Ongoing Research

Artificial Intelligence and Macroeconomic Dynamics: Growth, Pricing, and Distribution

Authors: Helen Popper (Santa Clara University), Christoph Schult (Halle Institute for Economic Research) Status: Draft (January 2026)

Abstract: This paper develops a tractable general equilibrium model in which artificial intelligence (AI) is produced by a monopolist that both learns by doing and uses AI recursively as an input. Explicitly modeling AI production clarifies how AI prices, adoption dynamics, long-run growth, and relative wages evolve in general equilibrium. Learning and self-use make the monopolist’s pricing of AI forward-looking and endogenous, generating two distinct adoption regimes—learning-first and scale-first—that shape macroeconomic and distributional outcomes.

For analytical clarity, the paper characterizes a generalized balanced growth path under a Cobb–Douglas benchmark. This benchmark delivers a closed-form condition for stable, non-explosive growth and implies that AI output grows faster than final output, the relative price of AI falls persistently, and real wages rise with overall output.

Extending the model to CES technologies, the paper derives a closed-form effective demand elasticity for AI that combines input substitution in production with product-market substitution across differentiated varieties. These forces jointly determine AI markups and price dynamics. Simulations show that complementarity in final-goods inputs generates a learning-first, capital-light adoption regime, while substitutability leads to scale-first adoption and attenuates the two-phase pattern. On the distribution side, the wage premium for specialized labor is lowest when final-goods inputs are complements and rises with substitutability; greater substitutability in AI production amplifies these effects.

Note: Ongoing research listed here is work in progress and not yet peer-reviewed. Items move to the Publications section once accepted or published.