CHAPTER 2

Most investigations of issues raised by industrial concentration begin with a study of its effects. There is, however, some advantage of logic in beginning with its causes.

A salient fact, agreed upon by both structuralist economists and their adversaries, is that about the same industries are concentrated in about the same degree in all modern industrial economies. From the study by Joe S. Bain, International Differences in Industrial Structure (New Haven: Yale, 1966) to F. L. Pryor's "An International Comparison of Concentration Ratios," 54 Review of Economics and Statistics 130 (May, 1972), the fact that concentration ratios and the rank order of concentration are roughly the same in all major industrial countries is common ground.

This international similarity of industrial structure leaves little room for the essentially local explanations, drawn from the field of antitrust, such as merger movements, individual genius, or predatory behavior by individual

firms.

In the search for more promising explanations, two recent investigations are worth noting. Professor F. M. Scherer found a strong correlation between minimum efficient plant size and concentration, in "The Determinants of Industrial Plant Sizes in Six Nations," 55 Review of Economics and Statistics 135 (May, 1973). This finding inferentially supports the popular impression that scale economies are a major factor in determining the degree of concentration in any given industry.

A second popular impression-that high technology is closely associated with high concentration-receives support from a study by Professor Raymond Vernon, "Organization as a Factor in the Growth of Firms," Chapter 4 in Industrial Organization and Economic Development, Jesse W. Markham and G. F. Papenek, editors (Boston: Houghton Mifflin, 1970). He found a strong rank order correlation among the percentage of an industry's employees who were scientists and engineers, its exports, value added per leading firm, and concentration ratio.

Large scale industrial operation, with its resultant concentration, makes a major contribution to economic performance. Professor J. Fred Weston published an estimate, tabulated on this page, in The Impact of Large Firms on the U. S. Economy.1

Professor Weston summarized the effect of the estimate as follows:

"Combining low positive with the negative estimates, the range is 3.4 percent to 4.3 percent. Subtracting the negative figures against the high positive estimate, the range of contributions of concentration is 6.5 percent to 7.4 percent. Translated into dollars for our present trillion dollar economy, the most probable range of net contributions of concentration to economic performance is from forty-three billion to sixtyfive billion dollars. These are minimal estimates of the direct benefits of large scale operations and industry concentration in the United States; they also indicate the order of magni

tude of the direct costs of breaking up large, efficient firms in the effort to atomize the U.S. economy."

Professor Weston's estimate is unique in that it seeks to balance gains and losses from large scale operations. There have been, however, some widely publicized estimates focussed solely on alleged losses from monopoly power. At the risk of digression from the present subject of causes of concentration, it seems useful to consider whether or not there is anything to the claim that the economic benefits of large scale organization are offset by economic wastes of "monopoly power."

Much of the early interest in "industrial organization" rested on the populist article of faith that "monopoly power" was a costly source of inefficiency in the American economy. With the rise of interest in quantitative economics, somebody was bound to try to estimate the inefficiency loss; and the result, published by Arnold C. Harberger in 1954,3 was an unwelcome surprise: The losses came out as negligible-of the order of .07 of 1% of output. Subsequent work-reviewed for example by Harvey Leibenstein in "Allocative Efficiency vs. 'X-Efficiency'", led to the conclusion:

"The empirical evidence, while far from exhaustive, certainly suggests that the welfare gains that can be achieved by increasing only allocative efficency are usually exceedingly small, at least in capitalist economies. In all but one of the cases considered all of the gains are likely to be made up in one month's growth. They hardly seem worth worrying about." P. 395.

The reaction to the findings summarized by Leibenstein has been twofold. First, the structuralists have synthesized more impressive numbers. Second, they have become experts in what Leibenstein called X-efficiency-i.e., efficiency in the internal management of business firmsand issued estimates of how costs could be lowered were competitive pressures greater. The names of F.M. Scherer, H. Michael Mann, and W. G. Shepherd are associated with some of the more spacious estimates. For reasons of space we limit ourselves to Scherer's, both because it has had the honor of Senatorial citation and because it is presented in a widely used textbook of respected authorship.

