Introduction

Abstract

The Review has a tradition of publishing papers on the subject of planning and of iterative planning procedures. On the whole, these papers have not addressed themselves at all to the important question of how the information from individual agents needed to implement a planning rule is to be elicited. The question is important because individual agents may well have an incentive to provide deliberately misleading information if this distorts the planning rule in their favour. The only minor exception to the general avoidance of this question came in brief discussions by Dreze and de la Vallee Poussin (1971) and by Malinvaud (1972), but their discussions were not thorough since the papers were more concerned with other matters. In recent years, however, particularly since the publication of Hurwicz's fundamental papers (1972), (1973), much work has been done on identifying incentive compatible planning mechanisms. These mechanisms, even if they do not necessarily encourage agents to reveal truthfully and directly the information a planner needs, at least lead to desirable outcomes even when agents do try to manipulate them. The reader who is completely unfamiliar with this field may wish to turn straight away to the early sections of the papers by Dasgupta, Hammond and Maskin and by Groves to see the incentive problem stated informally at greater length and then also formally. The papers in the present symposium have been collected and edited in collaboration with Ted Groves and Eric Maskin. They fall very naturally into an initial twofold classification, with the sole exception of Roberts. As he points out in the introduction to his paper, one can contrast global mechanisms in which each agent is concerned with the final outcome of the mechanism, with local mechanisms in which each agent is concentrating on the utility gain from each step of an iterative planning procedure. The first six papers and the last section of Roberts, all concern themselves with the global mechanisms. By contrast, the bulk of Roberts' paper, as well as those by Henry, by Tulkens and Zamir and by Schoumaker, all consider a particular local mechanism-namely, the specific adjustment procedure for planning with public goods which was proposed by Malinvaud (1972) and by Dreze and de la Vallee Poussin (1971). This procedure is called M.D.P. for short. The earlier papers showed that the procedure guaranteed convergence to a Pareto efficient outcome provided agents reveal their true marginal rates of substitution or demand prices throughout. Malinvaud (1972) also suggested that it would still converge, though rather more slowly, to a Pareto efficient outcome even if agents chose to misrepresent their true demand prices at each step in order to have their own utility increase as fast as possible. This slower convergence is demonstrated, for an economy with one public and one private good, by Roberts, who also shows the impossibility of guaranteeing Pareto efficient outcomes with a global mechanism, at least if agents just report their preference orderings. Henry refines Roberts' results for local mechanisms by restricting agents to announcing non-negative demand prices. He also considers the similar incentive problem which arises in a discrete version of the M.D.P. procedure, and shows that for this procedure too, convergence to a Pareto efficient outcome is guaranteed, given some mild assumptions about the agents' announcements of demand prices.