Title: Monetizing Steering (with Heski Bar-Isaac)
Abstract: Better-placed products enjoy greater sales, and so retailers on marketplace platforms prefer that the marketplace steer customers in their direction. A monopoly marketplace can earn profits through an ad-valorem fee, but how should the marketplace set it? A monopoly marketplace also chooses whether to steer through the design of an algorithm or through an ad auction that raises further revenues. We examine which of these the marketplace prefers. Operating in a single competitive retail market or when perfectly discriminating among markets, the marketplace can extract monopoly profits fully through an algorithm or through an ad auction with appropriate fees. The optimal fee for the marketplace depends on the steering method. With an auction, the optimal fee is 0; positive fees induce double marginalization. Thus, algorithms are preferred where fees are positive—if, for example, there are limits on the ability to steer or to counteract retail market power. However, (natural) limits to fee discrimination can favor steering by auctioned ads. This rationalizes using ad auctions in some markets and algorithms and commissions in others. The approach also speaks to self-steering and highlights that assessing policy impacts requires a holistic view.