The Informative Harm

When search costs decrease — when it becomes easier for consumers to find and compare products — standard economic theory predicts that competition increases and prices fall. Consumers benefit because they can find better matches; firms compete harder because they know consumers can easily switch. Cheaper search helps everyone.

The authors of arXiv:2603.25893 (March 2026) distinguish two ways search can improve. Search can become cheaper (lower cost per search) or more informative (each search reveals more about product quality). The standard prediction holds for cheaper search. But for more informative search — the kind that AI-powered shopping agents provide — the effect reverses. Better signal quality can weaken competition among businesses and reduce consumer surplus.

The mechanism: when each search reveals more information, consumers need fewer searches to make a decision. Fewer searches mean fewer competitors are evaluated per consumer. With fewer competitors per customer, each firm faces less competitive pressure and can sustain higher prices. The improved quality of each individual search reduces the quantity of searches, and the quantity reduction dominates the quality improvement.

The harm is conditional. If the market itself can observe the search transcripts — if firms learn what consumers learned during their searches — then the information effect can be positive. The firms adjust their offerings in response to revealed preferences, and competition operates through product improvement rather than price. But when only the consumer benefits from better information (the more common case with private AI agents), the information asymmetry tilts toward lower competition.

The structural observation: making individual decisions better can make the aggregate market worse. Each consumer makes a better-informed choice, but the system of choices produces less competition. The improvement is real at the individual level and harmful at the market level because the individual optimization (fewer but better searches) changes the competitive structure (fewer competitors evaluated). Local optimization and global welfare move in opposite directions when the optimization reduces the interaction count that drives competition.