Recommenders’ originals: The welfare effects of the dual role of platforms as producers and recommender systems

• We study a model of strategic interaction between a producer and a platform that employs a recommendation system • Producers have access to stochastic investment technologies and the platform employs a recommender system to provide consumers with information on product quality, following an information design approach. • We characterize the conditions under which the platforms dual role into the production market results in lower consumer welfare and in foreclosure of the independent producer. • Whether or not a policy separating recommendation and production or imposing unbiased recommendations is welfare improving depends on whether the platform’s revenue potential is large enough. • Our results highlight the importance of targeted restrictions on platform self-preferencing. We study a model of strategic interaction between a producer and a platform that employs a recommendation system, following an information design approach. Upon entry into the production market, the platform biases recommendations to credibly steer consumers towards its own goods. Despite the increased upstream competition, platform entry and self-preferencing can decrease consumer welfare and result in foreclosure of the independent producer. We then consider the natural policy remedy of separating recommendation and production or imposing unbiased recommendations and find it leads to welfare gains if the platform’s revenue potential is large enough, but to significant welfare losses when it is not. The ambiguity of such a policy’s welfare implications and the dependence on the industry’s returns highlights the importance of targeted restrictions on platform self-preferencing.