There are few circumstances where wars are globally welfare enhancing. One can imagine that wars can be individually optimizing, for example when we consider the old land-taking or enslaving war. But casual empiricism indicates that quite often fools engage in wars, like minnows tickling obviously overpowering giants (Irak, North Korea) or others who have little objective chance of winning (South Ossetia, Caprivi, Falklands). Is it because some belligerent are poorly informed or even irrational?
Clara Ponsati and Santiago Sanchez-Pages use Markov games with fully rational players to characterize wars, and even chronic wars. A country can lay a claim on another country, leading to bargaining or war, and it can only end if one surrenders. The problem is that parties do not know their relative strengths and can only learn about them by engaging in war. Add a dose of optimism, and you have a recipe for war. Were one to add some political economy (or populism) to this model, outcomes would be really depressing and worrisome. But I still have some faith in humanity.