Automation is eroding income-tax receipts and increasing fiscal pressure, and a scoping review finds that a 'robot tax' could recoup revenue and finance retraining to soften worker displacement; the proposal, however, is conceptual and rests on heterogeneous, untested evidence.

Main Finding

The paper conducts a qualitative scoping review and concludes that a well-designed "robot tax" can serve as a Pigouvian-style fiscal instrument to (a) compensate for employment‑related income tax erosion caused by automation, (b) fund retraining/reskilling and social protection for displaced workers, and (c) correct existing tax-policy biases that incentivise capital (automation) over labour. Implementation faces definitional, legal, administrative, and innovation‑risk challenges, and empirical evidence on the net fiscal impact of automation remains limited.

Key Points

• Research approach

  • Pragmatic philosophy; qualitative scoping review using the Arksey & O’Malley framework.
  • Literature reviewed across three themes: (1) concept/definition/design of a robot tax, (2) critiques/alternatives, (3) automation’s impact on employment and public revenue.
  • Three theoretical lenses: Smith’s Four Maxims of Taxation, Christensen’s disruptive innovation, and Pigouvian tax theory (Pigouvian view preferred).

• Rationale for a robot tax

  • Automation can reduce personal income tax receipts by displacing workers, increasing fiscal pressure and inequality—particularly salient for developing economies (example focus: South Africa).
  • A robot tax can internalise the negative externality of job displacement, finance retraining programs, and slow the speed of displacement to allow labour market adjustment.

• Design options discussed

  • Direct levies on automation/robot usage (sometimes called an “automation tax”).
  • Neutralising capital incentives: reducing or removing tax deductions/accelerated depreciation and VAT input claims on automation capital (example: South Korea’s reduction in incentives).
  • Imputed income approach: tax the hypothetical salary a robot replaces (requires attribution and/or legal personality debates).
  • Alternatives or complements: payroll tax adjustments, targeted subsidies for labour‑intensive firms, universal basic income, and retraining tax credits.

• Practical challenges and critiques

  • Definitional ambiguity: difficulty in defining what counts as a robot/automation and which degrees of autonomy should be taxed.
  • Legal/personhood issues: taxing robots directly raises questions about juristic personality; policymakers may instead tax firms that use automation.
  • Risk of stifling innovation and productivity gains—primary political objection (EU Parliament rejection in 2017; counterarguments from proponents like Gates/Musk).
  • Measurement and attribution problems: robots often perform parts of tasks; isolating displacement effects is empirically hard.
  • Sparse empirical evidence tying automation to long‑run tax revenue declines; Hotte et al. (2022) find negative effects pre‑2008 but no effect after 2008, indicating stage‑dependent dynamics and the need for more research.

• Contextual emphasis

  • South African case: generous tax incentives for machinery and accelerated depreciation vs. no equivalent tax advantages for labour; high unemployment and growing labour costs heighten policy relevance for developing economies.

Data & Methods

• Methodology

  • Scoping review (Arksey & O’Malley framework) — qualitative synthesis of academic, policy, and international examples.
  • Cross‑country literature used to draw lessons applicable to developing economies.
  • Theoretical integration: applied taxation principles (Smith), innovation theory (Christensen), and Pigouvian externality correction.

• Evidence base limitations

  • No primary empirical estimation in the paper; relies on secondary studies, policy reports, and theoretical models.
  • Empirical literature on automation’s fiscal effect is limited and mixed; causal identification is difficult due to confounding factors (productivity gains, new job creation, policy changes).
  • Case examples (e.g., South Korea) noted but effect attribution is inconclusive.

Implications for AI Economics

• Policy design and evaluation

  • Treat robot/automation taxation as a Pigouvian tool: set revenues aside explicitly for retraining, social protection, and active labour market policies to internalise external costs.
  • Rebalance tax incentives: consider removing accelerated depreciation or VAT input claims for automation capital, or phase such changes to avoid abrupt effects on investment.
  • Design must handle definitional and administrative constraints: prefer firm‑level levies tied to measurable automation intensity (e.g., robot/unit installed, automation‑enabled output), or imputed wage frameworks if attribution methods are robust.
  • Pilot, evaluate, and iterate: small‑scale pilots and careful monitoring (sectoral, firm‑level) before nationwide rollouts to assess impacts on employment, innovation, and productivity.

• Research agenda for AI economics

  • Causal estimates: use quasi‑experimental methods (difference‑in‑differences, instrumental variables) and firm‑level administrative data to estimate the effect of automation adoption on taxable income, employment, and tax revenues.
  • Microsimulation and incidence analysis: model distributional effects across workers, owners, and consumers; assess who ultimately bears the tax (firms, consumers, shareholders, or labour).
  • Measurement methods: develop standardized metrics for automation intensity and task displacement, including partial‑task substitution.
  • Dynamic welfare analysis: quantify long‑run tradeoffs between short‑run job displacement and productivity gains, and incorporate them into optimal tax rate design.
  • Policy interaction studies: analyze how robot taxes interact with R&D incentives, trade, and foreign investment—particularly critical for developing economies sensitive to capital flows.

• Considerations for developing economies

  • Fiscal constraints and institutional capacity mean simpler, administrable instruments may be preferable (e.g., tightening capital allowances rather than complex imputed income rules).
  • Equity and political feasibility: framing (tax on automation use to fund jobs/retraining) may improve public acceptability compared with direct wealth or income taxes.
  • International coordination: differing national approaches risk investment displacement—coordination or harmonisation may reduce competitive distortions.
  • Complementary measures: investing in education, active labour market policies, and digital infrastructure is essential to convert automation challenges into inclusive growth opportunities.

Overall, the paper argues that robot/automation taxation is a defensible fiscal policy option to address negative labour‑market externalities from AI and robotics, but its practical success depends on careful design, empirical evaluation, and coordination with broader labour and innovation policies.