scientific article; zbMATH DE number 1253572

Publication date 23 February 1999

compliance constraints renegociation proof equilibria self-enforcing contracts strategic bargagining

Research exposition (monographs, survey articles) pertaining to game theory, economics, and finance (91-02) Other game-theoretic models (91A40) Applications of game theory (91A80)

Review text

This book is a research monograph on the approach the author has developed to deal with a class of economic games. It concentrates on two person non-cooperative games of the prisoner's dilemma type in which it is assumed that there is a unique non-cooperative Nash equilibrium which is inefficient, i.e., the payoffs are less for both players than the payoffs obtainable in at least one strategy pair. The author proposes and analyzes a novel determinate solution concept, within the repeated play setting of supergames, for such economic games where uncoordinated behaviour does not lead to efficient outcomes and coordination cannot be enforced by outside parties. The book nicely motivates this particular setting, at the outset, by discussing several examples from international trade, industrial organization, public goods, environmental problems etc. where such a model appears as the natural framework. The author brings together two separate strands of the existing literature in game theory to formulate his particular solution concept. One of the strands is the theory of repeated games in which it is well known from the Folk theorem that efficient solutions may be sustained as non-cooperative equilibria provided discount factors are high enough. The second strand is the theory of strategic bargaining in the Rubinstein alternating offers model in which a bargaining solution is sustained as a subgame perfect Nash equilibrium, the solution again depending on the discount factors of the agents. The author synthesizes these two strands in a single solution concept for the particular class of games he considers. The focus of the book then is on economic games in which a certain strategic behaviour may overcome potential inefficiencies through coordination among players which is achieved by means of a self-enforcing contract between the players. A self-enforcing contract is one which does not depend on external threats or third party interference but rather on threats which are laid down in the contract itself in the form of specified future actions such that both agents will refrain from deviating from the agreed upon cooperative actions, which it is the purpose of the contract to sustain. The word contract here is meant to be an explicit or implicit understanding and not necessarily a formal legal document enforceable by a third party. The main elements of the author's approach are expounded in chapter 2 of the book. Following a brief overview of the literature on cooperation in a non-cooperative environment, he goes on to explicate what he means by a self-enforcing contract. The basic idea of the theory of repeated games is that repetition allows for punishment for deviation from an agreed upon cooperative outcome through trigger strategies such as choosing cooperation so long as the opponent does likewise but reverting to the non-cooperative equilibrium strategy for the rest of the supergame if the opponent deviates from the cooperative strategy at any stage, While such trigger strategies do constitute a subgame perfect Nash equilibrium, it is noted that it does not allow cooperation to be resumed if a deviation does occur at any stage from the cooperative outcome. The credibility of such a threat is in doubt given the repeated play setting, the inefficiency of the non-cooperative solution and the possibility of renegociation at any future period to achieve bargaining gains for both parties through cooperation. Accordingly, renegotiations, following a deviation from the cooperative strategies, is allowed for in the strategies to be considered. On the other hand, clearly some form of punishment is necessary to discourage deviation from the cooperative solution. The author chooses to adopt one of the various refinements of subgame perfect equilibrium suggested in the literature, namely the notion of renegociation-proof equilibrium in the sense of Farrell and Maskin. Agents may reconsider their strategies and consider revising these at any stage. Renegociations-proofness requires that the strategies chosen should be immune against such possible renegociations, i.e., even though the possibility of renegociations exists, deviations from the strategies chosen for the relevant subgame in the future periods should not be worthwhile in the sense that there is none which improves the payoffs of both agents (and of course neither agent can improve upon it unilaterally). Specifically, the strategies should satisfy the property that the discounted present value of the gains from cooperation, compared to the non-cooperative solution, exceed the gains from defection followed by the non-cooperative solution for all future periods. In addition, two other conditions must be met. One, referred to as ex ante compliance, requires that it must not pay for an agent to defect in one period and then to invest in a resumption of cooperation after \(n\) periods, compared to maintaining the cooperative strategy for all \(n+1\) periods; here \(n\) stands for the agreed upon number of periods during which a deviating agent is obliged to maintain its cooperative strategy during this punishment phase while the opponent may enjoy the benefits (compensation) from its best response strategy. A second condition, ex post compliance, ensures that a deviating firm finds it to its self interest to resume cooperation after the punishment phase because the discounted stream of such payoffs is larger than what it would be if it did not do so. As may be imagined, some of these conditions are implicit in others. Barring special circumstances, in general, there would be a range of possible renegociation proof equilibria. The choice of an unique solution, out of this plethora of alternative self-enforcing contracts, constitutes the second tier of the author's solution concept. The compliance constraints, mentioned earlier, determine a frontier of feasible payoff combinations. The author's proposed solution is the strategic bargaining solution of a Rubinstein alternating offers game where the offers involve points on the efficient frontier of the feasible payoff combinations corresponding to the compliance constraints. It is cast as an infinite horizon game where the time which elapses between alternating offers is understood to be ``negligibly small; these two features remove any asymmetries due to last mover or first mover advantages. It is a basic assumption of the entire work that the model under consideration is stationary, i.e., the same environment is repeated at each stage regardless of actions taken in the past (there are no dynamic stock effects). As the author points out, this is a restrictive assumption whose sole purpose is to simplify the analysis. Time consistency in such a non cooperative setting would naturally