Impact of the Scale Effect of Recommendation Systems on Competition in Digital Platform Sectors

Over the past quarter-century, digital platforms proliferated and became the world’s most valuable companies. Traditionally, the growth of digital platforms is explained by cross-platform network effects, which, in turn, are supported by recommendation systems—a set of algorithms that suggest the most suitable user of one type to a user of another type. The dependence of the accuracy ensured by algorithm predictions on the number of observation units and on the number and type of observations for each unit—returns to scale—affects the comparative competitiveness of large and small platforms, the structure of markets, and hence the choice of public policy instruments in relation to the platforms. The aim of the article is to systematize the data regarding the returns to scale of recommendation systems on digital platforms. The results obtained by empirical studies and the analysis of coverage and convergence indicators for some Russian platforms cast doubt on the significant positive return of recommendation system accuracy with regard to the number of users: it largely depends on the designed set of algorithms. Improvement in recommendation system algorithms will make it possible for even smaller Russian platforms to remain competitive with a limited number of users.