The firm is the company as a whole.

The plant is one of the company's facilities, normally one of its manufacturing facilities. The minimum efficient plant size is the minimum size of company required for the lowest cost production. [Diagram goes here - download the original pdf to see it.] Below the minimum efficient scale there are diseconomies of scale owing to the small size of the firm. In this zone, as the scale of output rises the company becomes more efficient and experiences increasing economies of scale. Above the minimum efficient scale there are also diseconomies of scale. As the size of the firm increases, the administration becomes less efficient, and the average costs rise. Therefore, there is pressure on any company to growth until it has reached the size of minimum efficient scale. Thereafter, a company may still growth through merger, acquisition or takeover but this will be dictated by corporate objectives as much as the need to achieve minimum efficient scale. It is commonly observed that in producing and distributing almost every economic good there is some systematic relationship between the size or scale of the plant and the production cost per unit of output, and a similar relationship between the scale of the firm and the unit cost of producing and distributing the good. Let scale be measured as the rate of output per unit of time which a plant or firm is best designed to produce (can produce more economically than any other size of plant or firm). Let us also suppose that any plant or firm is operated at its best rate of output. Producing at that output rate, a plant or firm will have a certain cost per unit of output for production, and a firm a certain unit cost for production plus distribution. But these unit costs will not necessarily be invariant to the scale of the plant or firm. (For formal analysis of the relationship of the scale to the rate of utilization of the plant or firm and for definition of the scale curve, or long-run average cost curve, see Viner 1931 and Stigler 1946, pp. 128142.) As the scale of a plant is enlarged from the smallest feasible size through progressively larger sizes, increases in the scale of plant will generally result in lower production costs per unit of output until some critical scale is reached beyond which further increases in scale will leave unit production costs unchanged. This critical plant scale may be designated as the minimum optimal scale of plant in any given industry. Similarly, as the scale of a firm operating a plant or plants in an industry is progressively increased to and beyond the size required for the operation of one plant of minimum optimal scale, the firm may realize reductions in the cost of producing and distributing a good, possibly until it is large enough to operate two or more such plants, although again it apparently tends to reach a critical scale beyond which further scale increases will not result in lower unit costs of production and distribution. Plant in this context means a factory, mill, or other assemblage of connected productive facilities located on a single site (frequently also called an establishment), and firm means an independent administrative and control unit which manages a plant or plants and distributes their outputs. The reductions of costs with increases in the scales of plants or firms reflect, in general, economies of scale; in particular, economies of large-scale plant and economies of large-scale firm (in essence economies of the multiplant firm). The economies in question are in general internal economies (economies internal to the plant or firm in the sense that they are attributable to increased size of the individual unit), as distinguished from external economies, which are reductions in unit costs that may occur as the result of increases in the size of an industry, accomplished by an increase in the number of firms. Realization of scale economies results in the decline of unit costs of production in a plant until a minimum optimal (lowest-unit-cost) scale is reached; thereafter, further increases in scale will not tend either to reduce or to increase unit production costs, and there will be an indefinitely wide range of optimal plant scales that are larger than the minimum optimal scale. The same may also be true of firms in that, once they have attained minimum optimal scales, further growth will neither increase nor decrease their unit costs. But there is the possibility that the growth of firms to very large size will result in an increase in their unit coststhat there are diseconomies of very large scale firms which place an upper limit on firm scales that are optimal. The actual existence of such diseconomies is both debatable and debated. Economies of scale may be either real economies, reflecting a reduction in the physical quantities of productive factors needed to produce a unit of output (and a corresponding reduction in money costs), or strictly pecuniary economies, reflecting only a reduction in the prices at which the firm acquires productive factors, no real cost savings being involved. Economies of large-scale plant typically have a technological basis and are real

