Communication Costs in Economic Mechanisms and Organizations

Stanford University, Stanford CA

Investigators

Abstract

A key function of an economic system is to acquire and disseminate the information necessary for consumers and firms to make socially desirable consumption and production choices. This function is often performed by market prices, which convey to individuals the relevant information, and motivating them to act in a socially efficient way. However, free markets may fail due to economies of scale, externalities, or the inability to price a large number of potential goods. In such situations, the "invisible hand" of the market is often substituted with the "visible hand" of managers in firms or with the "visible hammers" of auctioneers. This project will examine the design of such alternative economic institutions. One hypothetical economic mechanism would ask all participants to reveal their preferences and technology, and then implement an efficient outcome (the incentives to communicate truthfully can be provided with appropriately designed monetary compensation). The problem is that this mechanism would require a prohibitive amount of communication. The study will examine the design of economic mechanisms that economize on communication costs, using the models of communication costs developed in economics and computer science. One important application is the design of combinatorial auctions, which allocate heterogeneous objects among bidders (recent examples include auctions of spectrum licenses and online procurement). Since every bidder has a valuation for each subset of the items, and the number of such subsets is exponential in the number of objects, full communication of preferences would require prohibitive communication (e.g., with 30 objects, it would involve the announcement of more than one billion numbers). The study (joint with Noam Nisan, a computer scientist) will identify the communication burden of realizing socially efficient or approximately efficient allocations. In particular, it will examine how much communication is required to improve upon selling all objects as a bundle to the highest bidder, and how the answer depends on the designer's a priori knowledge of the structure of bidders' preferences. Another application is to the organization of firms. As noted by Ronald Coase in 1937, firms arise when the cost of "discovering what the relevant prices are" would be prohibitive. This idea can be examined in a complex coordination problem, in which pooling all agents' information to obtain an ideal decision may be practically impossible, but coordination may be achieved by "authority" of one of the agents. This model could explain why the "managerial task" of information processing and decision-making is concentrated in the hands of few, rather than being spread evenly among organization members. It can also identify the properties of the optimal management hierarchy, such as its height, spans of control, and the staff/operators ratio. It may also explain why higher-skilled individuals are often placed in higher positions, illuminating the role of organizational structure in determining the substitutability between workers of different skills, and consequently the market skill premium. Finally, it can be used to understand the organizational effects of information technology, which can reduce communication costs, changing the organizational structure and affecting the skill premium.