Common Property Externalities: Isolation, Assurance, and Resource Depletion in a Traditional Grazing Context: Comment

Abstract

Runge has made a useful contribution by challenging identification of problems of grazing on common property with prisoner's dilemma. It may be, as Runge argues, that nonseparable interdependencies of cost functions in context of traditional grazing make conventional model of prisoner's dilemma inapplicable in that context. But Runge is incorrect in stating that, even in prisoner's dilemma, introduction of can solve of voluntary cooperation. Runge confuses two different models of of voluntary cooperation. The first model, prisoner's dilemma, is characterized by fact that each individual has an incentive not to cooperate, regardless of what other individual does. This is a characteristic of pay-off matrix of prisoner's dilemma and is independent of expectations each agent has of other agent's action. In on other hand, it pays each agent to do what other agent is doing. In such a context, expectations of other agent's action clearly do matter. One game, therefore, cannot be transformed into other merely by introducing an argument about as Runge claims, saying that the structure of N-person prisoners' dilemma may be transformed into an assurance by dropping assumption that individuals formulate their choices independent of expected choices of (p. 602). This assumption is not made by analysts of public goods problems (e.g., Buchanan). Runge continues, each individual expects everyone else to stint, that individual will stint, too (p. 602). The point of prisoner's dilemma, especially when viewed as a model of free-rider in a public goods or externalities context, is that incentive to (or not contribute to provision of a public good) continues to exist, and, in fact, is greatest precisely when others are not stinting. The notion of expectations-that individuals will cooperate if only they know that others are cooperating-therefore, is plainly false in context of prisoner's dilemma, notwithstanding claims of Frohlich and Oppenheimer, as cited by Runge. Expectations can have an important role, however, if agents in prisoner's dilemma are allowed not only to choose simple actions (to stint or not to stint) but also to commit themselves to reactions (I will stint if you stint, but if you do not stint, I won't either.) This point arises out of Taylor's work on repeated prisoner's dilemma, Howard's on meta-games, Thompson and Faith's on hierarchical interaction, and of my own on (Guttman 1978, 1981). It is now generally recognized that if one or both actors can commit themselves to a strategy of matching each other's contributions to public good, free-rider can be solved in groups of all sizes. This notion of matching behavior, however, is not at all identical to notion of coordinated expectations. The expectations emphasized in literature on meta-games, supergames, and matching behavior are expected reactions by other actors to one's own actions, not expectations of unconditional actions of others. Runge's references to vast literature on prisoner's dilemma are misleading. One cannot properly associate work of Taylor on repeated prisoner's dilemma with that of Frohlich and Oppenheimer, as Runge does (pp. 602-03). The experimental results of Marwell and Ames, Smith, and Bonacich et al. also do not lend support to specific solution proposed by Runge. For an experimental work demonstrating existence of matching behavior in a public goods context, see Guttman (1981). It is therefore misleading to treat prisoner's dilemma and problem as two species of same general problem. Whether of grazing on common property is better viewed as a prisoner's dilemma or as an assurance problem, a purely voluntaristic solution may well occur. However, for prisoner's dilemma, reason why outside enforcement may be unnecessary is not possibility of expectations, but that of matching behavior.