A study of networking and location decisions in the advertising agency industry

This study investigates the interplay between the location decisions of advertising agencies and their desire for networking. We think of networks as a set of professional and business friendships and contacts---normally initiated and/or confined within close proximity---that allows members to share knowledge, expertise, information, and creative ideas. We show that these informal arrangements and non-market interactions have strong implications for the structure and efficiency of the advertising market.; Initially, we develop a model that provides for a micro foundation of network formation and knowledge sharing among advertising agencies, despite the industry's severe competition. Under incomplete information, agencies, which naturally possess more information than do advertisers about other agencies, undertake part of the search on behalf of advertisers towards betterment of the match. Early elimination of agencies in the search process eases competition, allowing for cooperation through "networks." These are small groups of related agencies, often connected by personal friendship, that share knowledge in the absence of a conflict of interest. Although agencies form networks to increase their chance of winning an account and raise their expected profit, the network formation game almost always results in a symmetric network size, where the chance of winning an account is independent of the network size. Surprisingly then, the formation of networks decreases the net profit of agencies because it introduces communication costs while revenue is unchanged. Nevertheless, competition among agencies forces them to form networks whose benefits accrue to advertisers.; In the second chapter, we work out a screening model of knowledge sharing and networking among single-unit advertising agencies and investigate its implications given heterogeneity in agencies' quality. Under a modest set of assumptions, the separation outcome is Pareto-undominated Nash equilibrium. A necessary condition for the existence of separating equilibrium (together with the rejection of the pooling alternative) is identified even in the presence of agglomeration economies from networking. This condition is then tested using confidential establishment-level Census data in the discrete location choice model. Our findings suggest that the necessary condition for separation is met and that separation and sorting on quality among agencies in their location decisions is strong.; Finally, in the third chapter, we investigate networking among close neighbors in Manhattan. We construct a model of local interactions that incorporates the spatial aspect of the contact within confined urban neighborhoods. We consequently identify a sharp decline in the benefits of networks by distance: agencies only benefit from others within 250 meters (surrounding city blocks). Interestingly, the benefits from immediate neighbors (in the same city block) are not much higher---a finding associated with severe local competition.