Communication and transportation networks provide the infrastructure for life in modern society. Whether messaging over the Internet, making voice or video calls, or choosing between routes on highways, the quality of one's experience depends critically on the design and control of such networks. A major concern of much of the research on networks has been congestion; in particular the congestion resulting from the negative impact of decentralized decisions on social welfare. This literature has shown that autonomous, self-interested users can experience increased costs or substantial losses in efficiency (e.g., delays in traffic) compared with socially optimal outcomes due to congestion in networks. Of course, not all decision making in decentralized networks gives rise to negative externalities. In fact, network industries are often characterized as those that have increasing returns to scale. Consider, for example, telephones or fax machines. The value of these technologies increases in the number of users who have adopted them; similarly, the value of computer software increases as more users purchase this technology.
In a departure from previous experimental research, we propose studying interactive decision behavior in networks, where choices result in positive externalities; where the utility that a given agent derives from her choice increases in the number of other users making the same choice. We are interested in a family of applications regarding public modes of transportation where agents, sharing the same route from a common origin to common destination, can also share part of the cost. We propose to conduct several laboratory experiments which differ in (i) the cost associated with alternative modes of transportation, (ii) the homogeneity or heterogeneity of the players, (iii) the information available to the players about previous decisions of other players, and (iv) whether decisions are made in a simultaneous or sequential manner. The research proposed is one of the first systematic studies that investigate cost-sharing of public transportation by independent, autonomous agents. We propose a novel approach to cost-sharing allocation in a dynamic setting that integrates both theory and experimentation. The examination of heterogeneity in users, the information that they have of one another, and issues such as segment capacity, all bear important implications for policy making related to network traffic.