economies. They are generally attributed to real savings in production costs resulting from specialization of labor, specialization of capital equipment, and specialization of management functions that large plant scales permit. Economies of the large or multiplant firm, so far as they are realized, may be real economies of specialization of management, or real savings in shipping costs through the operation of geographically dispersed plants where nationwide distribution is for some reason advantageous to the firm. They may also be strictly pecuniary economies resulting from lower prices paid for productive factors as a result of the monopsonistic power of large-volume buyers, or from access to financing on preferential terms. The realization of strictly pecuniary economies is of no advantage to society except insofar as passing them on to customers of large firms through lower output prices may have desired effects on the distribution of income. (An extensive analysis of the bases of scale economies is found in Robinson 1931.) Real economies of large-scale plant are due fundamentally to indivisibilities or lumpiness in the particular forms of specialized factors or agencies of production, which can be acquired or used only in certain finite sizes and in some cases are more efficient in larger than in smaller sizes. With infinite divisibility of every productive factor and agent, there would be no plant scale-economies. This fact explains why progressive expansion of a plant will lead to a minimum optimal scale beyond which the plant will be neither more nor less efficient. At some critical finite scale, the indivisibilities of specialized factors or agencies which favor large plant-size will be fully overcome (the advantages of specialization will be exhausted), and further increases in scale can do no better than duplicate plants of minimum optimal scale on a single site. (For more extensive analysis of this and related matters, see Bain 1952, pp. 110117. For analysis of the reflection of scale economies in production functions, see Boulding [1941] 1955, chapter 34.) The same phenomenon of exhaustion of scale economies at some finite scale is encountered with any real economies of the multiplant firm, and probably with its strict pecuniary economies as well. If diseconomies of very large firms are encountered (and the evidence bearing on this issue is inconclusive), it is because of the inherent inefficiency of very large administrative organizations, afflicted with inflexible regulations, red tape, and extensive internal communication problems. Some theorists have argued that management is essentially a fixed factor which cannot be expanded in size beyond certain limits without assuming an altered and less efficient form. Quantitative importance. What has been said so far refers entirely to the qualitative character of the relationship of the scale of the plant or firm to the unit cost of production and distribution of a good, and to the theoretical explanation of such a relationship. The important economic issues turn upon the question of how quantitatively important these scale economies (and diseconomies) are in various industries. This is because the extent of their importance in any industry determines the minimal degree of plant concentration and minimal degree of concentration of industry control by firms which are consistent with plants and firms of reasonably efficient scale; it also determines (if diseconomies of very large firms are involved) the maximal degree of concentration by firms which is consonant with efficiency. Specifically, therefore, we wish to know for each of a large group of industries and for types of industries similar in the relevant respects: (1) the approximate percentage of industry output which would be supplied by one plant of minimum optimal scale; (2) the approximate percentage of industry output which would be supplied by one firm of minimum optimal scale; (3) the extent to which production costs per unit of output of plants of progressively less than minimum optimal scale exceed lowest attainable costs (i.e., the slope of the plant scale curve relating unit costs to scale, over a range of suboptimal plant scales); (4) the extent to which unit production costs plus distribution costs of firms of progressively less than minimum optimal scale exceed lowest attainable costs (i.e., the slope of the firm scale curve); and (5) whether or not (and if so, to what extent) diseconomies of very large firms are encountered in practice. This information should permit us to determine for any industry, or for groups of similar industries, how large a number of reasonably efficient plants and how large a number of reasonably efficient firms the industry can support (as well as how small a number of firms, should diseconomies of very large firms be significant). And this determination should in turn indicate the extent to which the pursuit of efficiency in scale of plants and firms is consistent with degrees of concentration of industry control by firms which are respectively compatible with competition, oligopoly, and monopoly. It should also allow us to appraise existing degrees of concentration of control in individual industries, and particularly the