The Scherer estimate occurs in his Industrial Market Structure and Economic Performance (Chicago: Rand McNally, 1971) pp. 400-409. He reached an estimate of "social losses," as of 1966, equal to 6.2% of GNP, which on a trillion dollar GNP calculates to the oft-cited $60 billion number. His table at page 408 breaks this 6.2% into eight parts, briefly summarized as follows:

Of this 6.2%, 3.3%-more than half-is made up of items that are irrelevant to S. 1167. For purposes of moves against unregulated oligopoly, 0.6% of the 6.2% drops out as irrelevantly relating to the regulated sectors. Another 0.6% is due to deficient cost control by space and defense contractors-again the business of Government departments with ample powers, other than Justice or FTC. "Wasteful promotional efforts" accounts for another 1.0%; but this one again fails the relevance tests: What would any of the structural remedies so far proposed, such as S. 1167, do to convince advertisers that they ought to be more informative (and less effective), especially if competition is intensified? And are outsiders better able to judge waste than the businessman who is spending the advertising money and the consumer who responds to it?5

Another 0.6% item in Scherer's study is, "Excess and inefficient capacity due to industrial cartelization and the stimulus of collusive profits." Back at page 407, one finds that this number is compounded of oil proration and a record of low capacity utilization in the cement industry.

Neither the petroleum nor the cement industry is highly concentrated, at least in terms of national markets, and oil proration is government action. So, for purposes of S. 1167, there goes another 0.6%.

Next the old antitrust issues of basing point pricing and cross-hauling are down for 0.2%-a $2 billion estimate on our trillion dollar economy. The estimate is not especially believable, not obviously avoidable except by development of regional monopolies, and quite as characteristic of industries outside the reach of dismemberment legislation as of industries within its reach. In fact, without multiple assembly plants under single firm ownership, cross-hauling costs might be expected to rise rather than fall.

Then we have 0.3% (or $3 billion) for "Operation at less than optimal scale." The "less than" is intentional; the objection is to the preservation of inefficiently small business under the umbrella of oligopoly pricing, fair trade laws, and the like. Scherer does not use the large companies with even larger competitors for his examples; what he cites are: inefficiently small drug stores, liquor outlets, and gasoline stations. ." We suspect that the public does not appreciate that those who cite a $60 billion loss have in mind the elimination of large sectors of small business which are surviving the market test but are thought "inefficient" by the structuralists.

So much for the items not relevant to S. 1167 or similar deconcentration proposals.

Scherer included 2.0% for "Inefficiencies due to deficient cost control by market sectors insulated from competition." This number is Leibenstein's X-efficiency item. Scherer includes in his 2% "discretionary costs which could be cut under the impact of adversity with no impact on production." This test would be met by expenses not usually regarded as inefficiencies, such as outlays for R&D, quality control, plant modernization, after-sales service, etc. etc. He ends by characterizing his own estimate as "guesswork." Also, even if Scherer is successful in finding the havens from competition which are so elusive to others, there is an old business axiom that you make money from the price at which you buy, not from the price at which you sell; and this leaves the firm in possession of supposed "market power" already highly motivated to cut costs.

This brings us to Scherer's first item-0.9% for "Deadweight welfare losses due to monopolistic resource misallocation: unregulated sectors." The first thing that arouses one's suspicions is the fact that Harberger and Schwartzman, who did the primary research Scherer relies on, arrived at a figure between 0.06% and 0.1%-in other word, Scherer has multiplied his sources by a factor of 9 to 15. Suspicion of this multiplication process is accentuated by Leibenstein's observation to the effect that the low level of allocative efficiency losses is inherent in the geometry of the concept. 56 American Economic Review at 395-6.

How Scherer accomplished the 9-to-15 multiplication is summarized in a footnote on page 404.

Step 1 was to raise Harberger's .06% to 0.1%, "to take account of excessive aggregation and the use of too high a normal' profit return" The discussion of this point at page 402 is limited to statistical opinion; the evidence to get from here to there is missing.

Step 2 was to multiply 0.1% by 3 to extend it from the manufacturing economy, studied by Harberger, to the entire economy. The structure dispute has centered on the unregulated manufacturing sector. Much of the services sector is unconcentrated, and the areas of the services sector which might be thought concentrated include a number of regulated industries: communications, banking insurance. utilities. For others, such as newspapers and radio and TV stations, dismemberment would raise formidable First Amendment problems This Step 2 multiplication thus lacks the needed evidentiary basis for relatng it to deconcentration proposals or for assuming that Harberger's manufacturing sample is typical of the nonmanufacturing sector.

