By William F. Shughart II
Industrial concentration” refers to a structural characteristic of the business sector. It is the degree to which production in an industry—or in the economy as a whole—is dominated by a few large firms. Once assumed to be a symptom of “market failure,” concentration is, for the most part, seen nowadays as an indicator of superior economic performance. In the early 1970s, Yale Brozen, a key contributor to the new thinking, called the profession’s about-face on this issue “a revolution in economics.” Industrial concentration remains a matter of public policy concern even so.
The Measurement of Industrial Concentration
Industrial concentration was traditionally summarized by the concentration ratio, which simply adds the market shares of an industry’s four, eight, twenty, or fifty largest companies. In 1982, when new federal merger guidelines were issued, the Herfindahl-Hirschman Index (HHI) became the standard measure of industrial concentration. Suppose that an industry contains ten firms that individually account for 25, 15, 12, 10, 10, 8, 7, 5, 5, and 3 percent of total sales. The four-firm concentration ratio for this industry—the most widely used number—is 25 + 15 + 12 + 10 = 62, meaning that the top four firms account for 62 percent of the industry’s sales. The HHI, by contrast, is calculated by summing the squared market shares of all of the firms in the industry: 252 + 152 + 122 + 102 + 102 + 82 + 72 + 52 + 52 + 32 = 1,366. The HHI has two distinct advantages over the concentration ratio. It uses information about the relative sizes of all of an industry’s members, not just some arbitrary subset of the leading companies, and it weights the market shares of the largest enterprises more heavily.
In general, the fewer the firms and the more unequal the distribution of market shares among them, the larger the HHI. Two four-firm industries, one containing equalsized firms each accounting for 25 percent of total sales, the other with market shares of 97, 1, 1, and 1, have the same four-firm concentration ratio (100) but very different HHIs (2,500 versus 9,412). An industry controlled by a single firm has an HHI of 1002 = 10,000, while the HHI for an industry populated by a very large number of very small firms would approach the index’s theoretical minimum value of zero.
Concentration in the U.S. Economy
According to the U.S. Department of Justice’s merger guidelines, an industry is considered “concentrated” if the HHI exceeds 1,800; it is “unconcentrated” if the HHI is below 1,000. Since 1982, HHIs based on the value of shipments of the fifty largest companies have been calculated and reported in the manufacturing series of the Economic Census.1 Concentration levels exceeding 1,800 are rare. The exceptions include glass containers (HHI = 2,959.9 in 1997), motor vehicles (2,505.8), and breakfast cereals (2,445.9). Cigarette manufacturing also is highly concentrated, but its HHI is not reported owing to the small number of firms in that industry, the largest four of which accounted for 89 percent of shipments in 1997. At the other extreme, the HHI for machine shops was 1.9 the same year.
Whether an industry is concentrated hinges on how narrowly or broadly it is defined, both in terms of the product it produces and the extent of the geographic area it serves. The U.S. footwear manufacturing industry as a whole is very unconcentrated (HHI = 317 in 1997); the level of concentration among house slipper manufacturers is considerably higher, though (HHI = 2,053.4). Similarly, although the national ready-mix concrete industry is unconcentrated (HHI = 29.4), concentration in that industry undoubtedly is much higher in specific cities and towns that typically are served by only a handful of such firms.
These examples suggest that concentration varies substantially across U.S. industries. Trends in concentration vary from industry to industry, but most changes in concentration proceed at a glacial pace. So, too, does aggregate concentration: the fifty largest U.S. companies accounted for 24 percent of manufacturing value added (revenue minus the costs of fuel, power, and raw materials) in 1997, the same percentage as in 1992 (and as in 1954, for that matter). On some measures—the percentages of total employment and total assets controlled by the nation’s 50, 100, or 200 largest firms—industrial concentration in the United States actually has declined since World War II.
Concentration indexes calculated for a particular year conceal the identities of the industry’s members. In reality, turnover among the nation’s leading firms is fairly regular over long time horizons, averaging between 2 and 5 percent annually. Success at one point in time does not guarantee survival: only three of the ten largest U.S. companies in 1909 made the top one hundred list in 1987. Available concentration indexes, which are based solely on domestic manufacturing data, also ignore the global dimensions of industrial production.
The Causes and Consequences of Industrial Concentration
Some industries are more concentrated than others because of technical properties of their production technologies or unique characteristics of the markets they serve. Economies of scale, which allow firms to reduce their average costs as they increase their rates of output, favor large-scale production over small-scale production. Thus, industries for which scale economies are important (e.g., auto manufacturing and petroleum refining) are expected to be more concentrated than others in which costs do not fall as rapidly as output expands (e.g., cut-and-sew apparel manufacturing). Similarly, concentration tends to be higher in industries, such as aircraft and semiconductor manufacturing, where learning curves generate substantial production-cost savings as additional units of the original model or design are made.
Owing to so-called network effects, some goods increase in value as more people use them. Computer operating systems, word-processing software, and video recorder-players are examples of such goods, as are literal networks such as railroads, commercial air transportation, and wire line telephony. Because standard technologies and protocols that provide compatible interconnections are critical to the realization of network effects— allowing faxes to be sent and received or computer users easily to exchange files—consumers rationally favor large networks over small ones. The necessity of building networks that accommodate critical masses of users means that only a few providers will achieve dominant positions, and therefore the industry will tend to be highly concentrated. Such domination is likely to be temporary, however, since consumers will switch networks when benefits outweigh costs, as illustrated by the replacement of Betaformatted video tapes by VHS formatted ones, which in turn are being replaced by DVDs.
