1February 8, 1922

2[Écriture manuscrite autographe sur la couverture : This was the beginning of monop. Comp.]

3[1] This paper will take as axiomatic the general proposition that government regulation of any industry should attempt to approximate the results that would be attained under simple competition. Its purpose will be to examine the determination of rates in a railway system under what we may call the usual assumptions of economic theory to discover rules by which we may obtain this approximation more successfully. A comparison with industry in general is undertaken as an essential preliminary to any consideration of the rate problem to ascertain if the price reasoning of economic theory is applicable without qualification to the railroad industry, and if not, what qualifications are necessary.

Are Railways Fundamentally Different from Other Industries ?

4Most economic textbooks and writers on the railroad problem are accustomed to pointing out differences between railroads and other industries and concluding that they fall into a distinct class, for instance, that they are “naturally monopolistic” whereas other industries are naturally competitive, or that they are “diminishing cost industries” whereas other industries are typically “constant cost industries”. Thus they seem to be destined because of some fundamental differences for special treatment, and in the discussion of the determination of the price of the [2] service, new principles such as “charging what the traffic will bear” are introduced which were not found necessary in the general price discussion. I believe that few economists would dispute the statement that these differences are entirely differences in degree. Probably the consensus would be, however, that the degree is so great as to make them, practically speaking, differences in kind. While this position may be perfectly innocuous in a discussion of the advisability of government regulation, it is apt to lead erroneous conclusions when the subject for consideration is the determination of rates.

5Let us examine more closely the exact nature of these “differences”. Taking up specific points a moment later, we may first inquire just why it is that government interference is generally accepted as desirable in the case of the railroads whereas a policy in which “laissez faire” predominates is advocated for most other industries. Briefly, it is that the assumptions of economic theory are less perfectly realized in the railroad industry then in other industries. In the price analysis, the economist finds it necessary to make certain assumptions which are realized with varying degrees of accuracy in different industries. His conclusions as to the efficacy of freely-arrived-at prices to regulate any industry will depend upon the correspondence between such prices and the ideally perfect prices under his assumptions, in other words upon the degree to which his assumptions are realized in each case. In most industries capital is more mobile than in the railroad industry. The consequences of economically incor[3]rect and of irrational conduct (such as “cut-throat” competition) are less severe and hence play a less important part in the divergences from ideal results. The conditions, while perhaps far from the ideal assumptions, are at least so near that we believe them to be closer approximations than could be obtained by authoritative interference. If we did not so believe, we would interfere. But for the railroads, the assumptions are still farther from being realized, so far, in fact, that society believes that a policy of authoritative control can more nearly approximate the ideal results than a policy of non-interference.

6We are accustomed to saying that free competition does not work in the case of the railroads. The statement would be more clear if we used “laissez faire” instead of “free competition”. Granted the assumptions of economic theory, which include free competition, the railroads would not need regulation any more than any other industry, nor would the prices of transportation services be determined by any special considerations.

7In what specific ways are the railroads supposed to be sui generis, and how do they reveal themselves to be after all different only in the degree of divergence from ideal conditions?

81. Mobility of capital. It is indisputable that capital moves with exasperating slowness in the railroad industry. It has taken years to build some of our longer lines and existing roads have frequently enjoyed a practical monopoly for long periods, not because they enjoyed any absolute control of the supply, but because competitors [4] were slow in entering the field. But capital flows into any industry slowly, and temporary high prices due to the failure of new competitive capital to enter any line of productive activity are common occurrences. Moreover it takes time to get any large plant constructed and in operation, though perhaps there are few cases where it takes as long as for a railroad. Likewise, it is true that an unusually large amount of railroad capital is so highly specialized that once invested it cannot be got out again and converted to other uses. Equipment might be sold to other roads and real estate diverted to other uses, but fills, tunnels and bridges must be used where originally constructed and for purposes of transportation only. Similarly, manufacturing businesses in general may be able to convert large parts of their plant and equipment to other uses or sell them to competitors, but the operation of a business at a loss even for long periods simply because to stop operation altogether would involve a still greater loss is by no means uncommon, and is prima facie evidence of the impossibility of converting a large amount of the capital invested. Capital once invested in any industry cannot be converted to another without severe consequences being visited upon the entrepreneur concerned for his bad judgement.

9Under this head should be discussed the practica1 monopoly which a railroad may enjoy between intermediate points on its route. This is certainly not an absolute monopoly in its only legitimate sense of full control over [5] the supply of a commodity so that no competitor may enter the field (such as a patent or copyright or the ownership of all of the supply of a raw material). The control over price here is due to the fact that competition has not yet entered the field. Many of these temporary monopolies are being destroyed by the growth of the electric railway and or motor truck transportation. At bottom here the difficulty is not one of monopoly versus competition, but again one of mobility of capital. If capital were perfectly mobile, any return above the theoretically correct one in any industry would be immediately and automatically eliminated by a sufficient flow of capital to that point. Of course capital could not flow to the production of a good absolutely monopolized, by definition.

