Concentration Levels in the U.S. Advertising and Marketing Services Industry : Myth vs. Reality

This paper analyzes changes in concentration levels in the U.S. Advertising and Marketing Services (A&MS) industry using publicly released data that have been largely ignored in past discussions of the industrial organization of this industry, namely those available from the U.S. Census Bureau's quinquennial Economic Census and the Service Annual Survey. We define the A&MS industry in terms of nine sectors, each of which is represented by a separate 5 digit NAICS category. In so doing, we have sought to redress some of the measurement problems surrounding estimates found in the existing literature. Our main findings are threefold. First, in the case of the core and largest sector, Advertising Agencies, firm level concentration as measured by Herfindahl-Hirschman Index (HHI) increased slightly but remained relatively low from 1977 to 2002. All of the HHI estimates readily satisfied the standard widely used to characterize an industry as "unconcentrated." We find mixed support for the hypotheses that the ranks of mid-sized agencies were depleted by ongoing waves of mergers and acquisitions and resulted in a polarized size structure. The size distributions of agency revenue have become more polarized in the sense that over time they appear more skewed, more dispersed, and exhibit greater inequality. The share of total receipts realized by small agencies fell while that of large agencies rose. However, the position of mid-sized agencies appears to have changed little over the period 1977-2002, as measured by the shares of agencies and receipts they represent. Second, concentration levels in 1997 and 2002 varied across the nine sectors comprising the A&MS industry, but all were within the range generally considered as indicative of a competitive industry. Third, we developed concentration ratios at the level of holding companies (HC's) and find that the four largest HC's captured between a fifth and a quarter of total revenue from the A&MS industry, a share that remained quite stable over the period, 2002-2006. These estimates are lower by an order of magnitude than estimates often cited in the trade press. Reasons for the discrepancy are discussed