Whom to benefit? Competing platforms' strategic investment in recommender systems.

As digital technologies improve and platforms have access to more user data, platforms can use recommender systems (RSs) to discover users' interests and make better matches. The literature on platforms (two-sided markets) focuses mainly on the number of users and less on the effects of matching. We build a new theoretical framework to reconcile this gap. Users can enjoy a greater network effect (NE) from better matching if platforms invest more in RSs improvement. The results indicate the following. (1) The improvement of RSs can increase users' surplus but does not necessarily benefit advertisers. Advertisers' gains from RSs improvement depend on both the strength of the marginal utility of the network effect in advertisers (MUNE-A) and the extent to which platforms subsidize users. (2) Platforms' investment level in RSs not only relies on the marginal cost but also relates to the strength of the MUNE-A. If the latter becomes too strong, then it is optimal for platforms to reduce their level of investment in RSs due to heavy user subsidies. (3) Social welfare is more likely to increase with RSs improvement if users or advertisers can obtain stronger MUNE from each user.