Online reviews increasingly play a crucial role in the purchase decision process of consumers. Indeed, through the observation of the opinions and the experiences of prior purchasers, consumers can ex-ante estimate the quality of products to make comparisons between the remarkable amount of alternatives that are usually available in online markets. This thesis studies to what extent the presence of product choice and competition influences the process of information diffusion in a population of consumers who communicate through the mechanism of online reviews. Specically, we study a model of social learning in an oligopolistic market with heterogeneous consumers who are initially uninformed about the intrinsic quality of products. To decide which product to buy, if any, they estimate the quality of the dierent alternatives using the online review information displayed by an information aggregator (the platform). Purchasers of a given product experience its true quality plus a random disturbance, and publicly report whether they liked or not the product reporting a binary online review. Chapter 3 studies the case of a duopolistic market where consumers use a maximum likelihood estimation procedure to ex-ante assess the quality of the different products. We characterize the outcome of such inference procedure and we provide conditions for the asymptotic almost sure consistency of consumers' quality estimates. Chapter 4 instead considers an oligopolistic market where consumers naively identify the quality of a given product with the fraction of positive reviews for that product. We examine what consumers learn over time under this inference rule, and how this impacts on the consistency of their decision. A mean-eld approximation is then proposed to provide an insight on how the presence of product choice in influences the speed of the learning transients. Chapter 5 provides a heuristic analysis of the information control problem of the platform that controls products' relative positioning in the search results of consumers. We focus on myopic ranking policies with different objectives, such as greedily maximizing prots or optimizing a measure of the amount of information acquired from every transaction. We characterize these myopic policies, establishing whether they are asymptotically optimal or not, and numerically compare their performances.
Scheda prodotto non validato
Attenzione! I dati visualizzati non sono stati sottoposti a validazione da parte dell'ateneo