Front Page Titles (by Subject) III - Cost of Production and Price Over Long and Short Periods
The Online Library of Liberty
A project of Liberty Fund, Inc.
Search this Title:
Also in the Library:
III - Frank H. Knight, “Cost of Production and Price Over Long and Short Periods” 
Journal of Political Economy, April 1921, xxix, no. 4, pp. 304-335.
About Liberty Fund:
Liberty Fund, Inc. is a private, educational foundation established to encourage the study of the ideal of a society of free and responsible individuals.
The text is in the public domain.
Fair use statement:
This material is put online to further the educational goals of Liberty Fund, Inc. Unless otherwise stated in the Copyright Information section above, this material may be used freely for educational and academic purposes. It may not be used in any way for profit.
Over longer periods of time supply and demand take on still different interpretations, especially important in the case of the supply. The supply now means the amount produced, viewed as a continuous average rate, and becomes a variable, controlled by producers' calculations. From this point of view price tends toward the point where the rate of production and the rate of consumption are equal, both being functions of price. It is axiomatic that goods cannot permanently be consumed more rapidly than they are produced and will not be produced more rapidly than they are consumed.5 For short periods of time this equality does not necessarily hold, for the reason that accumulated stocks serve as a sort of buffer between production and consumption. Consumption may exceed production for a considerable time, drawing down accumulations, and production may exceed consumption by building them up; but it is evident that neither difference can exist permanently or for very long.
The form of the functional relation between rate of production and price is the most complicated problem in price theory and to this problem the remainder of the discussion will be devoted. But one more distinction must be drawn by way of defining the data or given conditions of the problem. A change in the production of any given commodity may be associated with a change in the total productive power of the society as a whole or it may be related to a shift or transfer of productive power from one use to another. In general, again, it is a matter of the time interval taken into account. Over relatively short periods of a few years or a small multiple of the production period for the commodity, referred to under Case II, changes will generally represent transfers of productive resources and will be correlated with opposite changes in the production of some other good or goods. For such periods of a few years the total productive power of society does not greatly change. It is therefore as natural as it is necessary to separate the consideration of effects of changes in total productive power from those of transfers from one field to another.
The effects of these long-period changes in the total situation will not be taken up in the present discussion at all. We assume that the fundamental conditions of economic life in the aggregate, on both the supply and demand sides of the relation, remain unchanged. These fundamental conditions include (a) the total supplies of productive resources ("land, labor, and capital"); (b) the "state of the arts" or the knowledge of productive methods and processes; and (c) the "psychology," tastes and habits of the people. Significant changes in these things are generally progressive in character, in contrast to the readjustments to accidental fluctuations which make up the changes considered under the three cases already enumerated, and may be grouped under the heading of Social Progress. A social setting in which all such progressive changes are abstracted but in which unlimited time is assumed for all adjustments to the given conditions in these fundamental respects to work themselves out to their natural equilibrium results, is approximately what is meant by the "static state" or Marshall's conditions for the establishment of long-time normal price.6
We turn now to the crucial problem of the relation between the supply of a commodity and its price (meaning by supply the rate of production of the commodity) or in other words the problem of the form of the long-time supply curve. If supply is some function of price the meaning of the price point as the condition of equality between production and consumption is clear. Diagram III is drawn on the superficially natural assumption that an increase in price, other things being equal, will increase the production of the good, that supply is a direct function of price. Demand (rate of consumption) is of course an inverse function, as in the other cases.
The production of the commodity depends on the action of producers who are governed by profit-seeking motives and it is in this connection that cost of production exerts its effect on price. It goes without argument that cost affects price only as it affects supply, that any given supply put on the market will sell at a price determined by the demand, irrespective of its cost. The general character of the reasoning is simple. If the price is above the cost of production (including a profit representing payment at the general market rate for the entrepreneur's own services), production will be stimulated and the increased supply will bring down the price. If price is below cost, production will be decreased and the price raised.
