Steering via Algorithmic Recommendations

Abstract

This paper studies whether market structure affects algorithmic recommendations in dominant platforms. We focus on the dual role of Amazon.com as a platform owner and retailer. We find that products sold by Amazon receive substantially more "Frequently Bought Together" recommendations across product categories and popularity deciles. To establish causality, we exploit within-product variation generated by Amazon stockouts. We find that when Amazon is out of stock, the identical product sold by third-party sellers receives 8% fewer recommendations. The pattern can be explained by economic incentives of steering and cannot be explained by consumer preference. Furthermore, the steering lowers recommendation efficiency.