AI pricing algorithms can push prices above competitive levels, study finds
When AI pricing systems miss competitors, they may inflate prices to monopoly levels.
A study examines whether simple algorithmic pricing systems can systematically produce collusive-like prices in multi-firm markets. Firms using an explore-then-exploit pipeline randomize prices during exploration, then estimate demand using a misspecified monopoly-style model that ignores competitors. The pipeline can converge to supra-competitive prices above the Nash equilibrium, particularly when firms explore similar price ranges on the same side of the Nash price. Under symmetric exploration, prices can reach monopoly levels. Simulations calibrated to a real multifamily rental market show these outcomes arise robustly beyond theoretical assumptions.
- Misspecified explore-then-exploit algorithms can converge to prices above Nash equilibrium.
- Under symmetric exploration, prices can reach monopoly levels.
- Simulations on real rental market data confirm robustness across finite horizons, heterogeneous products, and nonlinear demand.
Why It Matters
Implications for antitrust regulators: AI pricing systems may inadvertently collude, driving prices up in real markets.