Industrial concentration also is promoted by barriers to entry, which make it difficult for new firms to displace established firms. Barriers to entry are erected by government-conferred privileges such as patents, copyrights and trademarks, exclusive franchises, and licensing requirements. Existing firms may possess other advantages over newcomers, including lower costs and brand loyalty, which make entry more difficult.
The fundamental public policy question posed by industrial concentration is this: Are concentrated industries somehow less competitive than unconcentrated ones? Concentration would have adverse effects if it bred market power—the ability to charge prices in excess of costs—thereby increasing industry profits at consumers’ expense. In theory, industrial concentration can facilitate the exercise of market power if the members of the industry agree to cooperate rather than compete, or if the industry’s dominant firm takes the lead in setting prices that rivals follow. And, indeed, the evidence generated by hundreds of econometric studies suggests that concentrated industries are more profitable than unconcentrated ones. But that evidence begs the question. It does not tell us whether profits are higher in concentrated industries because of market power effects or because the firms in those industries use resources more efficiently (i.e., have lower costs).
Some economists have found that concentration leads to higher prices, but the link observed typically is both small (prices elevated by 1–5 percent) and statistically weak. A detailed econometric study by Sam Peltzman (1977) reaches the opposite conclusion. He reports that profits are higher in concentrated industries not because prices are higher, but because they do not decline as much as costs do as efficient firms expand their scales of operation. Analyses by Yale Brozen (1982), Harold Demsetz (1974), and others have found that the positive relation between industrial concentration and profits disappears altogether when firm size is taken into account. These results are consistent with the hypothesis that some industries are more concentrated than others because large firms have significant cost advantages over small firms. There is, in short, little unequivocal evidence that industrial concentration per se is worrisome. Just the reverse seems to be true.
Public Policies Toward Industrial Concentration
Consolidating production in the hands of fewer firms through mergers and acquisitions obviously is the most direct route to industrial concentration. Preventing transactions that, by eliminating one or more competitors, would lead to undue increases in concentration and the possible exercise of market power by the remaining firms is the mandate of the two federal antitrust agencies—the U.S. Department of Justice and the Federal Trade Commission—under section 7 of the Clayton Act (1914). That mandate was strengthened considerably by the Hart-Scott-Rodino Act (1978), which requires firms to notify the antitrust authorities of their intention to merge and then to hold the transaction in abeyance until it has been reviewed. Most transactions with summed firm values of fifteen million dollars or more had to file premerger notifications initially; in February 2001 that threshold was raised to fifty million dollars and indexed for inflation.
Two important factors that antitrust authorities consider in deciding whether to allow a proposed merger to proceed are the level of market concentration if the merger is consummated and the change in market concentration from its premerger level. (Note that the “market” considered relevant for merger analysis hardly ever corresponds to the “industry” defined by the Economic Census; antitrust markets may be defined more broadly or more narrowly; in practice, the definition of the relevant market usually is the key to whether a merger is lawful or not.) Concentration thresholds are laid out in the Justice Department’s merger guidelines, first promulgated in 1968, revised substantially in 1982, and amended several times since.
The guidelines state that proposed mergers are unlikely to be challenged if the postmerger market is unconcentrated (HHI remains below 1,000). However, mergers generally will not be approved if, following consummation, market concentration falls within the 1,000–1,800 range, and the HHI increases by more than 100 points or, if the postmerger HHI is 1,800 or more, concentration increases by more than 50 points.2 Exceptions are provided when the merging firms can demonstrate significant cost savings, when barriers to entry are low, or when one of the merger’s partners would fail otherwise. (In the European Union, by contrast, competition policy, including merger law enforcement, is shaped principally by fears of possible “abuses of dominant market positions” by large firms.)
Studies examining the enforcement of section 7 under the merger guidelines have found that they are not always followed closely. Mergers are, indeed, more likely to be challenged the greater the level of market concentration and the higher the barriers to entry are thought to be. But law enforcement also is found to be influenced significantly by political pressures on the antitrust authorities from groups that stand to lose if a merger is approved, including rivals worried that the transaction will create a more effective competitor. In fact, studies of stock-market reactions to news that a merger is likely to be challenged typically find competitors to be the main beneficiaries of such decisions.
The Economic Census has been conducted every five years since 1967, and before that for 1954, 1958, and 1963. Prior to 1997, it was known as the Census of Manufactures. That same year, industries began being categorized according to the North American Industry Classification System (NAICS), which replaced the Standard Industrial Classification (SIC) codes used until 1992. Industrial concentration also is reported by the Economic Census on the basis of value added. Industry concentration ratios and HHIs for the 1992 and 1997 economic censuses can be accessed online at: http://www.census.gov/epcd/www/concentration.html. Information on industrial concentration is not readily available for sectors of the economy other than manufacturing.
When firms with market shares of s1 and s2 merge, the HHI increases by (s1 + s2)2 − s12 − s22 = 2s1s2. So, for example, if a merger is proposed between the two largest firms in the hypothetical ten-firm industry described earlier, the HHI would increase by 2 × 25 × 15 = 750 points (from 1,366 to 2,116). According to the guidelines, that merger would in all likelihood be challenged.