Simcha Barkai describes the results of new research examining the impact of antitrust enforcement on U.S. economic activity with co-authors Tania Babina, Jessica Jeffers, Ezra Karger, and Ekaterina Volkova. Enforcement, they find, increases economic activity.
In new work, joint with Tania Babina, Jessica Jeffers, Ezra Karger, and Ekaterina Volkova, we hand-collect a complete history of Department of Justice (DOJ) antitrust lawsuits from 1971 to 2018. We use this dataset to study the real effects of these lawsuits on economic activity. Our paper provides the first systematic evidence of the real effects of DOJ antitrust enforcement. We find clear evidence that antitrust enforcement increases the level of economic activity (measured as employment), the creation of business, average wages, and the labor share.
A recent literature documents rising market power in the U.S. and its negative effects on aggregate wages, investment, and productivity. These patterns have sparked a renewed interest by policymakers, researchers, and the media in competition policy and antitrust enforcement. This wave of interest in antitrust enforcement echoes similar cycles of attention to anticompetitive behavior that led to the passage of the four key federal laws regulating anticompetitive behavior, beginning with the Sherman Act in 1890. But despite over a century of antitrust enforcement, there is little systematic empirical work measuring the effects of antitrust enforcement on economic outcomes. A key challenge to empirical research in the area of antitrust enforcement is the absence of standardized data on antitrust enforcement actions.
Our main source of information on antitrust enforcement is legal summaries of DOJ antitrust lawsuits that are compiled for lawyers and legal scholars. These case summaries cover all 3,055 DOJ antitrust lawsuits filed in court over the years 1971 to 2018. A typical case summary is two to three paragraphs long and describes the initiation and resolution of an antitrust lawsuit filed by the DOJ Antitrust Division in federal court. We manually review these summaries and collect a large number of standard variables such as the alleged violations, the name of the district court, and the case filing date. In addition to these standard variables, we collect detailed information on the geography and industry of alleged anticompetitive behavior. Specifically, we collect information that describes the location of the seller and the geographic scope of the alleged violation (ranging from city to international) and we manually match each case to a standard industry code. To ensure maximum accuracy, each case is read and coded independently by two individuals, and we then compare the entered values and reconcile disagreements.
It is important to distinguish between two types of antitrust enforcement, both carried out by the DOJ Antitrust Division. The first are conduct cases, which are all cases in which the DOJ alleges the existence of anticompetitive behavior, such as agreements between firms to fix prices. The second are merger and acquisition cases, which are efforts by the DOJ to prevent the merger of two firms when it believes that this merger could potentially lead to lower competition in the future. While our data collection covers both types of cases, our empirical analysis only studies conduct cases. The reason for this is simple: if the DOJ is successful in a conduct case, this should improve competition and economic outcomes. However, if the DOJ is successful in a merger case, this prevents a worsening in competition (but no improvement) and therefore should leave economic outcomes unchanged.
Our work highlights several key trends in DOJ antitrust enforcement. First and foremost, the annual number of DOJ antitrust enforcement actions, which had been increasing for nearly a century, declines sharply since the start of the early 1980s. Over this time period, it does not appear that the DOJ Antitrust Division changes the composition of cases that it pursued. Conduct cases make up a majority of cases in nearly every year and account for over 70% of cases in our sample period. While there are years of intense DOJ focus on specific economic sectors (e.g., construction in the early 1980s), the composition of sectors targeted has not greatly changed since the 1980s.
Turning to economic analysis, our goal is to understand the impact of DOJ antitrust enforcement on economic activity in the targeted industry. There are many reasons that an industry might grow or increase wages in any given year, many of which are unrelated to DOJ enforcement actions and changes in the degree of competition.
In order to construct a counterfactual for how a targeted industry would have evolved in the absence of an antitrust lawsuit, we focus our analysis on non-tradable industries, which are those industries in which businesses serve a limited and local geographic market. Specifically, we compare outcomes in industry-states targeted by a DOJ antitrust lawsuit (e.g., grocery stores in Massachusetts) to outcomes of the same industry in other states not targeted by the lawsuit (grocery stores in other states). This comparison accounts for common changes in both the production technology of and demand for the products of the targeted industry. In the example of grocery stores, this comparison can account for improvements in supply chain management and scanner technology as well as variation in demand from households that is common across geographic locations.
