Collective Action Games Embedded in Social Networks

The paper presents a model that synthetically incorporates several paradigms of collective action theories to bring up conditions when different types of collective action games would occur. Different from past literatures, the paper focuses on the analysis when various types of actors are allowed to play games interactively. The designation of game-playing partners follows the three mechanisms of network formation incorporated in the paper. A computational simulation is adopted to show how some parameters such as network density and heterogeneity influence the final contribution of collective goods. * Email: [email protected] Draft Yen-Sheng Chiang 2 Introduction In The Logic Of Collective Action, Olson (1965) categorized different types of groups for collective goods production distinguished by different incentive structures, mainly caused by group size. A privileged group is a group where group members are strongly induced to contribute even when only a small number of others are predicted to do so. A latent group is a group where individual’s contribution does not cause any noticeable effect on the total welfare of group. An intermediate group, by definition, is a group where the production of collective goods cannot be solely burdened on any particular individual, but any person’s partial contribution can have a remarkable effect on total welfare or can be easily monitored by other group members. Olson’s insight was then successfully reinterpreted by scholars using game theoretical approach. Examples, among others, include Hardin(1982), Taylor (1987) and Chong(1991). Olson’s main concern, group size, plus other kinds of incentives, are mapped into the payoff structure of a dyadic game, and then the Nash equilibrium can be determined. They present corresponding types of games to the types of groups proposed by Olson(1965): privileged games, assurance games (also often called as coordination games) and prisoner dilemma games. There are rich discussions about how different factors, such as instrumental and social incentives, are embodied as the payoff structures in a collective action game. But most of these focus on the analysis when collective action can be interpreted as one type of game instead of exploring the possibility that various of types of games coexist with respect to the same collective action. If free riders and zealots coexist, for instance, the following question is what kind of condition will situate individuals to be free riders or Draft Yen-Sheng Chiang 3 zealots, and what kind of outcome would be caused when they interact (Coleman, 1988)? In this paper, an attempt is made to connect the past game theoretical discussion of collective action to the critical mass paradigm in sociology (Oliver and Marwell, 1993), which provides insights in how heterogeneous individuals are motivated to participate in collective action. For the past decade, critical mass theory proposed by Oliver etc al. have aroused great discussions about how costly collective actions are made possible, given a certain extent of incentive to free ride (Oliver and Marwell, 2001).Their discussion of group heterogeneity in terms of resources and interest provides a useful tip to construct the payoff structures in a dyadic collective action game. In addition, the distribution of interest and resources of group members can be adopted as clues in detecting how individuals make social connections to each other, and thus determine “whom” they are assigned to play the game with. 1 Generally it is argued that degrees of interest and resources are positively associated with the willingness to participate in collective action (Oliver, Marwell and Teixeria 1985: 529). With the distribution of resources and interest among promising participants, it is easy to categorize them into different groups, similar to Olson’s taxonomy. Recall that from Olson’s argument it is implied that there is a certain threshold of group size, above which individual’s contribution makes trivial effects on the total welfare, and below which a remarkable effect from an individual’s contribution would be made. Analogously it is plausible to argue there exists a threshold value in either interest or resources as prerequisites to 1 One example of such criterion is the principle of homophily (McPherson, Miller; Smith-Lovin, Lynn and Cook, James, 2001), which argues individuals are more likely to interact with those sharing similar attributes with them. A discussion in more detail is presented in the following section. Draft Yen-Sheng Chiang 4 participate in collective action. It will not arouse much debate that the production of collective goods incurs a certain level of inputs of resources. It is often observed as well that some collective action requires some degree of interest to sustain participants’ commitment in risky social movements. Precisely where the cutoff line would be remains an empirical question, but if we accept the argument that thresholds do exist in both dimensions of interest and resources, then four different “groups”, categorized by their relative position to the threshold, are obtained and illustrated as follows. Resources R < R* R > R* I > I* Interest