Information, Managerial Choice and Stockholder Unanimity

Abstract

Stockholder unanimity plays a key role in the theory of production. Under certainty, profit maximization is justified as being in the interest of all stockholders. But the introduction of uncertainty renders profit maximization meaningless, since production decisions do not give rise to a unique level of profits. A fundamental question is whether stockholders with diverse expectations and attitudes towards risk will support unanimously a particular set of production decisions. Without unanimity, firms' decisions will depend upon the weighting of diverse stockholder and managerial preferences, with the possible non-existence of a preference ordering over actions by the firm.' And without a theory of the firm, a substantial fraction of positive economics disappears. Several recent papers have studied the relationship between financial and production under uncertainty. Implicitly or explicitly, stockholder unanimity has been a property of these models. Two sets of conditions have resulted in stockholder unanimity. The first set, studied by Wilson (1968) and Rubinstein (1974), requires identical investor expectations and tastes belonging to the linear risk tolerance class of utility functions. The second set of conditions requires the space of returns generated by marketed securities to contain the marginal returns resulting from small changes in the firms' decisions. Ekern and Wilson (1974) showed that spanning was a sufficient condition for ex-post unanimity. Leland (1973) showed it was necessary as well, and showed that, with an additional competitive price assumption, spanning assured ex-ante stockholder unanimity.2 A number of economic environments are consistent with the spanning property, including the complete markets models of Arrow (1964) and Debreu (1959), and the incomplete models of Diamond (1967), Leland (1974, 1975), and Forsythe (1975). While providing a class of environments consistent with stockholder unanimity, both sets of conditions seem overly restrictive. Individuals possess different information and different expectations. And investment projects or other decisions resulting in marginal changes in returns typically have project-specific risks which are not spanned by currently marketed securities. Furthermore, the two sets of conditions consistent with unanimity give little scope to the role of the manager. Clearly managers cannot have better information than investors if expectations are identical. And wvhile spanning permits different expectations, the optimal decisions by the firm will be independent of any particular set of expectations, including those of the manager. Why should a manager gather information if (as these models predict) better decisions will not result? In short, current approaches to stockholder unanimity are inconsistent with the fact that managers typically have (and are urged to obtain) better information about firm-specific risks. While informational asymmetry has been considered in models of principal-agent and hierarchical relationships (Ross (1973), Mirrlees (1976)), it has not been considered in models where there are multiple stockholders (principals) with divergent tastes and expectations.