The Economics of Welfare

Part II, Chapter XIV

THE CONDITIONS OF MONOPOLISATION

§ 1. WE may now return to the main argument. In Chapter XI. we supposed self-interest to act along the route of simple competition. We showed that in these circumstances, apart from divergences of the types discussed in Chapter IX., the value of the marginal private net product of resources in any industry tends to equality with the central value of marginal net products in general, and inquired in what circumstances the value of the marginal social net product of resources in the given industry would diverge from the value of the marginal private net product there. We have now to consider other routes along which self-interest may act. As has already been observed, an essential note of “simple competition” is that the supply of each seller constitutes so small a part of the aggregate supply of the market that he is content to “accept market prices without trying, of set purpose, to modify them.”
*28 When any seller’s output constitutes a substantial part of the whole, there is scope for various sorts of monopolistic action; and, when any sort of monopolistic action is present, self-interest does not tend to evolve an output such that the value of the marginal private net product of resources devoted to its production is equal to that yielded by resources employed elsewhere. In future chapters I shall examine monopolistic action in detail. Before that is done, however, convenience suggests that some study should be made of the conditions which determine the appearance of monopolistic power.

§ 2. First, other things being equal, circumstances, which,
when the aggregate scale of an industry is given, make it structurally economical for the typical individual establishment to be large,
pro tanto increase the likelihood that a single
seller will market a considerable part of the aggregate output of his industry; for such circumstances necessarily increase the probability that a single
establishment will market a considerable part of that output. Whether any single establishment will, in fact, become big enough, relatively to the whole of an industry, to procure an element of monopolistic power, depends on the general characteristics of the various industries concerned. Such an event is more than usually likely in industries that produce fancy goods liable to become “specialities.” For in these industries there often exist, within the broad general market, minor markets, to a certain extent non-competitive among themselves; and, when this is so, a single establishment may supply a considerable proportion of its own minor market without itself being of very great size absolutely. In a few peculiar industries, among those concerned with staple goods and services, it may also well be that the prospect of internal economies will lead to the evolution of single establishments large enough to control a predominant part of the whole output of the industry. One of the most notable instances of this is afforded by the industry of railway transportation along any assigned route. In view of the great engineering cost of preparing a suitable way, it will, obviously, be much less expensive to have one or, at most, a few railways providing the whole of the transport service between any two assigned points than to have this service undertaken by a great number of railways, each performing an insignificant proportion of the whole service. Similar remarks hold good of the industries of furnishing water, gas, electricity, or tramway service to a town. The existence of many separate establishments involves a large number of main pipes, wires and rails. But the whole business of any ordinary district can be worked with a very small number of these mains. Therefore the existence of many separate establishments implies the investment of a great quantity of capital in mains that are only employed up to a very
small proportion of their capacity. There is an obvious economy in avoiding such investment. This economy is the
ultimate reason for the tendency, which appears strongly in the class of industry just discussed, for individual establishments to furnish a large proportion of the total supply. The truth is partly veiled by the fact that the
immediate reason is, in general, unwillingness, on the part of national and local government authorities, to allow the right of eminent domain to be invoked, or the streets to be disturbed, on more occasions, or by more people, than is absolutely necessary. It is, however, the extra expense of such procedure that lies behind this unwillingness on the part of the authorities. In the general body of industries concerned with staple goods and services the conditions peculiar to railways and their allied industries are not reproduced. Internal economies reach their limit at different points in different kinds of industry, at one point in the cotton industry, at another in the iron and steel industry; generally at a less advanced stage where the part played by labour relatively to capital is large and at a more advanced stage where it is small; but always long before the individual establishment has grown to any appreciable fraction of the whole industry of which it is a part.
*29 When this happens, internal economies evidently cannot be responsible for monopolistic power.

