allowing cross-country and over-time replication.
• Outcome: classification/estimation of vacancies likely to experience shortages and analysis of variation across occupations and countries; assessment of association between rapid change in skill content and likelihood of shortage.

Implications for AI Economics

• Monitoring & forecasting: Real-time vacancy and skill-use data (job ads + worker surveys) provide scalable tools to monitor skill mismatch as AI alters demand — useful for evaluating displacement vs. complementarity effects.
• Dynamic skill change: Occupations with rapidly changing skill content are more prone to shortages; AI-driven shifts (new tools, platforms, tasks) will likely accelerate such dynamics, increasing the need for timely measurement.
• Policy targeting: A low aggregate share (~2%) masks concentrated pockets of shortage — targeted reskilling/upskilling, apprenticeship incentives, or selective immigration policies may be more efficient than broad measures.
• Training & retraining investments: As AI changes task content, continuous training aligned to observed employer requirements (as captured in ads) can reduce mismatch.
• Research & evaluation: The paper’s replicable measurement approach can be used to evaluate the labour-market impact of AI adoption over time and across countries, and to assess whether AI increases or reduces skill shortages in specific occupations.
• Limitations to consider when applying findings to AI policy: measurement choices and data coverage affect the 2% estimate; rapid AI-induced changes could make historical benchmarks less predictive, underscoring the need for ongoing, high-frequency monitoring.