Agents in a multiagent system are not typically entirely self-sufficient; instead, they frequently need to enlist other agents to perform tasks for them or to exchange goods or services with them. This creates a problem: how can an agent efficiently locate other agents to work or trade with? As the number of agents grows, the cost of this computation can become prohibitively large. One solution to this is for the system to self-organize into smaller groups of agents. In this paper, we apply the idea of congregating to a model of an information economy. We illustrate how participants in this economy can self-organize into a set of markets such that agents are able to find suitable partners while retaining low computational costs. We show how congregating can help allocation problems scale to large populations by allowing agents to interact locally.