Concentration (the share of a market’s output produced by its largest firms) is a key proxy for measuring market competition. This column introduces a cross-country dataset to measure concentration at the industry level between 2000 and 2019. It captures both domestic production and international trade flows as well as business group linkages. It finds that the average concentration across countries increased by five percentage points between 2000 and 2019, with notable differences across geographic competition levels. These insights are important for understanding competition dynamics and informing antitrust and economic policies to foster a competitive environment.
The report on European competitiveness, authored by Mario Draghi and published in September 2024, introduced several proposals to reform competition policy in Europe with the aim of emphasising the weight of innovation and dynamic aspects of competition in the authority’s decisions (Draghi 2024). However, before taking action, it is crucial to get a clear understanding of the state and evolution of competition in Europe.
Concentration – defined as the share of a market’s output produced by its largest firms – is a primary metric used to proxy for market competition. Empirical studies have documented a rise in industry concentration since at least 2000 in the EU (European Commission 2024), the US (Autor et al. 2020, Covarrubias et al. 2020), and Japan (Kikuchi 2024). This trend has sparked an animated debate among academics and policymakers regarding its implications for competition. There are two main issues at the heart of this debate. First, concentration proxies for competition but does not measure it directly, with its relationship to market power depending on specific market characteristics (Syverson 2019). Concentration may arise from high barriers to entry or collusion, reflecting a lack of competition. On the other hand, it may also exist in a competitive market where heterogeneous firms producing differentiated goods vary in productivity, and there are sunk costs to enter the market. Second, methodological choices and data limitations significantly influence how concentration is measured and interpreted (Shapiro and Yurukoglu 2024).
Using innovative methodologies and a novel dataset to measure concentration in a cross-country setting, Calligaris et al. (2024b) show that average industry concentration in a set of European countries, complemented with data from Korea, Japan, and the US, increased by five percentage points (p.p.) between 2000 and 2019. Their approach enhances concentration measurement by aligning more closely with the ‘relevant markets’ defined by competition authorities across three dimensions: geographical boundaries (accounting for much of international trade), industry scope, and the connectedness of firms within business groups.
To address these issues, in a recent paper (Calligaris et al. 2024a) we develop a novel dataset combining official National Accounts data to accurately measure domestic production with detailed goods and services trade data at the industry level. The resulting representative dataset measures the effective market size of 151, mostly three-digit level, industries within mining, manufacturing, utilities, and non-financial services. It spans 15 European countries, Japan, Korea, and the US from 2000 to 2019. This is combined with cross-country data on sales of firms, linked to their global ultimate owner, from Orbis.
In a related paper (Calligaris et al. 2024b), we introduce three innovations to measure concentration:
- Calculating concentration at the geographic competition level (domestic, European, or global). Crucially, this indirectly accounts for international trade within a market, without relying on cross-country firm-level trade data, which are scarcely available.
- Defining industries at a more granular level (primarily three-digit) than previous representative cross-country studies. This aligns more closely with the narrow markets analysed by competition authorities.
- Considering affiliated firms’ linkages within business groups in each competition geography. This recognises the growing role of international mergers and acquisitions in explaining concentration trends across countries.
Defining the relevant geographical market
We devise a taxonomy to classify the geographical competition level for each industry. By comparing domestic sales with international trade flows and relying on information on trade partners, this taxonomy characterises each industry as competing at either the domestic, European, or global level. As the global economy becomes increasingly integrated, competition often transcends national boundaries. However, not all markets are equally globalised, with significant trade barriers persisting even within the EU in some industries. Accurately identifying the extent of globalisation in each industry, and accounting for the relevant geographic market of interest from a European perspective, is crucial for effective competition measurement.
Out of the 151 industries considered, 40 are classified as competing at the domestic level, 85 at the European, and 26 at the global level. This taxonomy can be a very useful tool for practitioners and policymakers in any application that requires defining the relevant geographic market (Calligaris et al. 2024a). We apply the taxonomy to measure concentration at the relevant geographic level, thus accounting for international trade between firms across the countries inside the market.
Measuring concentration at the granular industry level
Most existing cross-country research has been conducted at highly aggregated industry levels and a unique geographic scope (primarily national). Critics have questioned whether industry concentration accurately reflects product market concentration, which is more pertinent to consumer welfare and the focus of competition authorities. While product markets reflect consumer purchasing decisions, industry classifications adopt a producer’s perspective. Discrepancies between these two perspectives could lead to different conclusions about the state of competition (Benkard et al. 2023). In a forthcoming paper, we also show a strong correlation between granular industry-level concentration and product-level market concentration (Calligaris et al. forthcoming).