§ 3. Secondly, other things being equal, circumstances, which, when the aggregate scale of an industry and the size of the typical individual establishment are given, make it structurally economical for the typical individual unit of business management—a number of establishments, for example, controlled by one authority—to be large,
pro tanto increase the likelihood that a single seller will market a considerable part of the aggregate output of his industry. This proposition has, in recent times, become of predominant importance, and it is, therefore, necessary to examine carefully the various structural
economies, for which large scale control may, in different situations, be responsible.

Much has been made by some writers of the fact that, when a number of parallel establishments are grouped under a single head, the different plants can be thoroughly specialized to particular grades of work; and of the other kindred fact that the orders in any place can be met from the plant nearest to that place, and that, thus, cross freights are saved. That the economies resulting from close specialisation upon particular articles or even particular processes may, in some circumstances, be very great has been abundantly proved in the British engineering industry during the war. But it does not appear that a single control over many separate establishments is essential in order to secure these economies. Even though the different establishments were to remain separate, it might be expected, when once their great importance is realised, that the industrial organism would tend, under the sway of ordinary economic motives, to evolve them. In the paper industry of the United States, for example, each mill confines itself as a rule to the manufacture of some one quality of paper;
*30 and in the Lancashire cotton industry, not only are fine spinning, coarse spinning, and weaving localized separately, but individual firms frequently specialise on a narrow range of counts for spinning.
*31 The same thing is true of the economies obtainable from the utilisation of by-products. Nor does it appear that those economies in respect of marketing, which some writers ascribe to large-scale control, are a dominating factor making for combination. For, “if a manufacturer is purchasing raw material, there is generally a market price for it which all must pay, and which any one can obtain it for, so long as he buys the customary minimum quantity; while, if what he requires is a partly manufactured article, purchases amounting in value to hundreds of pounds per annum, accompanied by prompt payment, can generally be made at the cheapest possible rate. The sole advantage enjoyed by the largest concerns in the purchase of raw materials seems to me to lie in the possibility of occasionally
clearing the market of raw materials or of a surplus output of partly manufactured stuff, by some purchase quite out of the power of a smaller concern to compass. Such an operation, however, partakes of the nature of a speculation, and the profit, when gained, is hardly to be called a cheapening of the cost of production, if only for the reason that the opportunity for such a special purchase cannot be relied upon to occur very often, and, when it does occur, is perhaps as likely to result in a loss as in a gain.”
*32 Nor, again, should much importance be attached to those advantages of large-scale management which have been summarised as “concentration of office work, provision of central warehouse for goods, centralisation of insurance and banking, establishment of a uniform system of accounts, enabling easy comparison to be made of the working of branches, institution of a uniform system of costing and of a central sales agency,”
*33 and so forth. For these economies are scarcely practicable under the lower types of price-fixing Kartel, which are common in Germany, and, even in fusions and holding companies,
*34 they are very soon outweighed by the immense difficulty of finding people competent properly to manage very large businesses.

