Graduate employability depends less on standalone skills than on whether employers believe a candidate's net value justifies higher local costs and cannot be readily replaced by lower-cost workers or AI; the proposed VALUE Framework offers practical levers—verifiable productivity, adaptive skill-stacking, local intelligence, unoutsourceable capabilities and employer‑relevant signalling—to lower perceived substitutability.

Main Finding

Graduate employability in high-cost labour markets should be reframed as employers’ cost-adjusted assessment of value versus perceived substitutability—not merely as graduates’ possession of skills. Employers decide hiring and pay by judging whether a graduate’s productivity (net of wage/cost) justifies the premium and whether the work can be replaced by lower‑cost labour or AI-enabled alternatives. Reducing employer-perceived labour substitutability is therefore the central pathway by which high-cost-market graduates remain competitive. The paper proposes the VALUE Framework (Verifiable productivity, Adaptive skill stacking, Local‑regional market intelligence, Unoutsourceable capabilities, Employer‑relevant signalling) as a practical model for lowering substitutability.

Key Points

• Reframing employability:
  • Move from skills-centric to cost-adjusted graduate value and employer-perceived labour substitutability.
  • Employers weigh (1) value generated relative to cost and (2) likelihood of substitution by lower-cost humans or AI.
• Theoretical contributions:
  • Introduces two constructs: cost-adjusted graduate value and employer-perceived labour substitutability.
  • Extends human capital and signalling theories to incorporate global labour arbitrage and AI-enabled task substitution.
  • Integrates literature across employability, signalling, HRM, global labour markets, and AI-enabled knowledge work.
• VALUE Framework (core practical model):
  • Verifiable productivity: documented, measurable output and impact (metrics, portfolios, trial projects).
  • Adaptive skill stacking: complementary, transferable combinations of skills that increase robustness across tasks and contexts.
  • Local‑regional market intelligence: contextual knowledge, networks, regulatory and cultural fluency that adds location-specific value.
  • Unoutsourceable capabilities: interpersonal, tacit, context-embedded, or time-sensitive skills less amenable to remote workers or automation.
  • Employer‑relevant signalling: credible cues (internships, projects tied to firm problems, certifications, employer endorsements) that reduce uncertainty about true value.
• Propositions and mechanisms:
  • Higher perceived substitutability lowers willingness to pay a local premium; lowering substitutability raises competitiveness.
  • AI and task automation magnify substitutability for decomposable, codifiable tasks but can be counterbalanced by AI-augmentation that increases verifiable productivity.
  • Strategic HR practices and university programmes that emphasize VALUE elements can shift employer perceptions and labor market outcomes for high-cost graduates.
• Practical guidance:
  • Universities: embed employer-aligned, evidence-based assessments and locality-relevant experiences; promote adaptive skill stacking and employer co-designed projects.
  • Employers/HR: design hiring and onboarding to capture verifiable productivity and signal unique value; use task decomposition to decide where to augment vs outsource.
  • Policymakers: support programmes that develop unoutsourceable capabilities, regional ecosystems, and measurement of graduate value; consider wage/contracting policies that reflect task substitutability risks.

Data & Methods

• Methodological approach: conceptual and integrative literature review and synthesis.
• Sources synthesised: graduate employability research, graduate capitals, human capital theory, signalling theory, strategic HRM, global labour arbitrage literature, and AI-enabled knowledge work studies.
• Case choice: Hong Kong used as a theoretically significant, high-cost labour market to motivate and illustrate arguments (contextual grounding rather than empirical testing).
• Output: a conceptual model, theoretical propositions, and the VALUE Framework. No original quantitative or qualitative empirical data were collected; the contribution is theoretical and prescriptive.

Implications for AI Economics

• Micro-level substitution dynamics:
  • Emphasises task-level substitutability as the key determinant of when AI and remote labor depress local wages or displace roles; not all skills are equally exposed.
  • Suggests models of automation should explicitly incorporate employer perception and signalling frictions—firms may under- or over-estimate substitutability.
• Complementarity and augmentation:
  • AI can raise verifiable productivity (making high-cost labour more attractive) or increase substitutability (for codifiable tasks). Policies and firm strategies that foster AI‑augmentation (not pure replacement) can preserve local employment premiums.
• Wage and employment outcomes:
  • Predicts heterogeneous wage effects across graduates depending on how successfully they reduce perceived substitutability via VALUE components—implying widening within-market inequality unless mitigated.
• Human capital investment and returns:
  • Returns to education/training should be modelled as net of locale-specific cost premiums and substitutability risk; investments in unoutsourceable capabilities and signalling mechanisms may yield higher risk‑adjusted returns in high-cost regions.
• Labor market institutions and policy:
  • Supports targeted interventions (reskilling, industry‑university partnerships, certification regimes) that lower information frictions and enhance local-specific value.
  • Suggests regulators and policymakers consider task‑level protections or incentives to balance global arbitrage and local economic stability.
• Research agenda for AI economics:
  • Operationalize and measure employer-perceived substitutability (surveys, vignette experiments, revealed-preference hiring data).
  • Task decomposition studies to estimate substitutability elasticities across occupations and regions.
  • Field experiments testing VALUE-based interventions (e.g., employer-aligned capstones, verified productivity signals) on hiring outcomes and wage premiums.
  • Dynamic models of how AI adoption interacts with regional labour costs, signalling, and training investments.

Overall, the paper urges AI economics research to move beyond binary “automation vs jobs” questions and to incorporate employer perception, signalling, local contextual value, and strategic augmentation as central variables shaping how AI and global labour arbitrage affect high‑cost labour markets.