The rise of industrial concentration throughout the US economy has brought antitrust policy to the center of public debate. In many different sectors, a small number of large businesses are increasingly dominant. An increasing popular view amongst both academics and policymakers holds that this is a problem, one that can be reversed in large part by more aggressive antitrust enforcement. The other side of this debate normally takes a benign view of the rise of big business. Here I outline a third, alternative position: that the rise of industrial concentration is a problem, but one that antitrust law will probably not be able to systematically address. Instead, we may have to look to policies that more deeply address the underlying causes of the growing technological distance between big and small business.
By Gabriel Unger1
I. INTRODUCTION
Interest in antitrust law is cyclical. But the cycles stretch over many decades, to the point that a young practitioner or academic in, say, 2021, might not realize the extent to which debates over antitrust policy today so closely mirror debates that took place in the 1960s and 1970s. And sure enough, a historically familiar perspective has ascended in policy circles today: that a small number of big businesses have taken over a large amount of market share throughout the U.S. economy, and this rise in industrial concentration, caused by excessively lax antitrust enforcement, is bad for the economy, and must now be reversed by much more aggressive antitrust enforcement. Votaries of this position (“Neo-Brandeisian” to its friends, “hipster antitrust” to its skeptics) alternatively assert direct harm to consumers in the form of higher prices, or – and perhaps more commonly – broader harm to the process of competition, even in the absence of higher prices.
I would like to briefly suggest here that this view, though well-intentioned and familiar, is predicated on both a confused understanding of what the basic situation of the U.S. economy is today, and a confused understanding of what antitrust law can and cannot accomplish. In its place, I would like to offer an alternative view that might be more productive for concerned policymakers to achieve their ultimate goals.
II. THE BASIC SITUATION OF THE U.S. ECONOMY TODAY
Over the past four decades, the U.S. economy has become significantly more concentrated. In most sectors of the economy, the market share of, say, the largest four firms (“CR4,” one popular measure of concentration), has increased – for instance, from less than 15 percent in the average 4-digit SIC retail sector in 1980, to over 30 percent by 2012.2 And even these conventional measures (within narrowly defined sectors) obscure the fact that there has also been a rise in horizontal integration: a large multi-product corporation might be expanding both its market share within given 4-digit sectors while also expanding the number of distinct 4-digit sectors it is now operating in. (For example: the largest seller of groceries – SIC 5411 – in the U.S. today is Wal-Mart, a company which would usually be listed first in SIC 5311 – Department Stores. We have – so far – no good metric of industrial concentration that parsimoniously expresses both vertical and horizontal concentration at the same time).
In public life and media discussion about monopolies, the face of “Big Business” is often a Silicon Valley giant: Facebook, Google, etc. But the trend of rising industrial concentration is clearly seen throughout almost every major sector of the U.S. economy. The data shows that in sectors broadly spanning the economy, from grocery stores to animal slaughterhouses, to cardboard box manufacturing, a small number of very large firms have substantially raised their market share.
This, however, is the consequence of deeper economic developments. The IT Revolution of the past forty years has disproportionately benefited large firms. Large firms have always been, on average, more productive and more capital intensive than small firms. But over the past forty years, large firms have particularly benefited from large investments in IT capital, investments that often only make sense at scale and in informationally intensive business models (e.g. simultaneously involving many different customers, suppliers, and product lines). These investments have in turn made these large firms even more productive. There has been a documented increased in the productivity differences between the very largest firms and the rest of the firms in many sectors, which have in turn led to the growing differences in market share.