There are, however, certain structural economies of large scale management which are of a different order and have a wider reach. First, greater size, implying, as it does, greater wealth, makes it possible and profitable to spend more money on experiment. The Committee on Scientific and Industrial Research report: “Our experience up to the present leads us, indeed, to think that the small scale on which most British industrial firms have been planned is one of the principal
impediments in the way of the organization of research with a view to the conduct of those long and complicated investigations which are necessary for the solution of the fundamental problems lying at the basis of our staple industries.”
*35 This is obviously a very important matter; though it is not clear why a number of small firms should not, while retaining full independence in other respects, agree to collaborate in promoting research. Secondly, the union into one of what would have been many firms means that, instead of each wielding only the secret processes discovered by itself, each can wield the secrets of all; and in some circumstances this may involve large savings. Thirdly, a business combining many establishments is, in general, in contact with a number of different markets, in which the fluctuations of demand are, in some measure, independent. It is, therefore, in a position so to adjust things that the output of each of its component establishments shall vary through a narrower range than it would do if the several components were under separate control. But, if an establishment produces an average output A made up of (A+a) units one year and (A-a) the next, its costs are bound to be less than if it produces the same average output made up of (A+2a) units one year and (A-2a) units the next; for in the latter case it must have a capital equipment adequate to a “peak load” of (A+2a) units instead of the one of (A+a) units. That this is an important matter is illustrated by the eagerness of co-operative societies—creameries and so on—to assure themselves of the “loyalty” of their members, and of Shipping Companies to “tie” their customers to them by deferred rebates or in other ways.
*36 Furthermore, even though, apart from combination, the sum of the ranges of variation in the output of the component establishments were already reduced to a minimum,
i.e. to equality with the range of variation in aggregate output, combination might still lead to economies, by enabling the bulk of the plant to be run steadily, and reserving, after the pattern of the Sugar Trust, one specially adapted plant to adjust its output to the fluctuations in aggregate
demand.
*37 Fourthly, since it is much easier to forecast the incidence of various sorts of good and ill fortunes upon the aggregate of a number of separate concerns than it is to forecast the incidence upon each one individually, the operation of a business combining many establishments involves in the aggregate less uncertainty-bearing than the operation of its parts would involve if they were separated. The general economy resulting from this fact may manifest itself in the greater facility with which loans can be obtained, or in the lower price that has to be paid for them, or in the smaller proportionate reserve fund that the concern needs to keep for equalising dividends, or in other ways. The essential point is that the general economy, however it manifests itself, is necessarily there. The larger the unit of individual control, the larger is this economy. After a point, indeed, its growth, as the unit grows, becomes exceedingly slow. But, until the unit has reached a very large size, it grows rapidly, and constitutes a powerful force making for larger units—though, no doubt, among commodities suitable for grading, a speculative market may be developed, and may enable small concerns, through the practice of hedging, to put themselves, for
some sorts of uncertainty, on a level with large concerns.
*38 One further point may be mentioned. In certain special industries large-scale control not only achieves a direct economy by lessening the uncertainty-bearing that is involved in given fluctuations in the individual fortunes of different firms; it also achieves an indirect economy by reducing the probability that fluctuations will occur. It does this in occupations where public confidence is important, and where largeness of capital resource is calculated to create confidence. This condition is fulfilled in banking—the more so since publicity of bankers’ accounts has become common. The reason that banks differ in this respect from other concerns is, of course, that their customers are their creditors, and not, as in most trades, their debtors.
*39

§ 4. So far we have considered exclusively what I have called
structural economies. There is also another sort of economy that, in certain circumstances, favours the growth of large-scale management. So long as an industry is occupied by a number of establishments separately controlled, expenditure is likely to be incurred by all in defending their market against the others. A large part of the expenditure upon advertisements and travelers is, as was indicated in Chapter IX., of this character. But when, instead of a number of competing firms, there appear, in any section of an industry, a number of firms under a single authority, a great part of this expenditure can, as was also indicated in that chapter, be dispensed with. A and B being united, it is no longer to the interest of either to spend money in persuading people, whether through traveling salesmen or in other ways, to prefer the one to the other. It was stated, in regard to railways, before the Board of Trade Conference of 1908: “It is well known that railway companies find it necessary to spend large sums of money in canvassing against one another, and, if competition were removed by judicious amalgamation, the greater part of this money could be saved.”
*40 This economy is, of course, liable to be largest where, apart from unification, “competitive” advertisement would be largest, namely, not in staple industries providing easily recognized standard articles, but in various sorts of “fancy” trades.
*41