Step 3 was to multiply again by 2.5 "to reflect more realistic (ie, higher) demand elasticities." This step also lacks support. What "realistic demand elasticities" may be is a matter of dispute; and in any case "monopoly profits" are likely to be low in industries where demand is highly elastic. Schwartzman used double Harberger's demand elasticity and still came out with about the same result as Harberger." The Harberger approach in effect assumed that "monpoly profits" will center in industries with relatively low demand elasticity; Scherer's contrary assumption seems unsound.

Step 4 was a multiplier of 1.4 "to take into account vertical distortions." On this, one encounters a brick wall at page 403-4. "When price distortion estimation techniques similar to Harberger's are employed, taking vertical distortion effects into account actually reduces the final estimate of welfare losses. When a more sophisticated distortion measurement technique is used But this is the brick wall; the "more sophisticated technique" is neither explained nor referenced.

Of course, the Harberger technique is itself vulnerable, since he estimated "monopoly profit" in terms of aboveaverage profit, without distinguishing between profit variations attributable to "market power" and variations attributable to the risk factor, normal disequilibrium effects, or a pay-off for innovation. The latter variations could not plausibly be regarded as a basis for implying an overall welfare loss, and the reality is the Harberger work itself almost certainly overstates the "deadweight welfare" cost of "monopoly power."

The estimates of "losses" from "monopoly power" all rest on the assumption-often unstated-that concentrated industries would perform at equally low costs if they were not concentrated. Both common sense and the similarity of experience of other advanced industrial economies favor Professor Weston's view that large scale operations in fact benefit economic performance in cost/ price terms to a degree which submerges what Liebenstein called the "exceedingly small" effects of monopoly

power on allocative efficiency. Similarly, the oft-repeated concept of sectors of the economy "insulated from competition," resulting in "X-inefficiencies," almost invariably turns out to involve a gross perceptual gap between those whose lives are, and those whose lives are not, safely insulated from competition in the sectors in question.

In an industry where scale economies produce concentration, a prohibition of concentration operates as a prohibition of scale economies. More generally, anticoncentration legislation is in effect legislation against the circumstances which produce concentration, particularly where cause and effect are mutually indispensable. But the circumstances which produce concentration vary from industry to industry. A brief listing of the more common causes of concentration will help give some idea of the nature of the unspecified targets of deconcentration proposals:

(1) The simplest explanation of large firm size is "scale economies." These may occur in engineering, manufacturing or marketing, or in all three. The subject of scale economies and what constitutes a "substantial" level of scale economies-is discussed in Chapter 9. Some of the factors mentioned in the following paragraphs are also in the nature of scale economies.

(2) Firms in concentrated industries usually are responsible for their own innovation, in contrast to atomized industry, where innovation is usually sourced from equipment, materials, and parts suppliers and (in the case of agriculture) from government.

(3) As a matter of observation, product complexity favors firm size. In other words, firms with products made up of many individual parts (e.g., automobiles) are likely to be large firms; and firms which convert relatively simple raw materials into a complex range of products are likely to be large firms. The explanation is conjectural. One plausible possibility is that the degree of management organization required to handle the design, production, marketing and servicing of complex products produces a kind of scale economy of overhead. The minimum overhead is uneconomic unless the sales base is large. Another possibility is that product complexity offers extra opportunities and penalties for mistaken decision making-with the result that variations in decisional efficiency produce winners and losers.

(4) Again as a matter of observation, products with extremely high catalogue prices-say, from $500,000 to $1,000,000 and up-are almost always "big firm" products. Even in relatively atomized industries like construction, the firms that bid on the very big jobs are not numerous. And again the explanation is conjectural. Game theory suggests that the kind of bet involved in contracting to produce such a product can be risked only by large firms; and learning curve theory would imply that, given the small total output of such products in units, costs would be substantially reduced by concentrating production experience in a (very) few firms.

