Demographic ageing and AI together threaten to hollow out labour’s historic role as the engine of consumption and tax revenue; the proposed 'labour liability ratio' warns that shrinking workforces and cognitive automation could compress fiscal and demand foundations of welfare states, forcing a rethinking of ownership, distribution and economic governance.

Main Finding

The paper introduces the labour liability ratio (LLR), a conceptual heuristic that captures how demographic ageing and advanced AI-driven automation jointly compress labour’s systemic role in advanced capitalist economies. The LLR shows that ageing raises the fiscal and care-related “liabilities” of labour while AI reduces labour’s contribution to output, demand formation, and tax bases. Together these pressures can destabilize the labour‑centred accumulation regime (Fordism/post‑war welfare capitalism) and reconfigure distribution, ownership, and geopolitical patterns of accumulation.

Key Points

• Problem framed: twentieth‑century macro‑stability depended on labour acting simultaneously as producer, consumer and fiscal base. Demographic ageing + AI threaten that nexus.
• Labour Liability Ratio (LLR): a heuristic ratio contrasting labour’s macroeconomic contribution (numerator) with the demographic, fiscal and biophysical costs of sustaining labour‑intensive societies (denominator).
  • Numerator (labour contribution): Lo (output), Ld (demand via wages/consumption), Lt (tax revenues).
  • Denominator (liabilities/costs): Fd (fiscal dependency costs – pensions/welfare), Br (biophysical/material costs of reproducing labour power), Hc (healthcare/long‑term care burdens).
• Compression dynamic: LLR falls because
  • Labour-related costs rise with ageing (higher pensions, healthcare, long‑term care; Baumol’s cost disease in care sectors).
  • Labour’s relative contribution weakens due to AI and robotics automating cognitive and physical tasks, and falling aggregate labour intensity (fewer labour hours, longer education, earlier retirement).
• Demographic inversion specifics: persistently below‑replacement fertility and rising life expectancy in many advanced economies → shrinking working cohorts and rising old‑age dependency; illustrative projections show large population declines under low‑fertility, zero‑migration scenarios.
• Limits of conventional responses: pro‑natalist policies and standard dependency ratios are insufficient because they assume labour remains central to production; automation can decouple population size from productive capacity.
• Political‑economic implications: potential reconfiguration of distribution (capital captures more of technological surplus), debates over ownership and taxation of AI capital, need for new fiscal instruments, risks of core–periphery divergence as automation changes pathways for industrialisation.
• Methodological stance: the LLR is a theoretical/heuristic tool (not an econometric index) rooted in Regulation School political economy and synthesising demographic economics, automation studies and world‑systems perspectives.

Data & Methods

• Research design: conceptual/theoretical synthesis rather than formal econometrics. The author develops a macro‑political‑economy framework (LLR) and uses empirical literature and illustrative demographic projections to support the argument.
• Key data sources and empirical inputs cited:
  • United Nations Department of Economic and Social Affairs (long‑run population projections).
  • OECD reports on fertility, labour hours, tax wedges, and dependency ratios (2024).
  • IMF analysis on AI exposure of employment (Cazzaniga et al., 2024).
  • Automation and AI literature: Brynjolfsson, Li & Raymond (2023); Acemoglu & Restrepo (2020); Eloundou et al. (2023).
  • International Federation of Robotics (2024).
  • World Health Organization on healthy life expectancy (2024).
  • Historical and political‑economy literature: Aglietta, Boyer, Keynes, Kalecki, Polanyi, Piketty.
  • Sectoral cost disease literature: Baumol (1967).
  • Illustrative table of population projections under low‑fertility, zero‑net‑migration scenarios (EU, Italy, US) is provided for plausibility (not a forecast tool).
• Methods summary:
  • Literature synthesis across demographic economics, automation studies and Regulation School political economy.
  • Heuristic formalisation of LLR (symbolic expression of numerator and denominator components).
  • Qualitative scenario reasoning about how demographic and technological trends interact to compress the LLR.
• Limitations noted by author:
  • LLR is intentionally heuristic and not a precise, directly measurable statistic.
  • The paper does not present new microdata econometrics or causal estimation of AI impacts; it frames questions and dynamics for further empirical work.

Implications for AI Economics

• Rethink the baseline: AI economics should integrate demographic structure (ageing, fertility, migration) into models of automation impact — not only labour market displacement but also demand, taxation and fiscal sustainability.
• Demand formation matters: models that focus purely on productivity gains risk missing second‑order macro effects if wage‑mediated consumption (labour demand) declines; AI could shrink aggregate wage income even while GDP rises in value.
• Fiscal modeling: incorporate declining labour‑based tax bases and rising care/health expenditures into assessments of long‑run public finance under automation scenarios. This motivates research on capital taxation, robot/AI taxes, and alternative fiscal instruments (e.g., ownership shares, sovereign wealth for technology rents).
• Sectoral heterogeneity: care, health and education sectors are subject to Baumol’s cost disease and may remain labour‑intensive; AI’s differential substitutability across sectors should be explicitly modeled (limited productivity gains in some high‑demand public services).
• Capital deepening and distribution: demographic contraction can increase capital per worker (capital deepening) and partially offset labour displacement — but gains may accrue to capital owners unless redistributive policies alter ownership of technological surplus. Empirical work should estimate incidence of returns (labour vs capital) under different ownership regimes.
• Global and developmental consequences: automation reduces one traditional development pathway (labour‑intensive industrialisation). AI economics must consider global division of labour, technological sovereignty, and how automation reshapes core–periphery dynamics.
• Measurement and empirical agenda:
  • Operationalize the LLR into measurable components (e.g., combine labour share, aggregate hours, tax receipts from labour, age‑structured public spending, and measures of labour reproduction cost) and test cross‑country/time trends.
  • Build integrated macro models and scenarios that combine demographic projections, sectoral AI substitutability matrices, and public finance modules.
  • Study firm‑level adoption of AI conditional on local demographic structures (labour scarcity vs labour surplus).
  • Evaluate policy interventions (UBI, negative income tax, capital taxation, public or cooperative ownership of AI platforms) in general equilibrium settings that include demand feedbacks from reduced labour income.
• Policy relevance for AI economists: evidence on the LLR dynamics strengthens arguments for:
  • Broader distributional policies tied to technological rents.
  • Investment in AI that complements care/health where feasible, while ensuring public provisioning where automation is limited.
  • International policy coordination to manage global inequality effects of automation.

Reference note: Philipp Humm, "The Labour Liability Ratio: Demographic Ageing, Artificial Intelligence and the Transformation of Labour‑Centred Capitalism", research article (posted Aug 6, 2026). DOI: https://doi.org/10.21203/rs.3.rs-10117242/v1.