§ 5. Let us next suppose that the size of the individual firm and the size of the individual unit of control in an industry have been adjusted to the structural and other economies obtainable, and that the units evolved in this way are not large enough to exercise any element of monopolistic power. It is then clear that monopolistic power will
not be called into being incidentally, as a by-product of developments that take place without reference to it. But there still remains, as an influence tending to produce it, the direct expectation of the gains to which it may lead. When promoters have reason to believe that amateur speculators will expect a particular monopoly to prove more profitable than it really will do, this fact promises extra gains to those who form amalgamated companies, because it enables them to unload their shares at inflated values.
*42 Apart from this special consideration, however, we may lay it down that the magnitude of the gains obtainable from monopolisation depends, the conditions of supply being given, on the elasticity of the demand—
i.e. the fraction obtained by dividing a
(small) percentage change in price into the associated percentage change in quantity purchased—for relevant quantities of the commodity concerned.
*43 The less elastic the demand, the greater,
ceteris paribus, are the probable gains. Incidentally, it may be observed, this circumstance, coupled with that noted in the next paragraph, makes it profitable for a monopolist to extend his control over products that compete
with his own; for example, for the “Big Five” meat-packers of the United States to absorb both (a) non-American meat and (b) non-meat foods.
*44 The principal conditions of highly inelastic demand have now to be set out.

The first condition is that the commodity shall be of a kind for which it is not easy to find convenient substitutes. The demand for mutton is made comparatively elastic by the existence of beef, the demand for oil by the existence of gas, and the demand for the service of trams by the existence of omnibuses. In like manner, the demand for the service of transport by rail is probably more elastic in England than in continental America, because “the long broken coast-line in England and the great number of ports” render the competition of water carriage exceedingly powerful;
*45 and the demand for the services of any particular line of railway is, in general, fairly elastic, even where no water competition exists, in consequence of the indirect competition of lines running to other markets.
*46 From another field a good example of the point I am now considering is furnished by Jevons, in his book on the
Coal Question: “When the Government of the Two Sicilies placed an exorbitant tax on sulphur, Italy having, as it was thought, a monopoly of native sulphur, our manufacturers soon had resort to the distillation of iron pyrites or sulphide of iron.”
*47 As regards the kinds of commodity for which it is likely that substitutes can be employed little of general interest can be said. It should be observed, however, that the products of a district, or a country, whose efforts are directed
to leadership in quality as distinguished from quantity, are less exposed to the competition of substitutes than other products. For example, the prime qualities of beef and mutton in Great Britain have not been affected by the development of the American and Australian trade to nearly the same extent as the inferior qualities.
*48 It is, therefore, a commercially important fact that English manufacturers enjoy a very marked leadership of quality in wall-papers, fine textiles and cables, whereas in the electrical and chemical industries they are in a decidedly inferior position.
*49 Obviously, from the present point of view, we must include among the substitutes for any commodity produced by a seller exercising monopolistic power the same commodity produced by other sellers. The larger, therefore, is the proportion of the total output of product that a seller exercising monopolistic power provides in any market, the less elastic the demand for his services will be. Inelasticity of demand for monopoly goods is, therefore, promoted in industries where importation from rival sources is hindered by high transport charges, high tariffs, or international agreements providing for the division of the field between the combined producers of different countries. Furthermore, in order that the elasticity of demand may be affected by substitutes, it is not necessary that the rival source of supply should be actually existing. In some industries manufacture by people who are normally purchasers is itself a possible rival source of supply. Thus the Committee on Home Work observe: “Unless the price at which these articles (baby linen and ladies’ blouses and underclothing) are sold to the wives and daughters of the better-paid working men and small middle-class people is low, those who would otherwise be purchasers will buy the materials and make the articles at home.” The same remark seems to apply to laundry-work and charing. The poor housewife has the power, if reason offers, to do these things for herself. Consequently, the demand for the services of specialists at such tasks is exceptionally elastic.
*50 For example, it has been
remarked of Birmingham: “The washerwomen are among the first to suffer in any period of trade depression, for, as the first economy in bad times is to do your own washing, the tiny laundry with a very local connection is soon emptied.”
*51

The second condition, making for inelasticity of demand, is that a commodity shall give rise to only a small proportion of the total cost of any further commodities in the production of which it may be employed. The reason, of course, is that, when this proportion is small, a large percentage rise in the price of the commodity, with which we are concerned, involves only a small percentage rise in the price of these further commodities, and, therefore, only a small percentage contraction of consumption. Dr. Levy suggests that this condition makes the demand for the ordinary raw materials of industry highly inelastic.
*52 A similar line of thought brings out the fact that the elasticity of the demand for commodities at wholesale will be smaller, the larger is the proportionate part played by retailing and transport expenses in the cost of the commodities to consumers.

