Order of Entry and the Geographic Distribution of CPG Brand Shares: an Application of Endogenous Sunk Costs Theory∗

Preliminary and Incomplete. Please do not circulate. Using a unique database for consumer package goods (CPG) brands, we analyze market structures across the largest US metropolitan markets. We find several striking empirical regularities in brand share patterns across markets. For each of the 31 industries covered in our database, we find that geographic markets are routinely characterized by a ”dominant-share ” brand. That is, markets are concentrated. However, the identity of the top brand varies from market-to-market. Furthermore, the identity of the dominant brand tends to be similar for geographically “close ” markets. Correspondingly, for a given brand we observe substantial variation in the average share across markets. At the same time, we observe strong spatial dependence in these shares. Finally, across brands within a CPG industry, we see very strong asymmetries in brand shares. We explain the underlying economics driving these patterns using a model of endogenous sunk costs with sequential entry. We test this theory using three specific industries for which we collect detailed historic information on the order-of-entry. In assessing the role of order-of-entry, we leverage a unique feature of our data whereby we observe variation across markets in the identity of the first-entrant. This property enables us to identify separately the first-mover effect from heterogeneity in relative firm competencies. Our data show that in each of the industries covered, concentration levels are bounded away from zero even in the largest markets. Moreover, historic order-of-entry patterns appear to be the best predictor for the asymmetric share patterns. Observed spatial dependence in the entry patterns also serves as a good predictor for the dependence in market shares.