National labour institutions, not just technological readiness, determine whether AI delivers corporate gains; strong unions and collective bargaining can blunt AI-driven restructuring, forcing multinationals to weigh labour relations and social license alongside tech deployment.

Main Finding

The paper develops a resource–institutional compatibility framework showing that the value of AI for firms—especially MNEs—is contingent on national labour institutions. Strong union power, collective bargaining coverage, and related institutional features can blunt or even reverse AI’s expected value by provoking negotiation, protest, reputational and regulatory responses. Thus AI is not a universally transferrable strategic asset: its return depends critically on institutional legitimacy and social license in host countries.

Key Points

• Conceptual contribution: Introduces "resource–institutional compatibility" — the degree to which firm-level AI capabilities are congruent with host-country labour institutions — as a necessary condition for AI to yield strategic value.
• Core propositions:
• AI adoption enhances competitive advantage at the firm level (standard RBV logic).
• The positive value impact of AI is larger for MNEs (due to cross-border knowledge integration and internalization).
• The positive effect of AI on firm value is attenuated (weakened) in countries with stronger labour union power.
• Mechanisms identified:
  • Negotiation–protest–regulatory feedback pathway (typical of advanced economies with strong unions): AI-driven restructuring triggers collective bargaining, strikes, regulatory pushback → higher implementation costs, delays, reputational risk → lower realized AI returns.
  • Absorption–cooperation–stabilisation pathway (more common in emerging markets with weaker unions): faster reconfiguration, less conflict, quicker capture of AI gains.
• Empirical examples and motivation: European cases where unions shape AI governance (e.g., Germany), recent strikes in transport/health/education demonstrating labour mobilization potential, contrast with many developing countries where collective action is limited.
• Conceptual risks: Without social legitimacy, AI can produce suboptimal outcomes (the paper coins the term "Artificial Idiocy" to describe failure to adapt/evolve because of institutional resistance).
• Managerial implication highlighted: MNEs must incorporate labour relations, union strength, retraining and benefit-sharing into AI deployment strategy—technical readiness alone is insufficient.

Data & Methods

• Nature of the study: The paper is conceptual/theoretical. It synthesizes literature across RBV, institutional economics, labour relations, and recent AI studies to build the resource–institutional compatibility framework and derive testable propositions.
• Methods used: narrative literature review and theoretical integration; development of formal propositions and causal pathways rather than empirical estimation.
• Suggested empirical operationalizations (implicit in the paper and useful for follow-up work):
  • AI adoption measures: AI-related investment, AI patents, AI-related job postings, use of algorithmic systems.
  • Labour-institution measures: union density, collective bargaining coverage, legal strike protections, historical strike frequency, labour law indices.
  • Outcomes: firm value (Tobin’s Q, market capitalization), productivity, employment changes, implementation cost overruns, reputational metrics.
  • Empirical strategies that would test the propositions: cross-country panel regressions with MNE fixed effects, difference-in-differences around AI adoption events, event studies of AI announcements interacting with country-level union measures, instrumental variables (e.g., exogenous variation in AI availability or sectoral AI shock), and case studies of negotiated co-governance arrangements (e.g., German works councils).
• Limitations: no original data or quantitative tests in the paper; claims are presented as propositions and conceptual mechanisms to motivate empirical work.

Implications for AI Economics

• Valuation and risk: Financial-valuation models that capitalize AI as an intangible should incorporate institutional risk (labour conflict, regulatory backlash) as an additional discount or risk premium. Expected cash flows from AI investments are endogenous to labour-institutional responses.
• Cross-border deployment strategy: MNEs cannot assume uniform returns to AI across countries. Optimal global AI strategies must consider host-country labour regimes, union strength, and social license; in high-union contexts, increased costs for negotiation, retraining, benefit-sharing, and slower restructuring should be anticipated.
• Policy and governance: Unions and labour institutions will be important actors shaping AI governance. Policies encouraging co-design, social dialogue, upskilling and benefit-sharing can increase AI’s legitimacy and therefore its realized value.
• Research agenda for AI economics:
  • Quantify how much institutional frictions cut into AI-derived productivity gains and market valuation.
  • Estimate heterogeneity of AI returns across labour regimes and legal environments.
  • Study the effectiveness and valuation consequences of mitigation strategies (retraining, profit-sharing, negotiated rollouts).
  • Incorporate labour-institution variables into asset-pricing and investment-cashflow models when assessing AI investments.
• Managerial takeaways: To maximize AI value, firms (especially MNEs) should invest in pre-deployment stakeholder engagement (unions/government), co-design workplace AI governance, allocate resources for retraining and compensation schemes, and tailor deployment timing/scale to host institutional compatibility.