Research · Per Ardua

Three Headlines from One Equation: A Replication and Sensitivity Analysis of “The AI Layoff Trap” (Falk and Tsoukalas, 2026)

The math does not adjudicate between them; the inputs do

AP-8 Applied DOI SSRN

Executive Summary

Falk and Tsoukalas (2026) present a task-based automation model in which a demand externality traps firms in an automation arms race, concluding that only a Pigouvian automation tax can correct the distortion. This paper replicates their model, verifies all ten propositions, and confirms the internal logic. It then shows that the model's conclusion is not a theorem about the world but a conditional statement whose sign and magnitude are determined by parameter choices the authors present as innocuous.

Under their baseline (η = 0.30, zero owner spending, homogeneous goods, no reinstatement feedback), the model produces catastrophe. Under equally defensible parameters reflecting standard features of real economies (η = 0.50, modest owner spending, product differentiation, endogenous reinstatement), the same model produces stability or even under-automation. Under optimistic but historically grounded parameters (η > 1, strong reinstatement), it prescribes an automation subsidy. Three parameterizations, three headlines, one equation.

The paper further documents that the model's baseline fixed-N, fixed-task structural closure — partially relaxed in the original's endogenous-entry extension, where the trap result persists — does not capture the empirically dominant margins of adjustment: U.S. business applications run 60% above 2019, AI startups captured 52% of global VC in 2025, and 60% of current U.S. employment is in occupations that did not exist in 1940. The demand externality mechanism is a genuine theoretical contribution. The policy prescription is a parameter choice.

Key Findings

  • Replication: All ten propositions verified; the internal logic holds.
  • Baseline parameters: Catastrophe, and a Pigouvian automation tax.
  • Standard-economy parameters: Stability or under-automation.
  • Optimistic, historically grounded parameters: An automation subsidy.
  • Missing margins: Endogenous entry adds more firms of the same type; the data show new firm types creating new occupational categories.

Code and Data

The paper source and replication code are archived with the paper at Zenodo: concept DOI 10.5281/zenodo.19619682 (resolves to latest).

Download Full Paper

Access the complete research paper with detailed methodology, empirical evidence, and formal proofs.

Download PDF