development of large multiplant firms, with an eye to finding to what degree they are justified by the pursuit of scale economies and conversely to what degree concentration is greater, and market structures less competitive, than considerations of economy actually require. There is not as yet available an adequate body of evidence on the quantitative importance of scale economies in large numbers of industries in all of the several sectors (e.g., agriculture, manufacturing, wholesale and retail trade, the service trades, construction, and so on) of any national economy. Bain has, nevertheless, developed data (1956b) on the importance of scale economies in a sample of 20 manufacturing industries in the United States as of the early 1950s, the majority of them having moderate to high concentration of control by relatively few firms. The findings for this sample may be viewed as tentatively and roughly indicative of the importance of economies of large plants and firms in American manufacturing industries generally. As for economies of large-scale plant, it appears that the proportion either of national industry capacity (where the industry has a unified national market) or of capacity supplying the largest regional submarket (where the industry is fragmented into regional submarkets) that would be provided by one plant of minimum optimal scale is from 15 to 25 per cent in two out of 20 cases, from 10 to 15 per cent in four cases, from 5 to 71/2 per cent in six cases, from 2 to 5 per cent in three cases, and below 2 per cent in five cases. In a number of industries in which the proportion of the market supplied by one plant of minimal optimal scale is appreciable, however, the elevation of unit costs at appreciably smaller plant scales is relatively slight, so that plants with substantially less than minimal optimal scale would be reasonably efficient; in only about 40 per cent of the industries do plants with scales from one-half to one-fourth the minimum optimal scale experience significantly higher production costs. Taking into account both the ratio of the output of a plant of minimum optimal scale to the size of the market and the relative flatness of many plant scale curves at suboptimal scales, we draw the following tentative conclusions about the 20 manufacturing industries. (1) In two, plant-scale economies are very important, in the sense that the output of a plant of minimum optimal scale exceeds 10 per cent of the designated market output and that unit costs would be at least moderately higher at half-optimal scale. (2) In five, plant-scale economies are moderately important in that the output of a plant of minimum optimal scale is 4 or 5 per cent of the designated market output and that unit costs would be moderately higher at half-optimal scale. (3) In nine, plant-scale economies are relatively unimportant, either because a plant of minimum optimal scale would supply a small percentage of the output in the designated market or because plants of suboptimal scale experience only slightly higher costs, or for both reasons. (4) In four, information does not permit definite classification, but important plant-scale economies may be present in two of the four cases. So far as we may generalize from these findings (1956b, pp. 7182), the importance of plant-scale economies is widely variable among industries generally and among industries with high concentration of control by a few firms. In a distinct minority of cases, plant-scale economies are sufficiently important to justify a moderately high degree of plant and firm concentration. In the substantial majority of cases, plant-scale economies are such that reasonably efficient production is consistent with the existence of a substantial number of individual plants serving an unsegmented national or principal regional submarket; high concentration of control by firms is not required to take advantage of economies of largescale plants. For the same sample of industries, economies of the large-scale or multiplant firm appeared to be relatively unimportant. In half of 12 industries for which pertinent data can be found, there appear to be no economies of multiplant firm whatever. In the other half, from slight to modest economies (involving a cost advantage of from 2 to 4 per cent) are found for firms controlling several optimal-sized plants. These economies are attributed in some cases to large-scale management and in others to reduced costs of distribution, typically through the use of regional plants to supply regional markets (1956b, pp. 8393). In the latter case, the economies are generally conditional on nationwide sales promotion of goods with appreciable transportation costs, and although they represent savings to the

firms involved, they are not savings to the economy, which could be as efficiently supplied by regional firms not engaging in this type of sales promotion. (Economies of large-scale distribution are discussed at greater length in Bain 1956a, pp. 343345.) With a few exceptions, realization of reasonably efficient production and distribution by manufacturing industries does not really require multiplant firms; efficiency requires firms with outputs no larger than one plant of minimum optimal scale. In the few exceptional cases, the efficient multiplant firm would ordinarily not need to have more than plant of minimum optimal scale in any regional submarket. Realization of economies of the multiplant firm, therefore, would not require higher concentration of control by firms in such submarkets than would realization of economies of large-scale plants aloneas seems generally to be the rule. Economies of the multi-plant firm appear to have been greatly overrated by various observers making casual judgments. Our investigation in general reveals no firm or systematic evidence that diseconomies of large scale tend necessarily to be encountered by very large firms in manufacturing industries. Some instances are found in which the