10To sum up briefly: Capital is generally speaking less mobile in the railroad industry than in most others, but this constitutes no fundamental difference between them.

112. Diminishing cost. “Diminishing cost” in this paper is always taken to mean an increase in total costs less than proportional to an increase in traffic. “Proportional costs” will always refer to the total cost divided by the number of units of traffic. The term “unit cost” is not used for a reason which will appear in a later connection.

12Railroads are almost without exception classified by writers as typically diminishing cost industries, although the statement is usually qualified as being true “within limits” or “most of the time”. Surely the matter merits close scrutiny, for it is diminishing proportional cost [6] that leads to “cut throat competition” and the tendency to monopoly under actual conditions.

13Let us consider first the case of the individual road. A railroad, in its growth and expansion, passes through the same stages as an industrial plant, but on a much grander scale. The initial outlay is always tremendous, and must be in order to run any trains at all. Frequently it takes years for the traffic to grow sufficiently to reach the point of greatest efficiency in the utilization of the plat. Extensions are also apt to be by large amounts, and hence there may be a succession of comparatively long periods when the road is “growing up” to its point of greatest efficiency with consequent decreasing costs, only to depart from it again with the next large investment and its consequent increasing costs. The same phenomena are observed, of course, in any industrial plant, except that the periods are much shorter. Even for a railroad, it must be true that investments subsequent to the first one may be and are made more gradually, consequently with less deviation thereafter from the most efficient “combination of factors”. Improvements in grading and roadbed, new and better bridges, stations, etc., and even double tracking can be done gradually and in places where the traffic is heavy and the new capital outlay most needed, thus making the expansion a more even process as it is in other industries. The doctrine that railroads are normally in a stage of diminishing cost probably had its inception during the period of extensive railway development, when, because of the large initial investment, most roads were actually in this condition. [7] With intensive development displacing extensive, it is doubtful if the roads are now forced to expand their plants much beyond the needs of the present of the very near future.

14But even though individual roads might, in some cases, find themselves in a condition of diminishing cost, it is difficult to see how the same could be said of the railroad system of the country taken as a whole. The more seasoned roads will tend to approximate closely the most efficient combination of factors at all times, expanding gradually as traffic demands. The newer roads will be expanding, perhaps more irregularly, with the result that costs on some will be decreasing and on others increasing. The expansions coming at different times on different roads should maintain the ratio of costs to volume of traffic comparatively constant.

15At this point we should distinguish carefully between the normal tendency in an industry and its condition at any particular time. It is the failure to make this distinction clear and the use of “diminishing cost” in a dual sense that has caused confused thinking on this point. We have, of course, been dealing with the “normal”, which would exclude fluctuations in demand. It was maintained that a railroad which was comparatively well along in its development, since it could add additional factors of any kind by small increments, would be able to expand slowly and steadily as traffic expands and hence always be operating at approximately the least proportional cost or greatest efficiency. The other phase of the problem ‒ the condition [8] of the industry at any particular time as affected by immediate demand conditions ‒ is so important in our present connection that we must consider it at some length.

16Slight fluctuations from “normal” conditions in any industry will cause that industry to be operating at a greater proportional cost than under “normal” conditions, where, presumably, the greatest efficiency prevails. Suppose the demand to decrease. Production will be cut down by the amount of the decrease, and costs will be cut down by the amount which was incurred on account of the units discontinued. In no case will the reduction in total cost be proportional to the reduction in output because there are always some constant costs (overhead) which are not affected. It is for this reason, of course, that proportional costs are increased. In ordinary manufacturing, the raw materials and labor, and in retailing, the cost of the goods sold, constitute a large share of the total, and since these costs are automatically reduced with a decline in volume of business, proportional costs are raised only slightly. In the case of a railroad, however, the percentage of costs which cease when traffic decreases is much smaller, and consequently proportional costs mount rapidly. The same percentage fluctuation in demand will produce a much wider fluctuation in proportional costs for a railroad than for most other industries. This is the only sense in which a well developed railroad is ever in a condition of diminishing cost ‒ a sense in which the same may be said of any business, but to a lesser degree. [9]

17A word should be said specifically about “cut-throat” competition under this head, for it is diminishing cost that provides the incentive for rate wars. There can be no doubt that the rate wars in our railroad history were due in considerable measure to an actual condition of diminishing cost on account of the early stages in the development of the roads at the time, and because of too many roads being built for the traffic available. This cause becomes of less importance as a system develops, and as it becomes possible to extend capital equipment more gradually. But the uneven demand for transportation services, due to seasonal fluctuations and the larger movements of the business cycle, will always be present. Most of the time the railroads will be in a position where increased output of their product will mean decreased proportional cost, just as all industries are most of the time in this condition. The incentive to “cut throat” competition is, of course, present in all industry (and is not unheard of in others), but it will always be much greater in railroading because of the rapidity with which proportional costs decrease with increases in volume of business.