So the basic situation of the U.S. economy is one of increased inequality on the supply-side, in terms of technology and productivity. In most sectors, there is a small number of very large, very productive firms, and a long tail of much smaller, less productive, and less capital intensive firms. Most of the growth in market share in the large firms over the past forty years has come from internal growth (“organic”), on the strength of their superior productivity, not from M&A. This phenomenon of growing supply-side inequality even has a geographic and perhaps more familiar analogue: a small number of very productive cities in the U.S. today (San Francisco, New York City, Boston Chicago, etc.) are pulling away from the rest of the country in terms of basic economic measures like income per capita – a growing regional divergence – while in an earlier historical period (1880-1980) both U.S. States and U.S. cities were converging towards each other in economic performance.3
None of this is to say there are not specific cases where it looks like some large business is inefficient but excessively protected by a suboptimal regulatory environment (there are many convincing case studies to this effect, e.g. for ISPs, or the hospital sector). But the data and the latest systematic empirical studies we have today largely suggest the picture of the U.S. economy above, of a growing divide in technological and productive capabilities between the largest firms, and the rest of the firm size distribution.
This supply-side inequality is a problem because it ends up abandoning too many firms, too many workers, and too much capital, to inefficient forms of production. An economy with an excessively long tail of unproductive firms will have difficulty in sustaining high levels of aggregate productivity growth. It is probably also intimately related to higher levels of income inequality between individual workers.4 And the small number of extant large productive firms in such an economy will be unable to absorb the entirety of the rest of the economy, either on the input side or the market share side.
III. THE BASIC SITUATION OF ANTITRUST LAW
The claim that some kind of decline in U.S. antitrust enforcement is the main cause of the rise in U.S. industrial concentration has become extremely, despite no direct evidence in its favor empirically connecting antitrust enforcement to industrial concentration. Antitrust law is passed off as both the main explanation of the economic developments of the past four decades, and also as the chief repository of policy solutions for the future. These claims risk dramatically over-representing what U.S. antitrust law actually is and how it works. Some very brief words of clarification here.
The basic architecture of U.S. antitrust law, while familiar to practitioners, is continually misunderstood and misrepresented by policymakers and even economists and others without a legal background. Antitrust law is not the functional equivalent of a generic tax on “bigness,” nor is it directly targeted at the variables that most formal economic models use to define “competition,” like markups, elasticities of demand, etc. As chiefly enforced by two separate federal agencies – the Department of Justice, and the Federal Trade Commission – it is instead the regulation of very specific actions, like: certain horizontal mergers (especially in highly concentrated markets), price fixing, illegal “tying” of unrelated goods and services, etc. Actions a regulator has authority over must be those clearly specified in extant legislation (e.g. the Sherman Act, the Clayton Act, the Robinson-Patman Act.).
When one of these agencies determines that there has been an infraction of one of these rules, they file a lawsuit, asking a court to compel the firm to stop the infraction, and possibly also pay some kind of fine.5 Antitrust suits can also be pursued by the Attorney Generals of individual U.S. States, or by affected private parties, or by some combination of the above. The legally mandated breaking up of a massive U.S. corporation simply because it has become “too big” is extraordinary rare – perhaps once every several decades – and still needs to be ostensibly tied to some more specific infraction (the most famous and most recent instance of this is the mandated divestiture of AT&T in 1982, and even then this was tied to a very specific claim about the illegal cross-subsidization of a regulated line of business by an unregulated line of business within AT&T – and the divestiture itself was actually proposed by AT&T during the settlement process!).
In general, the most direct and important relationship between antitrust enforcement and levels of industrial concentration centers around the regulation of horizontal M&A. Regulators might prevent a particular merger if they feel like the resulting company will have too much market share. But in principle there are no legal grounds to stop a highly productive company from organically growing as large as it can, provided it avoids specific infractions like those described above (price fixing, illegal bundling, etc.). Moreover, legal remedies for antitrust violations are specifically targeted at correcting past violations, not at hindering the company’s future growth (outside of potential damages or behavioral remedies introduced to specifically discourage future antitrust infractions).