largest firms in an industry have higher unit costs than somewhat smaller firms, but their cost disadvantages appear to be attributable to a variety of special circumstances not intrinsically linked with scale of firm per se. Diseconomies of large-scale firm thus do not tend to provide an effective check on degrees of concentration of industry control by firms that are much higher than required for efficiency in production and distribution. Indeed, the incidence of dominant firms which are from four or five to twenty or more times as large as is required for exploiting all visible economies of large-scale production and distribution is quite high in the 20 American manufacturing industries sampled. One reason that economies of large scale are not such as to justify high degrees of concentration of industry control by firms in the United States is that the national and major regional submarkets in this country are very largemost usually large enough to absorb the outputs of numerous plants and firms of reasonably efficient scale. In other industrialized countries outside the communist bloc, national and regional markets range from moderately smaller to much smaller, but in most such industries production techniques and methods do not differ enough from those of the United States that the scales of reasonably efficient plants are appreciably different. In these other countries, therefore, realization of efficiency in scale of plants and firms typically would require from moderately to substantially greater degrees of plant and firm concentration than is required in the United States. But actual plant concentration in their industries most frequently is not high enough for reasonable efficiency, although concentration of industry control by firms may be. (Findings on this matter are presented in Bain 1966, chapters 3 5.) As to attained efficiency in the United States, the findings based on the sample of 20 manufac-turing industries are that between 70 and 90 per cent of the output of a manufacturing industry is typically supplied by plants and firms of reasonably efficient scale, whereas the remaining 10 to 30 per cent is supplied by plants (and typically oneplant firms) that suffer from appreciable diseconomies of unduly small scale. Such an inefficient fringe of small plants appears generally to supply a larger proportion of industrial output in most other industrialized countries outside the communist bloc (Bain 1966, pp. 5566). Large-scale sales promotion. This discussion has not dwelled on possible economies of large-scale promotion, in part because the proper term would be advantages rather than economiescost additions rather than savings generally being associated with sales promotion. The question nevertheless remains as to whether there are systematic advantages to firms in large-scale sales promotion programs and expenditures. The evidence is largely negative. Although it is true that, in industries selling differentiated products, large established firms are frequently able to maintain large market shares with smaller sales-promotion costs per unit of sales than those incurred by smaller firms to maintain smaller shares of the market, it is not generally true that the smaller firms can match the advantaged position of the larger ones by simply increasing their salespromotion outlays. In effect, there is no unique functional relationship between size of promotional outlays and volume of sales, exploitable by all firms, comparable to the unique relationship between scale of plant and production costs. The term economies of largescale sales promotion is thus inherently misleading (Bain 1956a, pp. 339343). The foregoing discussion has emphasized the fact that although appreciable economies of large-scale plant are encountered at least in manufacturing, as well as some slight

economies of multiplant firms, they are not important enough quantitatively to justify high degrees of plant and firm concentration in the great majority of manufacturing industries in the United States. It is still true, however, that plants of absolutely very large size are required for efficiency in many lines of manufacturing and that substantially smaller plants suffer diseconomies of insufficient scale. Various findings which have been made to the effect that the smallest plants and one-plant firms in most manufacturing industries are, or can be, as efficient as or more efficient than plants and firms of larger sizes generally rest on statistical misinterpretation of defective cost data and of irrelevant profit data (see, for example, U.S. Federal Trade Commission 1941, pp. 1292; Kaplan 1948, chapter 5) and can at present be regarded as thoroughly invalidated. Inefficiencies of unduly small-scale plants indeed exist, and, as noted above, are encountered in an inefficient fringe of firms in most manufacturing industries. Much has of necessity been left unsaid about scale economies, particularly concerning their importance in industries outside the manufacturing sector. Scattered evidence suggests that economies of large-scale plant are exceedingly important in most publicutility industries and have some importance in construction industries. Significant economies of the large-scale firm are definitely encountered in some of the distributive trades, including especially the food-distribution industry. We lack systematic data, however, to support a general appraisal of the importance of scale economies in these sectors.