This paper studies welfare maximizing information structures in a public good setting.
In large groups less information is provided. In the limit, no information is provided but
the information is efficient as by the law of large numbers no information is needed to take
the efficient decision. This implies that the free rider problem is most severe not for large
but for intermediate group sizes.

We incorporate private information and strategic communication (Crawford and Sobel (1982)) into Simon's (1951) model of the employment relationship. We consider contracts between a principal and an agent that specify a bounded finite number of instructions and a fixed wage. Once the principal, privately, learns the state, she can either enforce one of the instructions from the contract or send a non-binding cheap-talk message to the agent. In the former case, the principal determines the action, in the latter case the agent does. All contracts partition the state space into `topics,' with each topic giving rise to a game in its own right. With little conflict, optimal contracts specify approximately the maximal available number of instructions, and there will be at least one topic in which there is (cheap-talk) communication. For `extreme conflict' optimal contracts are simple, specify a single instruction, and do not generate communication. In the uniform-quadratic specification, with sufficiently small conflict, topics from optimal contracts induce similar numbers of cheap-talk actions.

In this paper we investigate the role of private information in a patent race. Since firms often do their research in secrecy, the standard assumption in the patent race literature that firms know each other's position in the race is questionable. We analyze how the dynamics of the game changes when a firm's progress is its private information. Further, we address the question whether revealing it might be to a firm's advantage. We find that a firm has an incentive to reveal its breakthrough only if its rival has not done so, and only if research is inefficient.

We consider transfer-free binary-action rules for assigning a task among a group of agents. In contrast to a large literature on public-good provision and volunteering, we assume that agents have non-trivial preferences about who performs the task. We show that a utilitarian planner will use a threshold rule: every agent decides whether or not to ``volunteer''; if the number of volunteers exceeds a threshold number, the task is assigned to a volunteer; if the number is below the threshold, the task is assigned to a non-volunteer. We also show that any rule with a non-extreme threshold always has an equilibrium that yields a strict interim Pareto improvement over a random task assignment. This robust-improvement property depends on assigning the task to a non-volunteer rather than randomly among all agents if the threshold is not reached. The first best is approximated with a threshold that tends to infinity as the group size tends to infinity.