Inequality and the Organization of Knowledge

Since the seminal work of Lawrence F. Katz and Kevin M. Murphy (1992), the study of wage inequality has taken as its starting point a neoclassical constant-elasticity-of-substitution production function using as inputs capital and lowand high-skill labor. This approach assumes that the organization of production is fixed and determined by a particular specification of technology, and it ignores both the source of the interaction between workers and the organizational aspects of this interaction. These shortcomings are particularly important in light of growing empirical evidence that points, first, to the importance of decreases in the cost of processing and communicating information and, second, to the complementarity between organizational change and adjustments in the distribution of wages (e.g., Timothy F. Bresnahan et al., 2002). This paper argues that theories that seek to guide empirical research on these areas must put knowledge and information at the center of the analysis of organizations and link the organizational structure with aggregate variables via equilibrium frameworks. In Garicano and Rossi-Hansberg (2003), we present a model of this kind. It determines the patterns of organization, as manifested by the communication and specialization patterns, and the implied wage structure, that result from different costs of acquiring and communicating information. Here, we present a simple variant of this theory that allows us to focus on one of the main aspects of that framework: the sorting of agents into teams and the wage and organizational structure that accompanies that sorting. We use this simple model to analyze the changes in organization and wages that result from a very specific type of technological change: a reduction in the cost of communicating knowledge or information. This model allows us to consider the effect on within-class wage inequality, and the impact of information technology on the creation and form of organizations (e.g., size distribution of hierarchies). However, because knowledge is exogenously given, and agents cannot invest in learning, an important margin of the model in Garicano and Rossi-Hansberg (2003) is fixed, namely, the degree of “decentralization” or the extent to which problems are solved at lower levels. That model allows the simultaneous study of the acquisition of knowledge, spans of control, and matching in equilibrium. Moreover, it goes beyond the current analysis in that it allows for organizations with an unconstrained number of layers, and in that it studies two aspects of the impact of information technology: communication technology (like here) and the technology to acquire knowledge or information (e.g., processing power through cheaper database access).