AI does not imply a universal trade-off between jobs and productivity; instead, sectors face a ‘trilemma’ in which technical substitutability, market demand, or labor bargaining will bind, producing distinct patterns of displacement, scaling, or employment preservation and implying that policy must be tailored at the sectoral level.

Main Finding

The article rejects the simple productivity-vs-employment "AI dilemma" and proposes the Sectoral Trilemma: at the sector level, three objectives—rapid productivity growth via algorithms, employment stability or growth, and scalable output expansion—cannot all be achieved simultaneously. Instead, algorithmic adoption forces sectors into distinct regimes determined by which of three constraints binds: task substitutability (technical), price elasticity of demand for outputs (market), and institutional bargaining power over labor (political). The observed pattern—concurrent technological progress with wholesale clerical displacement in some industries—is a systematic outcome of how sectors resolve this constrained-optimization problem.

Key Points

• The AI dilemma (a uniform trade-off between productivity and employment) is a categorical fallacy; impacts are sector-specific.
• Three objectives form an unavoidable trilemma at the sectoral level:
• Fast productivity growth (through substituting or augmenting labor with algorithms),
• Stable or rising employment,
• Scalable output growth (ability to expand sales/production without proportional cost increases).
• A sector’s regime is defined by which constraint binds:
  • Technical substitutability: how easily tasks can be automated by algorithms.
  • Price elasticity of demand: whether increased output can be absorbed by demand at reasonable prices.
  • Institutional bargaining power: the degree to which wages, working conditions, and headcounts are protected/negotiated.
• Different combinations produce distinct outcomes (e.g., clerical-heavy sectors with high substitutability and weak bargaining → heavy displacement; sectors with inelastic demand or strong bargaining → employment-preserving adjustments).
• The framework synthesizes task-based and institutional approaches into a single diagnostic tool, shifting analysis from aggregate net effects to sectoral regime classification.
• Policy and firm strategy should be sector-tailored: targeted labor interventions, organizational design choices, and demand-side policies follow from identifying the binding constraint.

Data & Methods

• The article is primarily theoretical and synthetic: it integrates literature from task-based automation, demand-side analysis, and institutional/bargaining models to formulate the Sectoral Trilemma.
• Conceptual model: sectors face a constrained-optimization problem where algorithmic technology changes the production frontier and hence which of the three constraints binds.
• Suggested empirical inputs and operationalizations for diagnostics (for researchers testing or applying the theory):
  • Task substitutability: measures of routine/task routineness, occupation/skill task maps, AI-readiness scores, task-level automation probabilities.
  • Price elasticity of demand: sectoral/industry demand elasticities from price-quantity data, sales responses to price/quality changes, cross-sectional consumption patterns.
  • Institutional bargaining power: union density, collective bargaining coverage, minimum wage and labor regulation strictness, incidence of wage-setting institutions.
• Recommended empirical methods to validate and use the framework:
  • Sector-level panel regressions linking algorithm adoption (patents, AI tool diffusion, IT capital) to outcomes (employment, wages, output).
  • Difference-in-differences and event studies exploiting staggered AI uptake or regulatory changes.
  • Instrumental variables for adoption (e.g., exposure to relevant patents, technology shocks) to address endogeneity.
  • Case studies and firm-level analyses to trace mechanism channels (task change, hiring practices, price strategies).
  • Structural or calibrated models (e.g., constrained-optimization at sector level) and simulations to explore regime dynamics under varying parameter values.
• Data sources that map well to the framework: industry-level national accounts, firm-level administrative data, occupation-task surveys (O*NET, PIAAC), patent/AI adoption databases, labor institution datasets.

Implications for AI Economics

• Analytical shift: prioritize sectoral regime analysis over aggregate employment forecasts. Aggregate net effects mask large cross-sectoral reallocation and welfare heterogeneity.
• Policy diagnostics: policymakers should identify the binding constraint in each sector to choose effective interventions:
  • If technical substitutability binds: focus on retraining, task redesign, income supports, and incentives for job-creating product lines.
  • If demand elasticity binds: use demand-stimulating policies (subsidies, public procurement, product-market regulation) to preserve employment while allowing productivity gains.
  • If bargaining power binds: adjust labor institutions (collective bargaining expansion, wage floors, hiring protections) or employer-side regulations to distribute gains.
• Firm strategy: organizational design (whether to scale output, substitute labor, or preserve jobs) should be explicit about which constraint the sector faces; firms can leverage AI to pursue alternate equilibria (e.g., boost demand through new products rather than reduce headcount).
• Measurement and forecasting: more granular monitoring of AI diffusion, sectoral demand responses, and institutional changes is essential for credible forecasts and timely policy.
• Research agenda: empirical testing of the trilemma across sectors; development of diagnostic tools and indices for the three constraints; analysis of transition dynamics when sectors shift regimes (e.g., due to policy change or demand shocks).
• Normative considerations: the framework clarifies trade-offs and helps target redistribution or adjustment policies to sectors and groups most likely to be harmed by algorithmic restructuring, rather than one-size-fits-all remedies.

Limitations and open questions (for researchers and policymakers) - The theory requires operational measures for the three constraints; measurement error and causal identification remain challenging. - Interactions across sectors (input–output linkages) may propagate regime outcomes beyond single sectors and should be modeled. - Dynamic transitions—how sectors move between regimes over time as institutions, demand, and technology co-evolve—need formal modeling and empirical study.