From this point of view it is obvious that the costs which influence supply and price are the money outlays necessary to production. Ultimately these are the payments for the use of productive resources. We shall neglect the effects of taxation. We have no concern with the pains or subjective sacrifices involved in production, since it is not at all in terms of such "costs" that the entrepreneur makes his calculations on the basis of which he decides whether to produce the good or on what scale. He takes account of sentimental costs only in so far as they influence the outlays he must make to secure the services necessary to production. That is, he is concerned only with the price measure of his costs. Their magnitude in some other aspect will not influence his decision. Pains and sentimental repugnances are undoubtedly influences in limiting the supply of some sorts of services and raising their price, but in the aggregate they form a relatively unimportant element, and no one now contends that there is any tendency for the prices of productive services, still less of final goods, to bear any correspondence with these magnitudes. The relation between them is a separate inquiry, pertinent perhaps to an evaluation or criticism of the competitive economic order, hardly so to an explanation of its workings.
It would seem also to be almost too obvious for argument that in those costs which influence the entrepreneur's decisions and affect the supply and the price of a commodity, rent payments take their place among and in all respects on a parity with the outlays for other necessary productive services. They condition production. The entrepreneur must make these payments for the same reason that he must pay, say, wages; he must meet the offers of competing bidders for the use of the productive capacity represented. Even when all of these competing bidders are other producers of the same commodity the service being useless in any other industry,7 the payments are socially necessary as a means of effecting the distribution of the land among the different users and its rational combination with other agencies.8
Closely connected with the confusion involved in the interpretation of cost in subjective terms and the exclusion of rent is the notion of marginal cost, and the whole idea that one unit or portion of a supply costs or may cost more or less than another, under the conditions assumed for long-time normality. This point will be developed as we proceed,9 but since other confusions are also involved in the error, it should be mentioned here.
For the present let us first be clear as to what the assumptions or given conditions of our problem are. Progressive changes are eliminated, but unlimited time is assumed for the making of productive adjustments. That is, we are describing the tendencies operative in the relation of production to demand in terms of their final results in the absence of interference, as insisted upon at the beginning of the paper. Later on will be taken up the question of the advisability of a separate formulation of their effects when they operate for more limited periods (Marshall's short-time normals).
Under these conditions the supply curve is identical with a cost-of-production curve. The supply is a function of price because the cost of production per unit is a function of the supply, the amount produced. It follows at once from the relation between cost of production and price (see above, p. 313) that the amount which will be produced at any selling price (per unit) is the amount which can be produced at that cost per unit. That is, the same curve which shows output as a function of price shows cost as a function of output. In order to discuss the relations from the producer's point of view it is therefore advisable to reverse the axes of the diagram, treating supply as the independent variable and cost and selling price as functions of supply. This gives the same curves as before, but as seen in a mirror or looking through the paper from the back. (It is also evident that the demand curve may be regarded indifferently as showing selling price as a function of supply or the amount salable as a function of price, that these are two ways of looking at the same set of facts.) On the new diagram (IV), which represents a mirror image of Diagram III, the intersection of the curves shows in the more natural graphic way the equality between cost and selling price, which is the goal of producers' adjustments, though on either diagram, according to the direction in which it is read, it shows either equality of cost and selling price or equality of production and consumption.
Looking at the supply curve from this new point of view, it is evident that decreasing costs would mean that at higher prices less of the commodity would be produced than at lower prices. This certainly seems paradoxical and suggests that there is something wrong with the notion of costs decreasing as supply increases. The further course of the argument will show that decreasing cost as a long-run tendency is indeed impossible under a natural competitive adjustment of industry. Under the conditions assumed, an increase in the production of any commodity means a transfer of productive resources into the industry and a decrease in the production of some other commodity. But, other things being equal, this decrease in the production of other goods will raise their prices and increase the strength of the competing attraction which they exert on productive resources against the industry in question in which output is being increased. In simpler terms, an increase in the output of any industry involves increased demand for the productive goods used in it, which increased demand raises their prices, that is, raises the costs of production of the commodity turned out.
The implications of perfect competitive adjustment may now be briefly summarized and decreasing costs shown to be incompatible with the long-run tendencies of productive adjustments. In the first place, a perfect market for productive services is implied, that is, uniform prices over the whole field. The costs cannot be different to different producers or for different parts of the supply of any one producer, on this account. In the long run the same productive goods will cost the same prices and all differences of every sort in productive situations will be evaluated at their true worth under the influence of competition and be converted into costs which function in the same way as all other costs in the producer's calculations. Most of the apparent differences in production costs are undoubtedly due to imperfect evaluation of cost goods, and the tendency, however slowly it may work itself out, is manifestly toward a correct, uniform evaluation. Every productive good tends toward that position in the total productive system in which it has the greatest possible value, and tends to be priced at the value which it has in that position.
