Helping Behavior in Networked Organizations

I study organizations in which agents are connected through a fixed, un-directed, and unweighted network, and work collectively to produce a team output. Besides choosing own effort that contributes directly to the team output, agents can also exert helping effort to their network neighbors so as to reduce the neighbors' marginal dis-utility of own effort, and thus contributes indirectly to the team output. I characterize the sub-game perfect Nash equilibria in a two-stage game in which agents first simultaneously decide how much helping effort to devote to neighboring agents, and then decide how much own effort to expend, based upon the previously determined pattern of mutual help. The best response function of each agent with respect to the helping effort is nonlinear but the model is still tractable. I establish the existence and (generic) uniqueness of the equilibria. I show that a dense network might not necessarily sustain higher level of mutual help or achieve better performance, because many links in that network might be redundant. If the agents are homogeneous, the network in which agents are pairwise connected could attain the highest efficiency.