Our comparison of industry-states targeted by a DOJ antitrust lawsuit to outcomes of the same industry in other states not targeted by the lawsuit does not make sense for tradable industries, which are those industries in which businesses serve customers in all locations. For example, if the DOJ targets a car manufacturer in the state of Georgia, this case will indirectly impact all of its competitors including those located in Michigan. (grocery stores in Massachusetts do not compete for the same customers as grocery stores in Michigan or Georgia).
We combine our hand-collected data on DOJ antitrust enforcement, aggregated to the level of an industry-state-year, with confidential establishment-level microdata from the U.S. Census aggregated to the same level of observation. Using the combined data, we measure the effect of antitrust enforcement on the level of economic activity, business formation, average wages, and the labor share.
We find that DOJ antitrust enforcement induces a lasting increase in economic activity, measured as an increase in employment of around 5%. We present year-by-year estimates of the effect of antitrust enforcement on log employment, measured in a window of eight years around the filing of the DOJ antitrust lawsuit. This exercise consists in estimating a separate effect for every year after the antitrust enforcement and it allows us to see how such effect changes over time. . We then repeat the analysis in a difference-in-differences setting, which instead estimates the average effect over the entire post antitrust enforcement period, and find a long-run increase in employment of 5.4%. The estimate of the difference-in-differences analysis is similar in magnitude to the estimates in the later years of the year-by-year analysis, which implies that there is no later reversion or decline in employment, even though the average post-period length is 25 years.
We also find DOJ antitrust enforcement also induces a lasting increase in business formation. Year-by-year estimates show a clear and gradual increase in the number of establishments in targeted industry-states starting in the year of the lawsuit and stabilizing at an increase of nearly 3%. Difference-in-differences analysis shows a long-run increase in the number of establishments of 2.9%. The estimate of the difference-in-differences analysis is similar in magnitude to the estimates in the later years of the year-by-year analysis, which implies that there is no later reversion or decline in the number of establishments, even though the average post-period length is 25 years.
Finally, we study the effects of DOJ antitrust enforcement on payroll, sales, and the labor share, defined as the ratio of payroll to sales. We find an increase in payroll that exceeds the increase in employment, meaning that DOJ antitrust enforcement increases average wages. In addition, we find an economically smaller increase in sales (relative to employment) that is statistically insignificant. While we do not have separate measures of the quantity and price of output, the increase in production inputs (employment), together with a proportionally smaller (and statistically insignificant) increase in sales, strongly suggests an increase in the quantity of output and, at the same time, a decrease in the price of output. Last, we find a clear increase in the labor share.
In summary, we find that DOJ antitrust enforcement actions lead to a long-run increase in the level of economic activity, business formation, average wages, and the labor share.
There are three potential limitations to our research that may lead us to understate the overall effects of DOJ antitrust enforcement. First, our analysis is not able to capture the effects of general deterrence. Large efforts to detect and prosecute economic crimes are likely to reduce anticompetitive misconduct by firms. Second, due to the challenges of constructing a credible control group, our analysis is not able to study the effect of antitrust enforcement on nationally dominant firms. To the extent that these cases provide unique economic benefits they are not captured in our results. Last, it is possible that spillovers bias estimates toward zero. Once the DOJ Antitrust Division brings a case against a particular industry, there may be non-targeted firms in the same industry in different states that had been engaged in anticompetitive behavior but stopped after they learned of the lawsuit. This could lead to increased competition in the control group, thereby biasing our estimates toward zero, leading us to understate the true effects of antitrust enforcement.
To encourage future research on antitrust enforcement, we will make our hand-collected data available to other researchers.
Author Disclosure: We are grateful to the Washington Center for Equitable Growth for providing us with the financial resources needed to carry out the data collection and analysis. Any views expressed are those of the authors and not those of the U.S. Census Bureau. The Census Bureau’s Disclosure Review Board and Disclosure Avoidance Officers have reviewed this information product for unauthorized disclosure of confidential information and have approved the disclosure avoidance practices applied to this release. This research was performed at a Federal Statistical Research Data Center under FSRDC Project Number 2090. (CBDRB-FY23-P2090-R10259).
Articles represent the opinions of their writers, not necessarily those of the University of Chicago, the Booth School of Business, or its faculty.