In the second place, the conditions of perfect competition include the production of every commodity by an indefinitely large number of competing organizations, each of the most efficient size. The confusion between variation in the scale of operations of the single productive establishment with variation in the output in the industry as a whole is perhaps the most prolific source of error in this whole field of reasoning. Under perfect competition, neither increasing costs nor decreasing costs in the individual establishment affect output or price. All establishments will be forced to the most efficient size, and variation in the output in the industry means a change in the number of establishments, without change in their scale of operations individually. This does not mean that all must be of the same size, but that each, in the conditions in which it works, must be of the most efficient size and that the efficiency of all must be the same. This again is not the actual character of the competitive situation at any given time, but is its actual tendency, and it is the long-run tendencies which must first be grasped as a basis for discussion of conditions under which they are but partially realized.
The specification of a plurality of establishments each of the most efficient size eliminates at once both the possibility of decreasing costs due to increased efficiency under larger output and also the entire notion of marginal costs, referred to above (p. 314 [See footnote 9—Econlib Ed.]). If increased economies are available through larger-scale operations, then larger-scale operations will be introduced under competition, through an increase in the size of the establishment with a reduction in the number of establishments and without an increase in the output of the industry as a whole. The tendency to an increase in size and reduction in the number of establishments will go on, independently of change of output in the industry, until either all establishments reach a size of greatest and equal efficiency (not necessarily equal size) or else until there is only one establishment left in the industry. Competitive production is possible as a final adjustment only if the technological conditions and the demand for the product are such that a large number of organizations are left in the industry when all are at the size of greatest efficiency. Otherwise the tendency is toward the establishment of monopoly.
In the same way the notion of marginal cost is meaningless in relation to any final adjustment. Competitive price can never be determined in the long run by an equation of the cost of the final unit of the supply to the selling price, leaving a profit on earlier units. The final unit cannot be more costly than any other unit in the ultimate competitive situation; for (a) costs must in the long run be the same to all producers, as shown, and (b) there cannot be increasing costs in the individual establishment because that would mean that smaller establishments are more efficient than larger, and if so they will put the latter out of business or force their reduction to the most efficient size.
The same reasoning applies to different productive methods. In the long run all producers are forced to use the most efficient methods or give place to others who do. The long-run tendency is toward a price determined by the cost of production under the best possible conditions, not the worst, as so commonly stated, nor those of the average or representative establishment.
The final consideration and in some respects the most difficult of all is the relation between output and the capacity of fixed or specialized equipment in the industry. A considerable fraction of the productive equipment in an economic society can be transferred freely from one industry to another and another fraction can be transformed by being replaced by a different kind instead of the same kind when it wears out, but of another large part neither assertion is true.10 From our long-run point of view the two former are equivalent; both amount to effective fluidity or mobility. But even ultimately it is not admissible to assume perfect mobility for all types of productive goods. Even if the tendency is finally toward some degree of mobility for productive goods generally, the time involved would be so very long that it is pertinent to grant the point and raise in the present connection the question as to the effect upon the cost function of assumed permanent specialization of cost goods.11
It is commonly and naturally assumed that if there is fixed equipment in an industry, not transferable to other uses, payment for its use represents a fixed cost and that a reduction in the output of the industry will be accompanied by an increase in the cost per unit. But if the entrepreneur's, that is, the realistic, point of view is rigidly retained, it will be seen that this is not true. The entrepreneur's costs are the payments for the services of the cost goods, and if the demand for a product decreases, the rigidly specialized productive services used in making it will be revalued at lower levels and these costs also will decrease. In the long run, of course, such considerations as the fact that entrepreneurs may have contracted for these goods for a considerable period of time at fixed rates fall away.
