Social network and economic incentives

Many economic activities occur within the context of social networks e.g. job search through referrals, word-of-mouth recommendations of goods and services, enforcement of cooperative behaviors in small communities. In a given network, the role of social ties may range from being channels of information to constraints on search to reservoirs of personal favors. The potential economic values of social ties motivate both economists and sociologists to study whether and how the structure of social networks and economic behavior are related.;In Chapter 1, I model 3-point referrals and propose how the referrer's (relative) cooperative incentives and informaton vary with respect to the (relative) overlap of his network with those of the originator and the recipient (the possibility of the link between information and network overlap is included in the analysis essentially as a control variable). To test the main hypothesis, I analyze 3-point referral data from a leading online business and adopt a structural approach in the estimation to reflect the non-linearity in the model comparative statics. The results show that a referrer is biased towards the party whose network is more overlapped with his. On the other hand, despite possible negative endogeneity biases, network overlap does not appear to be correlated with the amount of information the referrer has on the outcome of the match. In addition to supporting the main hypothesis, the results also imply that when the cost of referral is sufficiently small, only the relative network overlap between all the parties, rather than the absolute levels, determines the success of the referral once it is initiated.;In Chapter 2, I conduct an alternative test of the main hypothesis by modeling how network overlap affects endorser incentives and subsequently, the effectiveness of public endorsements. Analyzing online endorsement and target audience response data, I find that endorsements by loosely overlapped endorsers are more effective in generating interest among the target audience. This result provides additional support for the main hypothesis. It also imply that an endorser who is more able to help spread positive reputation for an individual is, interestingly, the one who cares relatively little about that individual's reputation because his endorsement will be considered more credible. Moreover, a unique feature of the data allows us to separately identify the size and incentive effects of network structure on endorsement effectiveness.;Chapter 3 is an application of the results in Chapters 1 and 2. Having established which network structures are optimal for referrals, I analyze a panel of the online personal networks of about 3500 individuals and find evidence that (exogenous) changes in individuals' eagerness to expand their networks are accompanied by changes in personal network structures that are optimal for expansion through referrals. In contrast to previous (mostly theoretical) works on network efficiency, neither explicit models of the network formation process nor network stability are considered here. The focus here is on whether the dynamics of the network formation process somehow, by the choices of network participants or exogenous forces, deliver personal networks with the optimal transitory network structures. The welfare implications are similar to that of the literature on the return on schooling or training i.e. whether an individual who has the most potential to improve his current situation (personal network vs income) is able to do so (through referrals vs education). Thus, this chapter explores a somewhat novel dimension of network efficiency.;In this dissertation, I propose and empirically test the hypothesis that two individuals with highly overlapped social circles behave cooperatively with each other. Despite its conceptual simplicity, the link between network overlap and cooperative incentives has significant implications on trust and credibility when we consider multi-person or across-group interactions in social networks - two highly cooperative individuals will be perceived by third-parties as highly collusive and as a result, under asymmetric information, any potentially self-benefiting information they reveal may be received by others with skepticism. Based on this intuition, I find evidence of behavioral patterns consistent with the main hypothesis in referral and endorsement data from a leading online social network. At the macro-level, this result also has implications on what kind of network structure facilitates the flow of opportunities and information across groups/clusters.