AI adoption cuts high-skilled workers' relative wage-cost by roughly 7% in European regions, driven equally by wage and employment declines; the effect is concentrated where regions are highly specialized in AI, suggesting AI may dampen — not amplify — labor-market polarization.

Main Finding

AI adoption reduces the relative wage-cost position of high-skilled workers versus low-skilled workers by about 7% on average across European regions, with roughly equal contributions from wage adjustments and employment adjustments. Effects are concentrated in regions within countries highly specialized in AI technologies. Overall, this suggests AI may weaken — rather than reinforce — long-run labor-market polarization.

Key Points

• The authors develop a CES (constant elasticity of substitution) production framework that links AI to the wage-cost ratio between high- and low-skilled workers through two margins:
  • Wage margin: changes in relative wages.
  • Employment margin: changes in relative employment shares.
• Empirical estimates using European regional labor-market data find an average ~7% reduction in the high/low wage-cost ratio associated with AI.
• The observed decline is broadly split between reductions in relative wages and reductions in relative employment of high-skilled workers.
• Heterogeneity: effects are primarily driven by regions in countries that are highly specialized in AI technologies, indicating local exposure/intensity matters.
• Interpretation: instead of amplifying the historical trend toward labor-market polarization, AI appears to counteract it in the studied setting.

Data & Methods

• Theoretical approach: a CES production model that maps AI adoption into changes in the relative wage-cost of high- vs. low-skilled labor via wage and employment margins.
• Empirical approach: tests of the model’s predictions using regional labor-market data for Europe. Key outcome measures include the relative wage-cost ratio between skill groups and its decomposition into wage and employment components.
• Identification strategy (as described): exploits variation in regional exposure to AI and compares changes in relative wages and employment across regions; conducts heterogeneity analysis by regions’ AI specialization.
• Main empirical results: a near-7% average reduction in the high/low wage-cost ratio, with effects split between wage and employment channels and concentrated in AI-specialized regions.

Implications for AI Economics

• Distributional effects: AI can weaken the relative labor-market position of high-skilled workers, implying redistributive pressures that differ from the canonical “skill-biased technical change” narrative.
• Polarization dynamics: findings suggest AI may mitigate long-run polarization trends that have unfolded over the past two centuries, at least in the European regional context studied.
• Policy relevance:
  • Labor-market policies should consider both wage and employment adjustments when assessing AI’s impact.
  • Regional and sectoral exposure matter — policies could target regions with high AI specialization for retraining, mobility support, or complementary investments to manage adjustment costs.
  • Monitoring skill demand shifts remains important to anticipate and respond to heterogeneous local impacts.
• Research directions: further work should examine mechanisms behind the employment margin (task reallocation, occupational transition), long-term dynamics, and generalizability beyond Europe.