The third condition is that the further commodities, if any, in whose production our commodity is employed, shall be such that substitutes cannot easily be found for them. Thus the raw materials of the building trade should be subject, other things being equal, to a less elastic demand than those of the engineering trade, because foreign machines can compete with English machines much more easily than foreign houses can compete with English houses.
*53

The fourth condition is that the other commodities or services which co-operate with our commodity in the making of a finished product shall be easily “squeezable,” or, in technical language, shall have an inelastic supply schedule.

All these four conditions, which Marshall has distinguished, refer directly to the nature of people’s
desire for different commodities. There is yet another condition, dependent on the fact that the demand schedule derived from a given desire schedule for any commodity is only identical in form with the desire schedule, provided that the proportion of people’s incomes spent on the commodity is so small that variations in the quantity of their purchase make no appreciable difference to the “marginal utility” of money to them. When this condition is not realised, the following considerations become relevant. Suppose that there is only one sort of commodity in the world, and that it is impossible to store money. Then, whatever the form of people’s desire schedule for this commodity, their demand schedule is necessarily such that the same sum will be spent on it whatever the amount of it available; in other words, the elasticity of demand, in respect of all possible amounts of consumption, is necessarily equal to unity. From this we may generalise as follows: Given the elasticity of the desire for any commodity, the elasticity of the demand for it will diverge from this elasticity to a greater extent, and will approach more closely towards unity, the larger is the proportionate part of people’s income that is normally spent on this commodity. Thus, the demand for commodities which absorb a large part of people’s incomes cannot be either so inelastic or so elastic as it is possible for the demand for commodities which only absorb a small part of their incomes to be.
*54

§6. The preceding considerations suggest that units of control adequate to exercise monopolistic power will often be found, even though neither structural economies nor advertisement economies dictate their formation. The tendency towards this result is opposed by the difficulty and cost involved in bringing about agreements among competing sellers. This difficulty and cost depend upon the following general circumstances. First, combination is easier when the number of sellers is small than when it is large; for small numbers both facilitate the actual process of negotiation and diminish the chance that some party to an agreement will subsequently violate it. An attempt to form a Kartel in the German
match trade in 1883 is reported by Liefmann to have failed because no less than 245 separate producers had to be consulted.
*55 Secondly, combination is easier when the various producers live fairly close to one another, and so can come together easily, than when they are widely scattered. The reason why combination prevails in the German coal industry, and not in the English, is partly that, in Germany, the production of coal is localised, and not spread over a number of different districts, as it is in this country.
*56 A similar reason probably accounts, in great measure, for the excess of combination that appears among sellers in general, as compared with buyers in general; for, it may be observed, at auctions, where buyers also are closely assembled, combination among them is not infrequent. Thirdly, combination is easier when the products of the various firms are simple, of constant quality, not adjusted to individual tastes, and, therefore, capable of being intelligently and more or less precisely defined. Marshall wrote: “It is almost impossible to arrange a uniform price list for carpets or curtain stuffs into which wool of different qualities, cotton, jute and other materials are worked in varying proportions and with incessant changes in fabric as well as in pattern. There is no room for cartellisation of such things as biscuits, or ladies’ hats, in which versatility is demanded as well as high quality.”
*57 One writer suggests that a reason why English firms are combined to a less extent than foreign firms is that they concern themselves, as a rule, with the higher qualities, and the more specialised kinds, of commodities, rather than with “mass goods”;
*58 and another, in like manner, attributes the greater ease with which Coke Kartels are formed in Germany, as against Coal Kartels, to the greater uniformity of quality generally found in coke.
*59 Fourthly, combination is easier when the tradition and habit of the country is favourable, than when it is unfavourable, to joint action in general. When employers have been accustomed to act together in Chambers
of Commerce, in agreements as to discounts and rebates, or in negotiations with unions of workpeople, the friction to be overcome in making a price agreement is evidently less than it would be if they came together for the first time for that purpose. Thus: “The Association—such as the Merchants’ Association of New York—has, indeed, no monopoly power, but it is, nevertheless, of very great importance, owing to its socialising effects and its tendency to prepare the way for a stronger organisation, the combination or pool.”
*60 In like manner, the New Zealand arbitration law “forces employers into unions, for only thus can they defend themselves under the Act, and these naturally evolve into organisations for restricting competition.”
*61 Yet again, there can be no doubt that the various forms of joint action which British engineering firms, for example, were compelled to take during the Great War must have done much to smooth the way for future combination. Perhaps the opposing friction is also somewhat smaller when the producers concerned are companies than it is when they are private firms, in whose operations the sense of personal importance plays a larger part.

