Final Industrial Concentration Paper
March 09, 2022
It has become widely accepted that the U.S. economy is plagued by rising and excessive industrial concentration. In recent years, this view has become increasingly influential in policy discussions with both Democratic and Republican lawmakers raising concerns about industrial concentration and its economic implications.
The focus on industrial concentration has intensified over the course of the last year as the Biden Administration has made “combat[ting] the excessive concentration of industry” the focal point of its approach to competition policy. For instance, in July 2021, the Biden Administration released its “Executive Order on Promoting Competition in the American Economy” (Competition EO), which directed federal agencies and regulators to use antitrust enforcement to reduce industrial concentration and linked rising concentration to a host of ills, including higher prices, sluggish growth and stagnating wages.6 More recently, in January 2022, the Department of Justice (DOJ) and Federal Trade Commission (FTC) jointly announced that the agencies would seek to “modernize” the Horizontal Merger Guidelines and issued a Request for Information on Merger Enforcement citing the proposition “that many industries across the economy are becoming more concentrated and less competitive” as the primary motivation.
The notion that industrial concentration in the United States has reached excessive and harmful levels can be traced to three empirical studies discussed in detail below that use Economic Census data from the U.S. Census Bureau to analyze historical trends in industrial concentration. While these studies have been criticized by economists for focusing on industrial concentration –concentration within an industry as defined by the U.S. Census Bureau – rather than market concentration – concentration calculated using market shares in properly specified economic markets defined by consumer substitution patterns – the analyses nevertheless remain highly influential in policy circles. Thus, in this study, we set aside the (important) question of market definition and focus instead on investigating the underlying empirical foundations of the assertion that industrial concentration has risen to excessive levels in the U.S. economy.
The present concerns about industrial concentration in the United States are based on three empirical premises: (1) that industrial concentration is rising; (2) that industrial concentration is persistent – i.e., that concentrated industries tend to sustain existing levels of concentration or even become more concentrated; and (3) that industrial concentration is economically harmful. In this study, we use publicly available Economic Census data from 2002 to 2017 to investigate each of these premises.
Our methodological approach is designed to address several issues which have reduced the credibility of previous studies. First, we evaluate industrial concentration using the narrowest industry definitions available in the Economic Census data. Second, we use only the most reliable measures of industrial concentration available in the data. For the manufacturing sector, this is the Herfindahl-Hirschman Index (HHI), which is used by the DOJ and FTC in conducting merger reviews. For all other sectors, HHI is not available, and thus we use the four-firm concentration ratio (CR4), which represents the percentage of economic activity accounted for by the four largest firms within a given industry. Third, in characterizing overall trends in industrial concentration, we consider the entire universe of industries in the Economic Census data rather than selected samples of industries.
We also make five primary methodological contributions:
First, our analyses incorporate comprehensive data from the 2017 Economic Census, which were made public on a rolling basis from late 2018 to 2021, and therefore not considered in many previous studies of trends in industrial concentration.
Second, we examine not only average trends in concentration which have been emphasized in previous economy-wide studies of industrial concentration, but provide detailed analyses of changes across the entire concentration distribution. Examining the full concentration distribution allows us to investigate whether aggregate trends in concentration are driven by distinct underlying trends among more concentrated versus less concentrated industries.
Third, we analyze whether trends in industrial concentration exhibit a tendency towards mean reversion. This analysis is particularly important as the DOJ and FTC consider whether to include analysis of trends in industrial concentration in the merger review process. To the extent such a tendency towards mean reversion exists, it implies that trends in industrial concentration may reflect transient economic shocks that dissipate over time rather than structural economic changes.
Fourth, several industry case studies demonstrate that rising industrial concentration may often be the direct result of increased market competition and entry by new firms.
Fifth, in evaluating the relationship between industrial concentration and economic growth, we consider both variables related to consumer welfare (output growth) and variables related to labor markets (job creation and employee compensation).
Our main findings are summarized as follows:
There is no general trend towards increasing industrial concentration in the U.S. economy from 2002 to 2017.
- In both the U.S. manufacturing sector and the broader U.S. economy, industrial concentration has been declining since 2007.
- For manufacturing industries, average HHI fell from 821 in 2007 to 619 in 2017 – a decline of 203 points. As a result of this decline, the average HHI for manufacturing industries was 150 points below its 2002 level in 2017.
- For the broader U.S. economy, average CR4 fell from 36.9 percent in 2007 to 35.2 percent in 2017 – a decline of 1.7 percentage points. As a result of this decline, the average CR4 for all industries was approximately equal to its 2002 level in 2017.
- In both the U.S. manufacturing sector and the broader U.S. economy, the decreases in concentration were largest for the most concentrated industries. For instance, the 90th percentile of the HHI distribution for manufacturing industries fell from 1,904 in 2007 to 1,317 in 2017 – a decline of 586 points. Similarly, the 90th percentile of the CR4 distribution for all industries fell from 71.0 percent in 2007 to 66.8 in 2017 – a decline of 4.2 percentage points.
The evidence does not support the claim that high levels of industrial concentration have become a persistent structural feature of the U.S. economy.
- Higher concentration industries tend to become less concentrated over time while lower concentration industries tend to become more concentrated.
- Industrial concentration levels demonstrate a distinct tendency towards mean reversion suggesting that trends in concentration are influenced by transient economic shocks that dissipate in future periods.
- The share of economic activity accounted for by the most concentrated industries declined from 2002 to 2017. For instance, the share of economic activity accounted for by the most concentrated industries, with a CR4 of 90 percent or more, declined by more than 65 percent over this period.
The evidence does not support the claim that rising industrial concentration is generally associated with poor economic outcomes.
- Increases in industrial concentration are associated with output growth, job creation, and higher employee compensation.
- Evidence from several case studies shows that rising industrial concentration can be a direct response to increasing market competition.
The remainder of this paper is organized as follows. In Section II, we briefly outline the organizational structure of the U.S. Census Bureau’s Economic Census data and discuss the metrics available in the data to measure industrial concentration. In Section III, we review the recent literature on trends in industrial concentration in the United States. In Section IV, we provide an overview of our methodology for analyzing trends in industrial concentration in the U.S. economy and describe the construction of our data from the 2002, 2007, 2012 and 2017 Economic Censuses. In Sections V, VI, and VII, we use the Economic Census data to investigate whether industrial concentration in the U.S. economy is: (1) increasing, (2) persistent, and (3) economically harmful. Section VIII concludes with a discussion of the implications of our findings for policymakers.