Dynamic Pricing, Intertemporal Spillovers, and Efficiency

We present evidence on the potential efficiency gains of dynamic pricing algorithms. Using data from a staggered rollout in the context of restaurant food delivery, we find that high-frequency algorithmic pricing reduced average prices and increased output. Consumers appear to strategically time purchases across days of the week, hours of the day, and at higher frequencies (within the hour). Using a structural model, we estimate that production costs fell and consumer welfare increased with dynamic pricing. Counterfactual analyses indicate that intertemporal spillovers in demand have important implications for equilibrium price levels, time-series price variation, and costs.