What is true is rather that payments for permanently and rigidly specialized productive agencies do not exert a causal effect on the price of the good in whose creation they are employed. There is no exception to the principle that an increase in output represents an increase in cost per unit and conversely. Moreover it is difficult to give any definite practical meaning to questions of the causal relation between cost and price; such questions are metaphysical, having little bearing on problems of policy. The practically pertinent facts are summed up in the statement that under all conditions (a) every productive resource tends to be employed in that way and place in which it will make the greatest possible contribution to the output of consumption goods as measured by pecuniary demand, and (b) that it (that is, its "owner") tends to be paid for its use the value of the contribution which it makes. The statement that the cost of production and the price of any good are equal really signifies simply that productive resources are divided between the production of that good and the production of other goods for which they might be used in such a way that none of the resources can produce more value by being transferred either way. If cost is above price, some productive services are being used for the good in question which are worth more somewhere else, and if cost is below price, some productive services are being used for other goods which would be worth more to produce the good in question. To avoid false inferences commonly drawn it should again be emphasized that there is no necessary connection between pecuniary demand and real worth and hence this reasoning in no wise vindicates the competitive system, and would not do so even if its tendencies came to literal realization. Our present concern is merely the question of accuracy in describing its workings, in terms of their final, long-run tendencies, which should be done correctly before critical judgment is passed. Under the conditions necessary to competitive production, and looking to the final results of competitive tendencies, the cost of production is without exception a direct or increasing function of output.
A more or less important qualification relates to the extent to which cost necessarily increases with output. For commodities which do not represent an appreciable fraction of the demand for any productive resource which goes into them, the change in cost corresponding to probable changes in output may indeed be practically negligible. The function may represent virtually constant cost. For example, the case of steel rails may be contrasted with that of carpet tacks. A considerable change in the demand for steel rails means a considerable change in the demand for the ultimate resources used in producing them, and will make a marked difference in the prices of these resources, that is, in the cost of production. No probable change in the demand for carpet tacks would make an appreciable change in the demand for any ultimate productive resource and hence within the limits of accuracy of economic measurement the long-run tendency is represented by constant cost. The supply curve of Diagram IV is for such goods a horizontal straight line, in Diagram III a vertical one. But constant cost is the "limiting case" which in strict accuracy is never met with. There is no place for a tendency to decreasing costs, when the conditions are correctly stated.
[5.]The latter part of the statement does not fit certain types of "durable" goods such as gold, jewelry, works of art, ideas, etc., which are not strictly speaking consumed at all. The theory of normal price (price determined by cost of production) is wholly inapplicable to such things, in the form which is valid for ordinary consumption goods.
[6.]The expression "unprogressive society," though less compact, seems to this writer much better than the "static state" to designate this situation. The word "static" suggests the absence of change. The idea is not however to eliminate change, but only certain changes while discussing the natural readjustment of other things to the given condition of those assumed as unchanging for purposes of the argument. The term "dynamic" contrasted with "static" is still more objectionable and "progressive" has in this case the advantage of being more euphonious as well. The distinction between progressive change and fluctuations seems to be important enough to justify a generic division along this line. It is not always true that progressive changes become practically important only over periods of time long in comparison to those in which fluctuations work themselves out, but it is so generally true as to make the division all the more significant and to make it easier to visualize the separation.
[7.]A condition doubtfully more often true of "land" than labor, bearing in mind that mineral resources are not economic land.
[8.]The separation of land from "artificial" productive goods is to the writer one of the hardest things to account for in the traditional economic speculation. It simply is not true that there is any productive power in land which has not been "produced" in the only sense in which men produce anything; its value is due to the form it is in, which represents previous investment, and the supply is determined by free investment in competition with other fields. The speculative element in such investment may be larger on the average but in the writer's opinion the reverse is more probably true.
[9.]See below, p. 317.
[10.]The division lines cut across all the conventional productive factors. Some "land," some "labor," and some "capital" (capital goods) are transferable, some transformable (over a longer or shorter period of time), and some rigidly specialized. Here as elsewhere the conventional division is irrelevant; the writer has yet to run across any real economic problem in relation to which it has practical significance.
[11.] It should be noted that it is impossible to be sure that we are adhering rigorously to the assumption that progressive change in total productive capacity is absent. When productive goods are changed in form there is no clear and definite meaning in the assertion that they remain the same in amount. The equivalence can be approximately preserved, in so far as the new forms represent the same amount of some more fundamental productive resource (such as homogeneous labor) as the old, but some differences in the kind as well as amount of the ultimate investment are doubtless always connected with differences in the immediate form of the production good. The question really is the extent to which production goods differing in form and specialized to certain uses do ultimately represent the investment of unspecialized resources. It is undoubtedly true that for the most part they do; but even then, some such investments never wear out and give back the unspecialized productive power which went into them for use in creating goods of some other specialized form.