This implies that for the:
Declining Antitrust Enforcement ⇒ Rising Concentration
narrative to work, it almost certainly has to take the form of:
Declining Antitrust Enforcement ⇒ More Horizontal M&A ⇒ Rising Concentration
And even if the rise in industrial concentration had taken place mostly through large firms increasing market share by acquiring other firms, this would still not be proof of the antitrust-rising concentration theory, because merger waves are endogenous to many broader economic forces, and a rise in merger activity could simply follow from the rise of economies of scale. But there is an asymmetry here: if the rise in industrial concentration happens through internal, “organic” growth and not through M&A, it seriously discredits the main channel through which the declining antitrust-rising concentration story would have to run through. And in fact, this is what the microdata appears to show: that most of the rise in market share from the largest firms happened organically, not through M&A.6
To be clear, there is heterogeneity across different industries on this dimension, and there are specific industries where a high level of recent merger activity has been plausibly linked to higher industrial concentration and adverse economic outcomes (for instance, Hospitals, or Health Insurance).7 But when the U.S. economy is taken as a whole and every industry is included, it seems implausible that the rise in industrial concentration is a story primarily about M&A, instead of a story about a rise in organic growth from the largest businesses.
The “declining antitrust – rising concentration” story is further disfavored by a comparative perspective. U.S. antitrust law has probably become modestly less stringent on a number of dimensions over the past forty years (in large part because of the “law and economics” revolution in the 1970s: Robert Bork and others forcefully argued that antitrust law should be designed to maximize consumer welfare, instead of e.g. the welfare of small business owners). But if U.S. antitrust law has moved in one direction over the past several decades, antitrust law in virtually every other major economy has moved in the other direction, towards a position of greater scrutiny and tougher sanctions (as in the EU arguably is today8). And yet, it appears that the rise in industrial concentration is not a purely U.S. phenomenon.9
The vanguard of antitrust activists today are especially focused on a small handful of prominent high-tech firms (like Amazon or Facebook), which function as networks/platforms/two-sided markets, and pose a distinct set of regulatory issues. It may (or may not) be the case that conventional antitrust law now needs to be creatively modified to accommodate these firms (see e.g. the work Lina Khan, Glen Weyl, Jean Tirole), and that after such a modification, the expansion of such firms might slow down. But it is important to remember that such firms form a small subset of the economy (most estimates of the “High Tech” sector, broadly defined and including firms like Amazon, Google, and Facebook, put it at about 5 percent of GDP, and much less as a percentage of employment or of the number of total firms), and that the rise of industrial concentration is a far more general phenomenon, that stretches across industries from architecture to retail grocery stores. As intellectually curious a topic as platform regulation may be, it does not appear to directly bear on the future of industrial concentration in most industries or most firms.
Consider a thought experiment. Suppose a zealous regulator was allowed to take antirust enforcement to the hilt: all future M&A was to be outlawed, perhaps even all past M&A unbundled, all previously acquired companies spun off, every other kind of antitrust infraction ruthlessly punished. How much of the basic structural problem of supply-side inequality in the U.S. economy would be solved? My claim here is: not much. Almost all the deep inequalities between the largest, most productive, most technologically sophisticated firms, and the rest of the economy, would remain, and the former would continue to pull away from the latter.
IV. THE RECURRENT DEBATE
The debate over antitrust takes a historically recurrent form. One side takes a benign attitude towards high levels of industrial concentration, interpreting it as the consequence of economies of scale, good for overall productivity and good for consumers. The other side takes a cynical view, interpretating it as the consequence of lax enforcement, allowing unproductive big businesses to exploit smaller businesses, and harm consumers.
This is the basic tenor of much of the debate today, but it was also the case in the 1970s (the previous highpoint of public interest in antitrust) – see e.g. John McGee’s 1971 “In Defense of Industrial Concentration” on one side, or something like E.F. Schumacher’s 1973 “Small Is Beautiful” on the Other. And while antitrust law in the U.S. is usually understood as beginning with the Sherman Act in 1890, the same underlying argument about the relative value of big business goes back to the founding of the republic itself, and the debates between Hamilton and Jefferson over whether it would be better for America to be a nation of large national businesses or small yeoman farmers.
In modern American history, this debate also has a clear, if sometimes awkward or implicit, political valence. American conservatives typically defend the prerogatives of big business and the value of economics of scale; progressives are more likely to dismiss them. (Still perhaps a reasonable characterization of the situation today, give or take the tense relationship between the U.S. far-right and Silicon Valley).