(5) High technology products-that is, in an economic sense, products with high engineering expense-are

usually "big firm" products, even when the basic invention may have been made by an individual or a small firm. Part of the reason is that if engineering expense is high, it can be reduced relative to sales by spreading it over more sales. If it costs $50,000,000 to engineer a new jet engine for which there is a market of perhaps 1000 engines, the engineering cost per engine will be $50,000 if one firm only is in the market, $100,000 if two firms design similar engines, and $150,000 if three firms do it. If multiple design were mandated by legislation, the point would soon be reached where it would pay the airlines to buy their engines from a country with a more rational approach to industrial organization. Another reason may be that high technology products are likely to be complex (see paragraph (3) above) and sometimes very expensive (see paragraph (4)).

(6) Management technology and developments in business organization can affect firm size. In 1910, it would not have been thought that the corner grocery could support the overhead of a large corporate organization. Chain stores had changed this picture by 1930. The invention of the franchise system works in the other direction-toward giving small enterprises some of the advantages of their large, highly organized competitors. Firm size has been influenced by such developments as decentralized management and the "free form" conglomerate-though with less effect on size in particular markets than on total firm size.

(7) About the same industries are concentrated in most of the major industrial countries-a fact which (as stated above) invites the inference that accidents of managerial skill or financial manipulation (mergers, particularly) do not have much ultimate effect on firm size. They may, however, have a good deal to do with the shakeout process in the early history of an industry, when the leaders, survivors, and dropouts are selected.

(8) A few industries present spectacular initial capital requirements. This is, in economics, an aspect of scale economies; but given the obvious initial cost of any reasonable steel mill, for example, it must be evident that steel is a "big firm" product, quite apart from scale economies in even larger plants. This does not mean that there are "barriers" which permit existing firms to earn monopoly profits; if profit potential exists large aggregations of capital can be formed. It does mean that the market will not support a large number of competitors. Also, again as a matter of observation, very few industries of this kind end up with single firms operating minimum size plants, either here or in other countries. Having once learned how to build and operate Plant No. 1, it pays for a firm to try its hand at Plant No. 2.

(9) One of the obligations of the managers of an enterprise is to make the fullest use of the resources committed to their care: and in some instances this may mean branching out from a parent business into other businesses. This will not necessarily produce a "big firm" in a particular industry, but it may produce a firm large because of diversification or because of the comprehen

sive coverage of its product line. As a small example, a cable manufacturer who develops a new plastic insulation may branch into the plastics business, on the theory that by so doing it will earn more for stockholders on the new insulation than it would earn by licensing it to existing plastic manufacturers. A technology developed for one product, in other words, can easily lead-or even force the developer into diversification. Probably the most common form of "diversification" is, however, that most closely related to the original business: adding models, sizes, and accessories.

(10) Advertising and concentration are statistically related, but the causal relationships (if any) are uncertain. The relationship may be a statistical illusion: In some small-firm industries, advertising may not be a practical business possibility; hence statistical averages are almost bound to show more advertising in big-firm than in smallfirm industries. A "pull-through" market strategy may be a viable alternative to "push-through" in more big firm industries than small firm industries-and if this is the fact, the probable conclusion would be that advertising has a scale economy effect. The size required to advertise is not, however, necessarily very large; the evidence on scale economies in advertising itself has not withstood critical examination."

(11) There are a number of definable markets where a high degree of concentration has resulted from a consumer consensus in favor of the product of a particular manufacturer. Buyers would testify almost unanimously that the product they prefer is superior to any of the alternatives. This kind of consensus is surprisingly frequent where buyers, as a group, are in close touch with each other-competitive sportsmen, hobbyists, sometimes corporate purchasing agents in a particular industry, sometimes professional men or businessmen who compare notes at annual conventions. Given the propensity of courts to gerrymander markets, replacement of the behavioral test of monopoly by a simplified structural test could easily cause obvious injustice in these markets.

(12) High risk is a factor which seems to work toward concentration, and also toward compensatorially high profit rates. It is a very difficult element to measure from current statistics, because the most relevant data will relate to firms which are no longer in the industry. Also, the important risk is that associated with the investment of fairly substantial amounts of capital: small retail establishments with very high failure rates continue to attract entrepreneurs, but the larger the required investment the more difficult it becomes to think of a business with a high past failure rate which will attract entry without exceptional profit rates and probably also a high growth rate. The eventual outcome of a high failure rate and a low entry rate is an above-average concentration ratio, perhaps coupled with a profit rate that partially reflects the risk of entry. Firms already in the industry are not faced with the risk of inability to develop a viable management apparatus, but other risks may still lead them to determine their own new investment by reference to a