Algorithmic Recommendations and Human Decisions: Evidence from Hotel Pricing

Résumé

We study the interaction between algorithmic advice and human decisions in the context of dynamic pricing in hotels. We provide evidence of adjustment costs and inertia that induce a conflict of interest between the human manager and the algorithmic advisor, resulting in biased recommendations and suboptimal decisions. We develop a simple structural model of advice that is consistent with the data and allows us to quantify the benefits from a potential shift into a fully automated algorithmic pricing system. jointly with Juha Tolvanen and Alexander K. Wagner.