War and Peace in an Anarchic World
This paper considers a model in which a number of independent states of changing strength face opportunities to attack each other. Strong states can extract rents by attacking weak states, but in doing so they also risk overextending themselves, and hence making themselves targets of future attacks. The model yields several insights. First, there is no equilibrium in which there is peace for all configurations of player strengths. The reason is that, in such a scenario, states never have to fear future attacks--hence they can attack whenever it is myopically profitable. Second, there can be equilibria in which some configurations of player strengths are peaceful, even though there are players with myopic incentives to attack others. Third, the probability of war may be non-monotonic in its myopic profitability (from the attacker's point of view). Fourth, geographic constraints (preventing some players from attacking each other) or non-aggression pacts may increase or decrease the probability of war. When player strengths are stable, allowing players to enter non-aggression pacts freely at the beginning of the game always increases the probability of war.