§ 7. The preceding section has tacitly implied that, where the gain from unification exceeds the cost and trouble involved, unification will, in fact, occur. This implication, however, is not warranted. It does not necessarily follow that, because an opportunity for agreement advantageous to all parties exists, an agreement will in fact be made. The reason is that mutual jealousy may cause A and B to leave the melon of common gain uncut, rather than that either should allow the other to obtain what he considers a share unduly large relatively to his own. Shall “participation” be proportional to the capacity of the several combining firms, or to their average product during recent years, or to the amount of the investment that has been made in plant and goodwill, or to some other quantity? “One manufacturer has patents and special machinery, which have cost him a great deal of money, and by which he sets much store. He will not enter the proposed combination unless these costs are made up to
him. Another manufacturer may have a large productive capacity, fifty nail machines, for example. He may have been unable to find a market for the output of more than half his machines, but in the combination, he contends, all his capacity will become available. He, therefore, insists that productive capacity should be the basis on which the allotment of shares in the trust should be made. A third man, by the excellence of his equipment and the energy of his methods, has been able to run his plant at its full capacity, while his competitor, with a larger productive capacity but a less favourable location or a less capable body of subordinates, has operated only half time. The successful manufacturer contends that average sales should be the basis of allotment.”
*62 Disputes on these lines may easily prevent agreement if direct negotiation between the different firms is attempted. It should, however, be noticed that they can, in great part, be obviated, and that the difficulty of combination can be correspondingly reduced, when an amalgamation is effected gradually by the process of absorption (exemplified among English banks), or when a company promoter, undertaking to buy up and consolidate a number of competing concerns, negotiates terms separately with each of them, without stating into what arrangements he has entered with the others.

43.

If
x is the quantity purchased and φ(x) the demand price per unit, the elasticity of demand is represented by
If this is equal to unity for all values of x, the demand curve is a rectangular hyperbola. The verbal definition of the text is an approximate translation of the above technical definition, so long as the term
small contained in it is emphasised. But, of course, a 50 per cent fall in price must be accompanied, if the elasticity of demand is to be equal to unity, by a 100 per cent rise in consumption. It will be understood, of course, that, when we speak of the gains from monopolisation depending on
the elasticity of demand, there is a tacit implication that the elasticities as defined above, at the several points on the relevant range of the demand curve, do not greatly differ from one another. Mr. Dalton has suggested (
The Inequality of Incomes p. 192
et seq.) that, when the price of anything increases by any finite percentage, the term “arc elasticity” should be used to represent this percentage change divided into the corresponding percentage change of quantity. But, since there would generally be a different arc elasticity for every different amount of price change from a given starting point, this new term might, in unpractised hands, easily lead to confusion.