Recommender Systems and Competition on Subscription-Based Platforms

Subscription-based platforms offer consumers access to a large selection of content at a fixed subscription fee. Recommender systems (RS) can help consumers by reducing the size of this choice set by predicting consumers' preferences. However, because the prediction is based on limited information on the consumers and sometimes even on the content, the recommendations are often biased, a phenomenon widely evidenced in the computer science literature. Intuitively, if these biases systematically favour certain suppliers against others, this could impact competition between these suppliers. To study this intuition, we introduce a simple theoretical framework of platforms selling to consumers with a quasi-linear utility function via a recommender system. We find that RS biases lead to more concentrated markets, increased entry barriers, and increased homogeneity in the recommendations. We extend these findings to show that when users have limited attention, it can reduce the market concentrating impact of RS biases and harm top-selling product. The platform can counteract this effect by a choice architecture that gives more prominence to popular items. Self-preferencing can further increase concentration and ensure that the winners are the products preferred by the platform. Although encouraging more exploration can reduce these market consolidating effects, we show that they also reduce recommendation relevance in the short-run