Specialized digital skills pay off: workers with vocational data, programming or automation expertise earn roughly 14% more—far above the 6% premium for general digital literacy—and the reward is largest in Korea's big Chaebol where the premium approaches 19%, while advanced skills help TVET graduates close the gap with university peers.

Main Finding

Workers with specialized vocational digital skills (e.g., data analysis, programming, automation control) earn substantially higher wages than workers with only general digital literacy. After controlling for education, experience, and demographics, specialized digital skills are associated with a wage premium of about 14.2%, versus a 5.8% premium for general digital skills. Returns are heterogeneous across firm types and education tracks: the premium is much larger in large Chaebol firms (≈18.7%) than in SMEs (≈9.5%), and advanced digital skills acquired via TVET substantially narrow the income gap with general higher-education graduates.

Key Points

• Precise measurement matters: Differentiating general digital literacy from specialized vocational digital skills reveals much larger economic returns for the latter.
• Magnitudes:
  • Specialized vocational digital skills → ≈ +14.2% wage premium.
  • General digital literacy → ≈ +5.8% wage premium.
  • Chaebol (large conglomerates) → ≈ +18.7% for specialized skills.
  • SMEs → ≈ +9.5% for specialized skills.
• Skills-scale complementarity: Larger firms capture greater returns to specialized digital skills, suggesting complementarities between worker skills and firm scale/technology adoption.
• TVET implication: For graduates of Technical and Vocational Education and Training, acquiring advanced digital skills markedly reduces the earnings gap relative to graduates of general higher education, supporting the value of vocational upskilling.
• Policy relevance: Findings support prioritizing vocational digital-skill development as part of Korea’s Digital New Deal and for other East Asian economies facing similar labor-market polarization pressures.

Data & Methods

• Data source: Korean Labor and Income Panel Study (KLIPS) micro-data (panel microdataset used to analyze individual wages and characteristics).
• Empirical model: Extended Mincerian wage equation augmenting standard human-capital controls with indicators for general digital literacy versus specialized vocational digital skills.
• Controls included: Years of formal education, work experience, demographic characteristics (age/sex/etc.).
• Analyses performed:
  • Estimation of wage premia for different digital-skill categories.
  • Heterogeneity analysis by firm size/type (Chaebol vs SMEs).
  • Subsample analysis by education track (TVET graduates vs general higher-education graduates).
• Note: Results reported are conditional associations from regression analysis (controls included). The summary reflects the paper’s empirical findings; causal interpretation depends on identification strategy and robustness checks reported in the full paper.

Implications for AI Economics

• Returns to task-specific digital skills: AI and advanced digital technologies raise the value of specialized, task-oriented digital competencies more than general computer literacy. Models of labor demand should account for heterogeneity in skill content, not just aggregate “digital” possession.
• Complementarity with firm capital/scale: Higher returns in large firms imply complementarities between firm technology adoption (including AI) and worker skills. Predictions about wage inequality and distributional impacts of AI must model firm-level heterogeneity.
• Policy design for upskilling: Public investments that target vocational digital skill formation (TVET modernization, industry-aligned certifications, on-the-job training) can be high-return interventions to raise wages and reduce polarization.
• SME-targeted support: SMEs capture smaller returns to advanced skills; policies should combine skill training with measures that help SMEs adopt complementary technologies and reorganize tasks to exploit those skills.
• Labor-market dynamics: Upskilling in vocational digital skills can narrow traditional education-based wage gaps, suggesting a route to mitigate stratification driven by automation and AI diffusion.
• Research directions: Incorporate task-based measures of digital/AI skills into wage models, estimate causal impacts of training programs, and study how firm-level AI investments interact with worker skill premia to shape distributional outcomes.