On the theoretical economic study of the industrial organization of digital platforms

Resumen

Internet-of-Things (IoT) technologies are the new digital revolution. A revolution that is based on digital multi-sided platforms. An interesting and representative IoT market is the wearable market, or more specifically, the fitness-tracker market. In that sense, the study of this market and its properties can be helpful in the launching of other IoT technologies based on multi-sided platforms. We address the state-of-the-art of the literature on multi-sided platforms, and we highlight the most relevant open questions in the literature with regard to the fitness tracker market. In that sense, we highlight that there are two main areas: compatibility or data sharing among databases, and the launching phase of the platforms. The contribution of this Thesis is twofold. On the one hand, we develop a theory on the role of compatibility in multi-sided business models. On the other hand, we develop a price competition algorithm for multi-agent systems that allows us to implement theoretical frameworks in agent-based models. In that way, we can simulate the optimal pricing of platforms during the launching phase of their products. This Thesis highlights several interesting insights for managers and public authorities with regard to the pricing of digital platforms. On the one hand, we show that users' expectations matter. The optimal pricing depends on how users perceive the value of the platform. It is easier to launch a platform if there is a device that users value than when all the value relies on attracting other groups of users. On the other hand, we show that compatibility is profitable for the companies, and they are interested in compatibility agreements because they reduce competition. Platforms increase their market power and their profits. Nonetheless, this feature lead to the creation of barriers to entry. Therefore, public authorities have to mitigate those agreements if they want to promote competition and foster new ecosystems. Platforms tend to avoid the entrance of new competitors by launching new brands, and we show that platforms have incentives to do so. We also point out that public authorities have to be cautious with platforms that sell data because there are incentives to create monopolies by splitting the market. One company that provides the data, and another one that provides the devices. However, public authorities may have another bigger problem. We prove that platforms tend to subsidize one side to attract another group of users but also, they can even make temporal cross-subsidies to attract future consumers. In that sense, prices of digital platforms can be even lower than what one-shot theoretical models predict. The last contribution of the Thesis is with regard to the simulation of digital multi-sided platforms. We prove that addressing the launching of multi-sided platforms with the available theoretical models can be intractable, and we propose an algorithm for the simulation of those models that allows us to keep the tractability of theoretical models when we relax their simplifying assumptions. We prove that the algorithm resembles the best response map, but without assuming any particular theoretical model or function. We apply it in three launching cases, and we obtain three relevant conclusions. First, how users get aware of the existence of products is as important as other theoretical features, such as the differentiation. Optimal pricing must track how is the diffusion of information among consumers. The faster, the better. Second, not all users are equal. Some users are more important than others in transmitting information. The presence of those users increases the profits of platforms, without them, failures in launching are more likely. Third, when launching platforms with the idea of disrupting a previous market, platforms' owners have to provide significant value to users. It is not enough with providing a similar solution than the existing one. It has to be significantly better if not, the platform will fail. Our simulation maps some theoretical thresholds to determine how much value is necessary to avoid the failure in launching.