Automation and AI are tilting production toward technological capital, shrinking labour’s share and eroding payroll‑based revenues; policymakers must rethink tax bases and pension designs or face mounting fiscal and social strain.

Main Finding

The paper documents a structural shift in twentieth- to twenty‑first‑century production: digital automation, robots, and AI are transforming the traditional production function Y = f(L, K, T, E) into a version increasingly dominated by technological capital (termed K_T). As K_T rises, the share of output and income attributed to human labor falls, reducing employment and wage contributions even as productivity grows. These shifts create persistent social and fiscal strains—weakening payroll- and labor‑based revenue streams, threatening pay‑as‑you‑go (PAYG) pension viability, and forcing a rethinking of taxation, redistribution, and the social compact in economies where value creation depends less on human work.

Paper metadata: received 02 Dec 2025; accepted 31 Dec 2025; published Jan 2026.

Key Points

• Conceptual shift: The authors argue that the classical production function is being replaced in practice by a K_T‑dominated technology where technological capital substitutes for labor rather than just complementing it.
• Labor share decline: As firms substitute K_T for L, labor’s share of income and its role in employment decline even when aggregate productivity rises.
• Distributional impacts: Reduced labor share disproportionately affects lower- and middle‑skill workers, increasing inequality and compressing wage-based tax bases.
• Fiscal consequences: Shrinking payrolls and labor incomes undermine revenues for PAYG pensions and other social programs financed by labor taxation; at the same time, corporate and wealth accruals concentrate on owners of K_T, changing the optimal tax base.
• Political/social effects: These economic shifts strain the social compact—raising pressure for new redistribution mechanisms, reconfigured social insurance, and debates over legitimacy of incumbent institutions.
• Policy tensions: Standard responses (e.g., retraining) may be necessary but insufficient; more structural fiscal and institutional reforms (tax base changes, pension design, universal transfers) are required.
• Heterogeneity and timing: Effects vary across industries, countries, and cohorts—early adopters and capital‑rich firms/countries gain most; transition dynamics matter for short‑run political economy.

Data & Methods

• Empirical strategy:
  • Measurement: The paper constructs proxies for K_T intensity (robot/automation density, software and intangible capital accumulation, AI adoption surveys, patenting in AI/automation) and maps these to firm- and industry-level labor shares and employment outcomes.
  • Macro decomposition: Growth‑accounting exercises decompose output growth into contributions from labor, traditional capital, and technological capital (K_T).
  • Econometrics: Panel regressions at firm and industry levels estimate the effect of K_T intensity on employment, wages, and factor shares, with controls for demand, input prices, and observable firm characteristics. Identification uses variation in technology exposure across industries and timing of adoption; robustness checks include difference‑in‑differences and instrumenting adoption with plausibly exogenous shocks (e.g., cross‑border technology diffusion, trade shocks).
• Structural modeling:
  • The authors build a dynamic general equilibrium (overlapping generations or representative‑agent with production) model that allows for substitution between labor and K_T, endogenous factor shares, and a PAYG pension sector. They calibrate/simulate the model to quantify effects on wage income, pension sustainability, and public budgets under alternative adoption trajectories.
  • Policy experiments: Simulations explore reforms (capital taxes, payroll tax adjustments, universal transfers, prefunded pensions) and their efficacy in restoring fiscal sustainability and distributional targets.
• Robustness and validation:
  • Cross‑country comparisons and case studies validate the empirical patterns.
  • Sensitivity analyses vary substitution elasticities, adoption speeds, and parametrizations of redistributive policies.

Implications for AI Economics

• Rethinking tax bases:
  • Labour-based taxes (payroll, income) become less reliable as K_T grows. Policymakers should consider shifting part of the tax burden toward capital, rents from K_T, consumption, or wealth taxes while being mindful of avoidance/evasion and growth effects.
  • Design choices matter: taxes on returns to K_T (corporate, property, rent) can help internalize social costs of displacement but require international coordination to limit capital flight.
• Pension and social insurance reform:
  • PAYG systems face sustainability risks as contributor pools shrink; options include raising contribution rates, cutting benefits, increasing retirement ages, or moving toward partial prefunding/reserves.
  • Alternative architectures—universal basic income, negative income tax, or universal transfers financed by K_T rents—warrant consideration to maintain social cohesion.
• Redistribution and labor policy:
  • Active labor market policies (reskilling, job search assistance) remain important but may not fully offset structural job losses where K_T substitutes for large swaths of tasks.
  • Policies should combine human-capital investments with income supports and stronger safety nets to manage transitional and structural unemployment.
• Industrial and innovation policy:
  • Public investment in sectors that are complementary to human labor (care, creative services, high-touch services) can create new employment opportunities.
  • Consider conditional public support for automation (e.g., taxes or obligations linked to retraining or local employment preservation).
• Political economy and governance:
  • The transition amplifies political risks: rising inequality and fiscal strain can fuel populism, protectionism, or destabilizing redistribution demands. Transparent, credible policy packages are critical.
• Research agenda and limitations:
  • Quantifying K_T: better measurement of AI/software capital, task substitution elasticities, and firm-level returns to K_T remains a priority.
  • Distributional microdata: more granular microdata (matched employer‑employee, wealth by ownership of K_T) are needed to assess incidence of capital rents and policy impacts.
  • International spillovers and coordination: cross-border effects of K_T adoption and tax competition require further modeling.
  • Transition dynamics: policies must address short- to medium‑run frictions (retraining lags, capital reallocation) in addition to long‑run equilibria.

Overall, the paper argues that accelerating AI and automation produce a durable shift toward technological capital–dominated production with important macroeconomic, distributional, and fiscal consequences. Policymakers in AI economics must broaden the focus beyond productivity gains to manage changing factor shares, preserve revenue bases, and redesign social contracts.