An increasing body of empirical evidence documents trends of rising concentration, profits, and markups in many industries around the world since the 1980s. Two major criticisms of these studies are that concentration and market shares are poorly measured at the national industry level and that firm level revenues are a poor indicator of product sales. Indeed, because of data limitations, almost all existing studies measure concentration based on industry classifications and/or on firm balance sheet data, mostly aggregated at the national level.
This is highly problematic. Industry classifications of products may be either too large or too small in an antitrust sense. They may be too large since they include products that may not be substitutes, thus potentially including suppliers, customers, and non-competitors. They may be too small if they do not include relevant substitutes. Like the product market dimension, the geographic dimension of national industry-level aggregates may also be either too large or too small. They may be too small if relevant markets are actually at a supra-national level, such as worldwide or other groups of countries, or they may be too large if markets are actually more local than national. Even using more disaggregated data, such as census data, at the regional level does not completely solve this problem as market definition often does not coincide with geographic boundaries.
In a recent study, BCCP Fellows Pauline Affeldt, Tomaso Duso, and Joanna Piechucka, along with co-author Klaus Gugler, propose assessing the issue of concentration by using a completely different data source. They base their analysis on a novel dataset constructed by analyzing the merger control decisions of the European Commission. They collected information on almost the complete population of DG COMP merger decisions from 1990 to 2014, generating a dataset comprising 5,196 merger decisions. Since in each merger case potentially different markets – either in terms of products or in terms of geography – are affected, the final dataset contains 31,451 antitrust markets. Yet, because market shares are not always or fully reported, concentration measures can only be calculated for around two-thirds (over 20,000) product/geographic antitrust markets affected by over 2,000 mergers.
With this data at hand, they show that the concentration measures in these relevant antitrust markets are larger, by a factor of four to ten times, than what the literature documents so far. They also confirm that concentration has indeed increased over time, on average. However, they document that there is a great deal of heterogeneity across several dimensions. The extent of the geographic market as well as the broad sector of activity play a crucial role in this assessment.
Concentration appears to have increased more in broad worldwide markets than in more narrowly defined national markets. Moreover, concentration seems to have increased more in the service sectors than in manufacturing. Even within these broad sectors, they observe quite some heterogeneity across and within industries.
They further identify important elements that correlate with these concentration measures. Most importantly, barriers to entry are unambiguously positively correlated with concentration, irrespective of time periods, sectors of activity, and geographical market dimension analyzed. Although strict past merger enforcement negatively correlates with concentration, it appears that this correlation was stronger in the 1995-2004 period than thereafter. The intangibility of investments displays a consistent positive correlation with concentration only for wider than national – EU and worldwide – services markets. In contrast, it is negatively correlated with concentration in national markets.
Their main conclusion is that a strict merger and, more generally, competition policy enforcement that reduces barriers to entry are key tools to keep markets open and competitive. However, tearing down barriers to entry is not the sole task of antitrust authorities. Other policy areas such as regulation, institutions setting norms and standards, as well as international cooperation agreements must contribute. Notwithstanding this conclusion, there are circumstances in certain antitrust markets – such as high intangible asset industries in geographically wide services markets – where increasing concentration may indeed be likely related to increasing efficiency. It is the task of antitrust authorities to strike the delicate balance between these forces.
The full paper “Market Concentration in Europe: Evidence from Antitrust Markets” is available as DIW Discussion Paper No. 1930.
This piece also appeared as a promarket column and as a DIW Wochenbericht (in German).