Accounting for business group linkages
Mergers and acquisitions have been on the rise. Firms may purchase other existing firms or establish new affiliates operating in the same industry to consolidate market share. It is therefore essential to account for ownership linkages between all connected subsidiaries operating in a single industry when measuring concentration. We apply the procedure of Bajgar et al. (2023), combined with the taxonomy of geographic competition, to account for cross-border linkages in internationally competing industries and measure concentration at the business group level.
The market share of the top four companies per industry increased by five percentage points between 2000 and 2019
Applying these methodological innovations, our analysis reveals that average concentration across the countries considered increased by five percentage points between 2000 and 2019. The results are reported in Figure 1. Concentration rose more in industries competing at the domestic level (six percentage points) compared to those competing at the European and global levels (four percentage points each).
Figure 1 Concentration across geographical buckets
Note: The chart shows the unweighted average across industries (and countries, for the domestic bucket) of CR4 levels (Panel A) and cumulative growth (Panel B). Industries included in the analysis are a mix of two and three-digit industries belonging to mining, manufacturing, utilities, and non-financial market services and are classified as either domestic, European, or global, depending on the taxonomy developed by Calligaris et al. (2024b). The countries included in the sample are BEL, DEU, DNK, ESP, FIN, FRA, GBR, GRC, HUN, ITA, NOR, POL, PRT, SVN, and SWE for the domestic and European bucket, while in the global one also JPN, KOR, and USA are included.
Source: OECD calculations.
Methodological innovation matters for the measurement of concentration
We investigate the role of different methodological decisions on the measurement of concentration levels and trends. Defining industries at more disaggregated levels results in higher concentration levels but similar trends (Panel A of Figure 2). At broad-level aggregations (i.e. at two-digit level) even the largest business groups are unlikely to cover all the activities of the industry, and therefore their market share at such an aggregate level is relatively low. In more narrowly defined industries, large players are more likely to be active in all specialisations of the industry and, therefore, account for a larger share of production.
Expanding geographic market boundaries beyond national borders, i.e. considering some industries as competing at European or global level, reduces both concentration levels and growth rates compared to measures computed solely at the national level (Panel B of Figure 2). Note that by aggregating firms’ activities at the relevant geographic level at which competition takes place, our concentration measure accounts for firms’ sales to other countries within the same geographical market. That is, it accounts for firms’ trade between the countries in the market, even without the use of firm-level trade data, which is rarely available in a cross-country setting. This is particularly important given the recent findings of Amiti and Heise (2024), who show that concentration in the US has not increased when accounting for the role of trade.
Figure 2 Concentration levels, different industry aggregations, and the role of taxonomy
Note: The chart shows the unweighted average across industries and countries of CR4 levels. In both Panels, the solid blue lines refer to the baseline measure of concentration used in the paper, computed using the taxonomy and at disaggregated industry level. In Panel A, the dashed ones represent concentration computed using the taxonomy but considering industries at A 64 aggregation level, while in Panel B, concentration is computed considering all industries as competing at the domestic level. Industries covered for this exercise include all the two-digit industries whose three-digit sub-industries belong to a unique geographical bucket, belonging to mining, manufacturing, utilities, and non-financial markets. The countries included in the sample are BEL, DEU, DNK, ESP, FIN, FRA, GBR, GRC, HUN, ITA, NOR, POL, PRT, SVN, and SWE. Industries competing at the global level also include JPN, KOR, and USA.
Source: OECD calculations.
Finally, accounting for ownership linkages between firms results in higher levels and faster growth of concentration between 2000 and 2019 (Figure 3). In the baseline concentration measure, the sales of subsidiaries belonging to the same business group and active in the same market are summed up. This baseline is compared to an alternative measure in which each subsidiary is considered as a separate entity. As expected, not considering the business group dimension leads to lower levels of concentration across all geographical buckets, with the difference being larger in the European market. When neglecting business group linkages, concentration is relatively flat in industries competing at the European and global level, whilst still rising in domestic industries. The greater rise in concentration when accounting for business group linkages suggests that the development and acquisition of subsidiaries has been an important component of the rise in concentration, especially in the European markets.
Figure 3 Concentration levels: Firms versus business groups
Note: The chart shows the unweighted average across industries and countries of CR4 levels across geographical buckets. The solid blue lines refer to the baseline average level. The dashed ones represent concentration when the ownership structure of business groups is neglected. Industries are a mix of two and three-digit industries belonging to mining, manufacturing, utilities, and non-financial markets. The countries included in the sample are BEL, DEU, DNK, ESP, FIN, FRA, GBR, GRC, HUN, ITA, NOR, POL, PRT, SVN, and SWE. Industries competing at the global level also include JPN, KOR, and USA.
Source: OECD calculations.
Overall, while the different methodological refinements affect both the average concentration level and its evolution, our findings consistently indicate rising concentration under most specifications. These insights are pivotal for understanding competition dynamics and informing antitrust and economic policies to foster a competitive environment.
Source : VOXeu
