We find three sets of empirical evidence in line with increasingly stronger economies of scale.
Timing of rising concentration in an industry is stronger in periods with higher investment intensity in R&D and IT. Rising concentration is also stronger in periods with more breakthrough patents.
Breakthrough patent data from Kelly et al. (2021), available for manufacturing subsectors plus mining, construction, and agriculture.
Rising concentration aligns with greater prevalence of fixed operating costs.
Industries that experience higher increases in concentration are also the ones that experience higher growth in real gross output; correspondingly, their shares in the economy expand as well. This feature is consistent with economies of scale expanding the total volume of production.
Globalization since around the 1980s does not appear sufficient to account for the evidence by itself, but having access to broad markets (domestically or internationally) could have supported. the influence of economies of scale.
International trade for the U.S. (relative to GDP) was largely declining before the 1960s, and trade expansion began around 1970s. In our data, rising concentration for manufacturing was stronger before the 1970s (whereas rising concentration was stronger in non-tradable sectors in the period of globalization). The timing suggests that liberalization of international trade does not appear sufficient. In addition, in a simple model, if the reduction in barriers to international is the only force, then concentration in U.S.-based production would increase when exports are included in sales, but not when exports are excluded from sales. In the data, sales concentration increases even when exports are excluded from sales. Nonetheless, although trade by itself does not appear to account for the entire story, having broad markets (domestically or internationally) could increase the appeal of economies of scale
The past century witnessed several regimes of antitrust enforcement, but rising corporate concentration has been a secular trend throughout different antitrust regimes. Other regulations also do not seem to fully explain the secular trend and the timing of rising concentration in different industries.
Since concentration does not necessarily measure competitiveness (higher concentration can be associated with less competition or with more depending on the setting), our findings do not speak to the strength of market power or the effectiveness of policies targeting market power. What we can speak to is the following question: if we are interested in the role of large firms in the economy (e.g., the share of assets and sales in the economy that belong to larger firms), do regulations have a major impact? Over the past century, antitrust policies shifted through several different regimes. We also do not observe any significant relationship between our main facts and standard aggregate antitrust enforcement measures such as the annual number of antitrust cases filed by the Department of Justice (DOJ). Overall, we do not find evidence that antitrust shapes the economy-wide business size distribution, although it could have a more visible impact on the market for a particular product (which is closer to the domain of antitrust analyses).
Several other types of regulations may also affect the size of businesses, such as restrictions on size (e.g., interstate banking restrictions), subsidies for small businesses, or policies that affect the dissemination of knowledge. To explain our findings, such policies need to have shifted in favor of large firms over the past 100 years. Furthermore, they need to have been particularly important for the growth of large firms in manufacturing rather than in services in the early 20th century, and then switched focus in the later decades. At the moment, we are not aware
of such a pattern in regulatory policies.
Recent research also discusses lower search frictions, declining population growth, or low interest rates as potential contributors to rising concentration.
It is less clear why such changes would have affected different industries at different points in time (i.e., why they would result in stronger increases of concentration in manufacturing in earlier decades and stronger increases of concentration in services, retail, and wholesale in earlier decades). U.S. population growth declined since the 1960s and real interest rate declined since the 1980s--these forces may contribute to a higher level of concentration in recent decades, although they may not be the full story over the long run.