require that the result is time independent, since no agent can credibly commit himself to a different strategy at different points of time when the circumstances are identical. So the agreed upon strategies are identical over time and bargaining for payoff levels translates into bargaining for a single strategy level (for all future realizations) in a strategic bargaining game for one realization if contracts were enforceable. Subgame perfect bargaining solutions are studied in this context and given an interpretation in terms of concessions, i.e., a change of the strategy combination which an agent is prepared to accept in a period if he can realize an agreement in the preceding period when it is the turn of the opponent to make an offer. Subgame perfectness requires that neither concession increases either agent's payoffs and that no agent can gain by waiting. Under sufficient regularity conditions on the feasible payoff frontier -- it is concave and piecewise differentiable -- it is shown that there is an unique strategic bargaining solution, This is the proposed solution of the game. The solution depends on the discount factors of the agents and the relationship between the two is carefully studied. Some interesting conclusions are brought out. The solution proposed need not achieve an (unrestricted) efficient payoff combination, i.e., if agreements could be enforced by a third party then an improvement in both agent's payoffs may be possible, compared to the solution of the game. This is because the solution considered is a self-enforcing contract and is restricted to those which satisfy the various compliance conditions. It is possible, for instance, that, under appropriate parameter values, the entire feasible region of payoffs determined by the compliance constraints lies strictly within the unconstrained feasible region. Furthermore, even if the compliance constraints admitted (unconstrained) efficient solutions the solution may turn out to be inefficient. For instance, in the extreme case when one of the agents has all the bargaining power and can choose the solution to its best advantage, the objective of the agent is not to maximize joint payoffs in any sense but rather to do the best for itself given the characteristics of the opponent and the consequent nature of self-enforcing contracts. It may be necessary, when choosing the most advantageous payoff, that, in order to ensure compliance by the opponent, some concessions may have to be made in terms of payoffs and efficiency. ``Therefore, focusing explicitly on agreements on the contract curve is likely to overlook an important set of outcomes which are chosen deliberately by both agents as a result of rational bargaining. Intuition from the standard relation between changes in the discount factors and payoffs for the agents in the solution may not carry over in this setting. In strategic bargaining models, the division of the spoils of bargaining depends on the discount factors of the agents, the more impatient an agent is the less its payoffs compared to the opponent. This, however, may not be true in the setting considered here. This is because, as is evident from the compliance conditions, the compliance curves or the concession lines of each agent depends on the discount factor. If the solution is on the compliance curve for one agent and the discount factor of that agent increases then the effects of future punishment following a defection is smaller and this may necessitate more concessions on the part of the opponent to ensure compliance. The increase in the discount factor, however, always decreases the bargaining power of that agent in strategic bargaining. The effect of an increase in the discount factor, therefore, may be ambiguous, Specific examples are provided. This leads to interesting questions about what kind of opponents an agent would prefer to select and team selection questions. Questions of this kind, amongst others, are examined in the rest of the book, within the context of detailed studies of several examples, serving as special instances of the general framework expounded in chapter 2. Chapters 3 and 4 deal with joint provision of a ``collective good, distinguishing between cases, such as in oligopolistic markets, where the institutional setting would naturally try to discourage cooperation and cases, such as the provision of a public good or externality related problems, where the institutional policies would actively encourage cooperation. Chapter 3 deals with the problem without transfers while in chapter 4 transfers are allowed. In the main the analysis is carried out assuming special features of the reaction functions but providing some indication of how results may carry over in more general settings. Chapter 5 discusses a buyer and seller problem. Applications in duopolistic markets, international environmental policy, trade policy etc. are considered. Some simulation experiments are carried out with mixed results. A final chapter discusses some open problems, while an appendix outlines some further refinements of the solution concept. In the opinion of this reviewer, there is an air of unreality about the very notion of self-enforcing contracts. The possibility of cooperation following a defection needs to be considered because the threat of perpetually following a low level equilibrium strategy is considered incredible since both parties may gain if a deal can be struck. However, as the author notes, some form of punishment is needed as a deterrent to defection. The solution accordingly provides for a punishment phase but satisfies conditions so that, even though a sequence of actions contingent on defections are laid out, this is less profitable to actually follow, compared to sticking with the cooperative strategy. But what is there to prevent one agent to try to extract the ``temporary gains from a defection, banking on the possibility that it may be able to persuade the opponent to renegociate even prior to, or in the middle of, the so-called ``punishment phase. After all, it may threaten to follow the non-cooperative strategy which would mean losses sustained by the opponent as well. If this is considered ``incredible because the defecting agent would itself be suffering similar losses, one has to bear in mind that so would the opponent and there is (should be?) credibility and power in the ability to inflict damage. This is, however, a problem not specific to the author's approach but applies generally to the literature on repeated games as a way of introducing deterrence to defection. As the author points out, if the basic model encourages some readers to employ this approach for studying further problems of this kind and carry things forward, the book will have achieved its goal. Researchers in this area should find plenty of material of interest and certainly a perusal of at least the first three chapters and the concluding one should be rewarding and may well tempt one to go through the entire work.

Recommendations

Cited in

(15)

This page was built for publication:

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q4226818)