The dominance of top firms in specific industries is uncontroversial. Stories abound about the increasing national market share of companies like Walmart, Starbucks, Home Depot, or Amazon. In fact, several studies have shown that standard measures of national concentration have increased substantially in the US in the last few decades (e.g. Council of Economic Advisors 2016, Autor et al. 2017, Gutiérrez and Philippon 2017). A narrative has emerged which describes this increasing national market concentration as being responsible for declines in product market competition, as evidenced by higher profits and markups (Eeckhout and de Loecker 2017).
The problem with this argument is that in most industries markets are not national, they are local. The opening of a coffee shop in San Francisco does little to reduce the price of my morning cup in Chicago. The local retailer in Richmond does not compete with retailers in Washington, let alone in Dallas. The presence of transport costs and the imperfect substitutability between goods imply that markets are (at least to some extent) local and specific to particular products. Hence, if we hope to get clues about the evolution of competition over time, in most industries we need to measure product-market concentration locally, instead of nationally as is typically done. In a recent paper (Rossi-Hansberg et al. 2018), we examine local product-market concentration and dig deeper into the underlying market structure that has given rise to the increase in national concentration.
Local concentration has decreased, not increased, in the US
Our results turn the conventional narrative on its head. We find that, in contrast to the increasing national trend, local concentration has decreased, on average, for all major sectors and for the large majority of narrowly defined industries (industries accounting for roughly 70% of US employment and sales). The findings hold for a variety of geographic definitions (CBSA, county, or ZIP code) and industrial aggregations (from 2 to 8 digits). Naturally, the negative trends are more pronounced when the definition of the market is narrower (ultimately, aggregation will get us back to national measures). The results also hold for a variety of measures of concentration, including the Herfindahl-Hirschman Index, the adjusted Herfindahl Index, and the share of the top firm in a market.
Between 1990 and 2014, industries with decreasing local concentration can be found in all major sectors. Perhaps more surprising, 8-digit industries with diverging trends (increasing national and decreasing local concentration) are pervasive as well. They can be found in virtually all 2-digit industry aggregations. How can this be? How can national concentration in a narrowly defined industry increase while, concurrently, local concentration declines?
What is driving the decline in local concentration?
To shed light on the forces behind our finding, we study the role that top firms have played in generating these diverging trends (we define an industry’s top firm based on 2014 sales in that industry). Excluding the establishments of the top firm (or top three firms) in an industry, we recalculate the trends in national and local concentration. The results are striking: among industries with diverging trends, top firms have, of course, contributed to the increase in national concentration, but they have also contributed significantly to the declining trend in local concentration. That is, top firms have made the decline in local concentration more pronounced.
What is going on? The answer is simple. Top firms are expanding by adding more establishments in new locations. When they do, they increase their national share of sales while also decreasing concentration in the markets where they enter. On average, a new local establishment of a top firm does not drive out other local producers one-to-one; instead, it shares the market with them, which decreases local concentration (other local producers can be single-establishment local firms or other national firms with many establishments).
This view is reinforced when we compute event studies of the effects on local concentration when top firms open a local establishment. We find that, in industries with diverging trends, when a top firm opens an establishment, concentration falls and remains down for at least seven years. In contrast, if we do not take into account the establishments of the top firm, local concentration remains roughly constant. Top-firm establishments do not eliminate local competition; they simply become another producer in a less concentrated local market.
Concluding remarks
So no, top firms have not increased relevant measures of local concentration; they have contributed to their decline. This is not the case in all industries, of course, but it is the case on average and in industries that account for most of the economy’s employment and sales. The link between concentration and competition is complicated and depends on the particularities of an industry’s technology, as well as entry costs and related dynamic considerations. However, we know of no mechanism in the literature that could link the large documented declines in local concentration with declines in local product-market competition. In fact, virtually all standard theories would imply the opposite, that by decreasing local concentration, top firms have made product-markets more competitive.
Of course, even if top firms increase local competition, they could still abuse their increasing national share. For example, they could be gaining monopoly power in labour markets, or be acquiring political power to change regulations and other policies in their favour. However, the particular nature of the way top firms gain national market share, namely by adding local establishments, seems to prevent them from exploiting market power in product markets.
How do we square these results with evidence on increasing mark-ups and profit shares? This is still somewhat controversial, but our results might shed some light on why studies seem to be finding contradictory evidence on mark-ups and profits in particular markets and sectors. Perhaps more controversial is the interpretation of specific statistics as indicating the presence of market powers. Firms do not have to compete in prices; they can compete in quality, bundled services, locations, and other dimensions. Profits might be payments to unmeasured, or poorly measured, intangible assets and skills, and firms’ reports could be responding to tax incentives and regulatory standards. One thing we know, though, is that simple measures of average local concentration have been falling for the last 25 years. So enjoy the variety of coffee shops on Main Street!
References
Autor, D, D Dorn, LF Katz, C Patterson and J Van Reenen (2017), “The fall of the labor share and the rise of superstar firms,” NBER Working Paper No.23396.
Council of Economic Advisors (2016), Benefits of competition and indicators of market power.
De Loecker, J, and J Eeckhout (2017), “The rise of market power and the macroeconomic implications,” NBER Working Paper No. 23687.
Gutíerrez, G, and T Philippon (2017), “Declining competition and investment in the US”, NBER Working Paper No. 23583.
Rossi-Hansberg, E, P Sarte and N Trachter (2018), “Diverging trends in national and local concentration”, CEPR Discussion Paper No. 13174.
