The AI Layoff Trap
The paper identifies a demand-side externality in AI-driven automation: when firms displace workers, they capture full cost savings but externalize the demand destruction to rivals. In competitive markets, this creates a Prisoner's Dilemma where rational firms over-automate beyond the collective optimum, generating deadweight losses for both workers and owners. The analysis shows that only a Pigouvian tax on automation can correct this failure, while UBI, capital taxes, and worker equity programs cannot.
This paper delivers a sharp theoretical contribution by identifying a novel market failure in automation—the 'AI Layoff Trap'—where competitive pressures drive firms to automate beyond the social optimum because each firm bears only $1/N$ of the demand loss it creates. The model is analytically tractable with closed-form solutions, and the policy analysis is comprehensive in demonstrating why standard remedies (UBI, wage adjustments, profit-sharing) fail. However, the results depend critically on strong assumptions: quadratic automation costs, linear demand, zero MPC for owners in the baseline, and a static one-shot game structure. While the extensions partially relax these, the empirical validation relies on illustrative layoff anecdotes rather than systematic causal evidence of the externality.
The demand externality mechanism is logically unassailable and formally derived. Proposition 1 establishes that the wedge between Nash equilibrium $\alpha^{NE} = (s-\ell/N)/k$ and the cooperative optimum $\alpha^{CO} = (s-\ell)/k$ equals $\ell(1-1/N)/k > 0$, strictly increasing in $N$. The policy analysis in Sections 4.1–4.6 rigorously distinguishes instruments that shift profit levels (UBI, capital taxes) from those operating on the per-task margin. The 'Red Queen effect' in Proposition 6 is particularly important: higher AI productivity $\phi > 1$ widens rather than narrows the wedge because market-share motives amplify automation incentives while the planner's optimum remains unchanged.
The model assumes specific functional forms—quadratic integration costs $k\alpha_i^2/2$ and linear demand—that yield corner solutions and clean comparative statics but may not generalize to arbitrary convex costs or non-CES preferences. The baseline assumption that owners have zero marginal propensity to consume (MPC) in the sector is strong; while Section 5.4 relaxes this, the externality persists under plausible parameter restrictions requiring implausibly high recycling rates. More critically, the one-shot game structure ignores repeated interaction: in a dynamic setting with trigger strategies, firms might sustain the cooperative outcome without taxation, especially when automation decisions are persistent and observable. Finally, the empirical evidence cited (Block, Salesforce layoffs) illustrates displacement but does not demonstrate the profit erosion or deadweight loss central to the model's predictions.
The paper effectively distinguishes its mechanism from Acemoglu & Restrepo's task-based labor market framework and Beraja & Zorzi's (2025) credit-constraint externality, correctly noting that its product-market demand spillover requires competition and persists even with complete credit markets. The comparison to Murphy, Shleifer & Vishny's 'big push' models as a mirror image (individually profitable but collectively destructive automation versus individually unprofitable but collectively profitable investment) is apt. However, the paper does not empirically validate the key comparative static—that more competitive industries (higher $N$) exhibit larger over-automation wedges relative to concentrated markets. The cited layoffs demonstrate displacement intensity but do not establish the causal link between fragmentation, demand externalities, and profit erosion predicted by Proposition 2.
As a theoretical contribution, the paper provides complete analytical derivations in Appendix A with closed-form solutions for all propositions. No computational code, data, or simulation scripts are provided or required. The numerical illustrations (Figures 1–2) use illustrative parameters ($c/w=0.30$, $\lambda=0.5$, $\eta=0.30$, $N=7$) rather than calibrated values from empirical moments. Extensions to endogenous entry, wages, and AI productivity (Section 5) are purely analytical. Independent reproduction would require only algebraic verification of the first-order conditions and comparative statics; the quadratic cost structure and linear demand ensure global concavity and unique equilibria without numerical computation.
If AI displaces human workers faster than the economy can reabsorb them, it risks eroding the very consumer demand firms depend on. We show that knowing this is not enough for firms to stop it. In a competitive task-based model, demand externalities trap rational firms in an automation arms race, displacing workers well beyond what is collectively optimal. The resulting loss harms both workers and firm owners. More competition and "better" AI amplify the excess; wage adjustments and free entry cannot eliminate it. Neither can capital income taxes, worker equity participation, universal basic income, upskilling, or Coasian bargaining. Only a Pigouvian automation tax can. The results suggest that policy should address not only the aftermath of AI labor displacement but also the competitive incentives that drive it.
Pick a starting point or write your own. Challenges run in the background, so you can keep reading while the AI investigates.
No challenges yet. Disagree with the review? Ask the AI to revisit a specific claim.