One smart agent

I propose an equilibrium concept for a class of games in which players make irreversible costly decisions; these games have been widely used in the recent Industrial Organization literature. The equilibrium concept is defined not in the space of strategies, but in the space of (observable) outcomes. It is weaker than perfect Nash equilibrium, and it involves combining a form of "survivor principle" with an assumption about entry. This assumption involves only a very weak rationality requirement: if a profitable opportunity exists in the market, there is "one smart agent" who will fill it. This weak equilibrium concept is sufficient to imply some empirically interesting regularities in the area of market structure.