UCLA

Social and economic networks play an increasingly significant role in people's lives. The formation of such networks do not only provide exchange of material or monetary benefit, but also transmission of information. By creating direct links and indirect connections with others, on one hand the delivery of goods and financial transfers are made such as in trade networks or in buyer-seller networks, and on the other hand agents learn valuable, payoff related information from own observation or peer reviews. For this dissertation I studied network formation and related problems from various aspects. These studies have pointed to overlooked but crucial factors in the information structure among strategic agents and the process of information acquisition, which once taken into account have produced theoretical predictions in stark contrast to the existing literature, and have reconciled long-existing inconsistency between theory and data.

Chapter 1 studies the problem of social learning through observation. Social learning is the study of how dispersed information gets aggregated in a society of strategic agents, and what kind of structure of information acquisition the society needs to facilitate efficient information aggregation. The existing literature always assumes that the observation structure is exogenous, in other words, the structure of who observes whose actions is exogenously given by some deterministic or stochastic process. However, it is more natural or realistic to assume that observation is costly and strategic. My study takes this alternative assumption and shows that in contrast to the condition of expanding observation -- meaning to observe a close predecessor -- in the literature, a sufficient and necessary condition for the highest level of social learning with costly endogenous observation is infinite observation, i.e. to observe an arbitrarily large number of predecessor. Endogenous observation also brings about a great difference in terms of individual behavior and social welfare.

Chapter 2 addresses the question of how networks form and what their ultimate topology is, under the much more natural yet hardly adopted assumption of incomplete information: agents do not know in advance -- but must learn --the value of linking. This study shows that incomplete information has profound implications for the formation process and the ultimate topology. Under complete information, the network topologies that form and are stable typically consist of agents of relatively high value only. Under incomplete information, a much wider collection of network topologies can emerge and be stable. Moreover, even with the same topology, the locations of agents can be very different: an agent can achieve a central position purely as the result of chance rather than as the result of merit. All of this can occur even in settings where agents eventually learn everything so that information, although initially incomplete, eventually becomes complete. The ultimate network topology depends significantly on the formation history, which is natural and true in practice, and incomplete information makes this phenomenon more prevalent.

Chapter 3 again provides an analysis of the network formation process, but in another understudied strategic environment -- one where agents are foresighted and care about both current and future payoffs. The related theoretical literature has often adopted a framework with homogeneous agents, and predicted that in generic cases it is impossible to sustain efficient networks even if such a network generates a positive payoff for every agent. However, these predictions are contradicted by data from existing real-life networks. This study analyzes a dynamic model allowing for agent heterogeneity and foresight, and establishes a Folk Theorem of networks, characterizing the set of sustainable networks in equilibrium for patient agents, and find that efficient networks can be sustained in equilibrium as long as it guarantees every agent a positive payoff. In the widely studied connections model which adopts a particular class of this valuation structure, a full characterization of efficient network is presented, which turn out to bear a striking resemblance in topology to networks observed in real life.