Revenue Management with Network Effects

University Of Minnesota-Twin Cities, Minneapolis MN

Investigators

Abstract

A product is said to exhibit network effects if individual consumers value it more when greater numbers of other consumers purchase or use it. The expansion of social media and internet technology has led to the emergence of new products that display network effects (for example, multi-player online games, group buying opportunities). It has also served to strengthen network effects for traditional products such as movies, television programs, and books by allowing individuals to easily participate in online communities built around those products. In the meantime, advances in information technology have enabled firms that sell such products to collect large quantities of data in hopes of better understanding how customers choose among multiple potential purchases. For these firms, a significant challenge is to find ways of making product pricing and assortment planning decisions that account for network effects. This research project will address this challenge and, in doing so, will provide methodologies that hold the potential to improve the profitability of firms engaged in internet commerce. Graduate and undergraduate students will be involved in the project and will gain exposure to techniques of mathematical modeling and optimization, thereby providing them with sophisticated technical skills as they enter the workforce. Consequently, the research will benefit the United States economy as well as society. This research project will study mathematical models of customer choice behavior with network effects, and will use those models as the basis for multi-product pricing and assortment planning optimization problems. The project will employ variations of the multinomial logit choice model, suitably modified to incorporate network effects. In these modified models, the utility functions that drive individual customer purchase decisions are affected by overall consumption or sales levels of the products. This feature of the choice models adds difficulty to the optimization problems, because product purchase probabilities are not merely functions of product attributes, but rather are endogenous. This project will involve the development and analysis of approaches needed to solve these problems and will study the structure and properties of the resulting solutions.