THE THEORY OF SYNDICATES

WE SHALL DEFINE a syndicate to be a group of individual decision makers who must make a common decision under uncertainty, and who, as a result, will receive jointly a payoff to be shared among them. Our concern is to analyze the decision process of a syndicate when the members have diverse risk tolerances and/or diverse probability assessments of the uncertain events affecting the payoff. Of particular interest is the possiblity of constructing a surrogate "group utility function" and a surrogate "group probability assessment." Such constructions potentially have a role in the theory of finance; e.g., for determining the forms of organizational charters and financial instruments, as well as the modes of delegating the group decision process to professional managers. The present treatment, however, is confined to tractable features embodying only a small measure of the complexity of practical situations. Of comparable importance are the ramifications for welfare theory; in particular, we shall be able to specify conditions under which Pareto optimal behavior by the group satisfies the Savage axioms [15] foy consistent decision making under uncertainty, and to isolate the inconsistent characteristics in the contrary case. Arrow's original treatise [1] has been the source of most of the work on group decision theory. Marschak [13], Radner [17], and Bower [6] have considered the case of a team, in which there is a joint utility function for the members. Harsanyi [9] and Theil [16] have considered the criterion that the group decisions satisfy the Von Neumann-Morgenstern axioms, and others. Madansky [12] has imposed the "external Bayes axiom" in the case of a common utility function but differing probability assessments among the members. Christenson [7] has constructed an axiomatic system for the case of an investment banking syndicate that is a special case of the present study, except for certain institutional factors. Borch [3, 4, 5]