At a high level, the terms of the debate always seem to be either (1) you accept high industrial concentration as a good thing, or (2) you interpret it as a bad thing, and you respond by attacking big business. But there is a third position available, one that may be both more realistic while also being more radical: perhaps high industrial concentration is a bad thing, but a symptom of a deep inequality in production technology between firms and requires a new kind of industrial policy to overcome.
V. THE ALTERNATIVE WAY FORWARD
Consider the American economy as it existed for much of the 19th century: largely agricultural. In the South, large-scale plantations. But in the North, an abundance of small family farms. These small-scale farms were nevertheless capable of high levels of productivity. This productivity in turn was directly supported by a creative set of national government policies. “Land grant universities” – established in the 1862 Morrill Act – were created by the federal government to research agricultural technologies and then disseminate this knowledge to farmers (examples of land grant universities still around today: MIT, Cornell). Small farmers were able to sometimes pool resources together in distinctive legal arrangements of property and contract (e.g. sharing grain silos or tractors), enabling them to replicate economies of scale while still being small. The government gave would-be small farmers free land from the public domain, through long-standing “homesteading” programs.
The basic alternative in trying to overcome high industrial concentration today is to ask the question: what would the 21st century version of this 19th century industrial policy look like? What policies could we experiment with that might enable small and medium businesses to better leverage investments in new production technologies, and remain competitive? What would a 21st century version of a land-grant university look like? We can invent new legal forms of property and contract that could help small businesses pool data, or pool hardware? Are there new forms of contractual collaboration between large businesses and the smaller firms they might otherwise try to acquire, that might better support a more decentralized and innovative economy?10 Is there a 21st century equivalent of a homesteading program for small businesses? How could Silicon Valley specifically be encouraged to develop new production technologies with smaller businesses in mind?
And while the answers here are not obvious, today we are not even asking these questions. Instead, the focus of those concerned about high levels of industrial concentration has been almost exclusively on a narrow antitrust agenda. This focus is unpromising. A policy program that instead attempted to help these smaller firms become more productive would be more likely to succeed in solving the fundamental problem. It would then prove far more radical in its effect, while paradoxically also being both far more legally and politically palatable.
1 Economics PhD student at Harvard University. For helpful comments and discussions, I thank Louis Kaplow, Larry Katz, and Habin Chung. The views expressed here are mine alone, as are any errors.
2 Autor et al, “The Fall of the Labor Share and the Rise of Superstar Firms,” QJE, 2020.
3 Ganong Shoag, “Why Has Regional Income Convergence in the US Declined?,” JUE, 2017.
4 See “Firming Up Inequality,” Song et al, QJE (2019).
5 In the case of the FTC, if it is not seeking an injunction, a challenge may also take the form of an administrative proceeding before the FTC, instead of litigation before a court.
6 See e.g. “The Micro-Level Anatomy of the Labor Share Decline,” Kehrig Vincent, QJE, 2021.
7 See e.g. “The Price Ain’t Right? Hospital Prices and Health Spending on the Privately Insured,” Cooper et al, QJE, 2019.
8 While the EU and U.S. antitrust enforcement regimes share many doctrinal similarities, there are also several institutional and doctrinal differences, including the ability of many EU antitrust bodies to enforce actions like blocking mergers without court approval. Action can be taken by either the EU antitrust body, DG Comp, or by a national agency within an EU country. And recent doctrinal developments in the EU, like the theory of “conglomerate mergers,” make merger blocking even easier. These differences combine to create a climate of much greater antitrust scrutiny than that of the U.S. (and this still does not take into account more nebulous but relevant political and cultural differences, which perhaps also contribute to the EU’s more aggressive antitrust stance).
9 See Autor et al (2021), which documents similar patterns in the evolution of industrial concentration throughout the OECD economies.
10 See “Contracting for Innovation: Vertical Disintegration and Interfirm Collaboration,” Gilson Sabel Scott, Columbia Law Review (2009), which details new forms of contractual relations between firms in the pharma/biotech industry that take the place of M&A, allowing firms to achieve the advantages of scale while maintain the advantages of decentralization.
