Abstract Collective Action under Uncertainty

Collective Action under Uncertainty Michael Thomas McBride 2002 I examine collective action under uncertainty in public good games and social networks. The first chapter analyzes discrete public good games in which the threshold number of contributions needed for provision is randomly selected from a known probability distribution function. Simultaneous voluntary-contribution equilibria are often inefficient under binary contribution decisions. Wider uncertainty about the threshold can reduce inefficiencies, however, because contributions are higher when there is greater uncertainty about the threshold (in terms of second-order stochastic dominance) if the value of the public good is sufficiently high. Games with continuous contribution decisions and sequential contribution decisions are also considered. The second chapter presents results from public good experiments specifically designed to test the main qualitative predictions of the first chapter. As predicted for within-session changes, contributions are higher under wider uncertainty when the value of the public good is high, and contributions are lower when the value of the public good is low. I use a proper scoring rule to elicit data on beliefs, and these data exhibit qualitative features of standard learning models. Using these data to proxy for actual beliefs, I show both parametrically and non-parametrically that aggregate decisions are not consistent with expected payoff maximization. Decisions are more consistent with a game-theoretic decision rule that accounts for risk aversion and innate cooperativeness. The third chapter examines equilibrium network structures under uncertainty. Bala and Goyal (Econometrica 2000) model network formation as a non-cooperative game under complete information. I extend their approach to include uncertainty about the network structure and uncertainty about the benefits of network participation. I define an incomplete-information imperfect-monitoring equilibrium concept, called Generalized Conjectural Equilibrium. I apply this concept to find strict equilibrium network structures in twenty different informational environments. I find that the unique equilibrium architecture of complete information games—the center-sponsored star—is still unique even under large decreases in information. I show exactly how the type of equilibrium structure depends on the informational environment. A main contribution of this paper is the characterization of network uncertainty. Another contribution is the definition and use of a new non-Nash equilibrium concept. Collective Action under Uncertainty A Dissertation Presented to the Faculty of the Graduate School of Yale University in Candidacy for the Degree of Doctor of Philosophy by Michael Thomas McBride Dissertation Director: Benjamin Polak