Knowledge Sharing and Sorting: The Case of Advertising Agencies∗

This paper provides a model of knowledge sharing and networking among the single unit advertising agencies and investigates the implications of this model in the presence of heterogeneity among the agencies. In a stylized model, we show that under the modest set of assumptions the separation outcome is supported as undominated Nash equilibrium. We reject the possibilities of pooling equilibrium even in the presence of general economy of scale from networking. This results in the concentration of the high quality agencies in the locations with higher wages and rents. We derive the maximum profit of an agency and show that the conditions that guarantee the prevalence of separation equilibrium have strong implications in the specification of the agency’s profit function. We estimate the parameter of the profit function, using the Census confidential establishment level data, and show that the networking effect and separation are the predominant factors in the profit function of agencies, and consequently, in their location decisions. ∗I am very grateful to Vernon Henderson and Rajiv Vohra for all of their advice and supports. I also thank Andrew Foster, David Weil, and others at Brown for their many insights and suggestions. All remaining errors are my own. †[email protected]