Two economists have published a mathematical paper arguing that large-scale AI adoption could, under certain conditions, lead to a self-reinforcing economic downturn.
The peer-reviewed study is based on formal economic modeling, titled “The AI Layoff Trap” (March 2, 2026), and is authored by researchers affiliated with the Wharton School at the University of Pennsylvania and Boston University.
The paper’s central claim is that in a fully competitive market, widespread automation can trigger a feedback loop: firms replace workers with AI to reduce costs, other firms follow to remain competitive, and employment gradually declines across sectors.
Because displaced workers also function as consumers, reduced employment leads to weaker aggregate demand. In the model, this declining demand then incentivizes further cost-cutting and automation.
The authors describe this as a “layoff-demand feedback loop,” where rational firm-level decisions collectively produce a macroeconomic contraction.
The study evaluates several policy responses within its model It concludes that most standard interventions only partially mitigate the effect under assumed conditions.
The only intervention found to fully break the modeled feedback loop is a Pigouvian automation tax: a per-task levy applied when firms replace human labor with AI systems, designed to internalize the loss of consumer demand created by automation.
The paper argues that no major economy has yet adopted such a mechanism at scale, and current labor market trends in technology sectors show significant layoffs alongside rapid automation adoption.
The authors conclude that without structural policy intervention, economies could face increasing pressure toward high productivity with weakening consumer demand.