18It should be mentioned, before dismissing the subject of diminishing cost and “cut-throat” competition, that diminishing cost in either sense could never cause ruinous competition if human beings were rational, so that we need make no qualifications on this score in constructing a rate theory. But more of this later.

193. Joint Cost. The costs to be discussed here are put [10] under the head of “joint” costs because that is the term most frequently applied to them in economic writings. Several principles will be set down, which will need little proof, and which will have a vital connection with the theory of rates.

20First, to dispose of the term and substitute a better one. The term “joint costs” was originally used in economics and should continue to be used to denote costs which are necessary to the production of two or more commodities which must inevitably be produced in a fixed ratio. Thus the cost of growing cotton is a joint cost of cotton and cotton seed oil, since for every pound of cotton produced, a fixed amount of cotton seed oil is also necessarily produced. Clearly the costs usually called joint in the railroad industry are not of this character. A large percentage of the total costs are common to several commodities hauled, but a train of cars may be made up of goods hauled in any proportions. Hereafter, in this paper, the term “common costs” will be used to denote costs which are incurred for the traffic as a whole or for a group of commodities, but which cannot be allocated to a particular commodity in the sense that they would not be incurred if that commodity were not hauled. Those which can be so allocated will be called “specific costs” of the commodity.

21Secondly, the total costs of furnishing transportation services cannot be divided into (1) common and (2) specific without further elaboration and explanation. In classifying costs, we have at the one extreme the specific costs of [11] transporting a unit of a good, and at the other, the costs common to the transportation of all goods and persons. Between these two, costs are, in a sense, both common and specific. The specific costs of hauling a freight car are common costs of all the goods in the car. The specific costs of running a train between two cities are common costs of hauling all the goods on the train. The specific costs of the passenger service (for instance, the conduct of passenger depots) are common costs to the transporting of all passengers, but not to the transportation service as a whole, and so on. The importance of this classification will appear in the discussion of rates.

22Thirdly, common costs are not unique in the railroad industry, but are found in all fields and are usually designated as “overhead”. It is surely true, however, that they comprise a disproportionate share of the total railroad costs, and hence the problem of determining the price to be charged for the transportation of a good is one of exceptional difficulty, although, as we shall see, no new principles are involved.

23To sum up briefly the discussion thus far: Railroads are not an industry sui generis. Since they operate under conditions further removed from the perfect conditions assumed in economic science, a greater degree of control by the state becomes necessary than in other fields. But the difference being only in degree, we should expect to find the values of transportation services determined under the same laws that govern the values of produced [12] goods in general. It shall be our next task to consider this proposition more in detail.

What Principles Govern the Value of Railroad Services?

24In a brief paper of this nature it would be impossible to develop and set down a complete theory of railroad rates. Indeed, if the conclusions so far are valid, such a theory would consist merely in the generally accepted value theory of economics, perhaps with a few changes in emphasis. No different principles are necessary to explain the determination of the value of transportation services than the value of any other economic good. This section will be devoted, therefore, to a brief discussion of two phases of the problem, which would seem at first thought to call for modification. We shall consider, in order, these two questions: (1) In so far as diminishing cost is a factor, what is its influence on the value of the product? and (2) how are the common costs to be distributed between the various goods hauled?

251. Diminishing Cost. In the first place, let us dispose of the fluctuations from “normal”. It will be recalled that most businesses are at almost all times able to diminish their proportional costs by an increased output, because of the uneven flow of demand. The incentive to “cut-throat” competition is comparatively weak in most businesses for at least two reasons: (1) the decrease in proportional costs with a larger volume of business is small because the constant costs are small; and (2) the variable costs are [13] great – hence the limit to the possible price cut is set so early that the resulting increase in demand would be almost negligible. The incentive is great in the railroad industry for the converse reasons. Of course “cut-throat” competition is essentially irrational; no one looking beyond the immediate future would initiate such a policy in any industry, although one irrational competitor would surely compel all of his perfectly rational rivals to follow him in such a policy to avoid worse consequences. So far as our present problem is concerned, we may rule out once and for all such irrational action, and inquire simply what would be the result if railroad traffic managers conducted themselves during office hours like sensible human beings. So long as fluctuations in business activity persist, it is almost inevitable that rates should follow such fluctuations to some extent. The more immediate forces working at any particular time would be, as in other industries, those of monopoly value. Either in time of a slackened demand or a demand in excess of the possible output, the rates set by free competition of rational men would be identical with those which would result from conscious cooperation – those which will yield the largest net return to the industry as a whole. The return over a period of years would, of course, be approximately commensurate with the risk involved compared with other industries, and would, in every sense, be a competitive one, except in so far as the railroads were in the early stages of development, and hence, for the time being, “normally” in a condition of [14] diminishing cost. This case will be next be considered.

26This phase of the problem contemplates a railroad, or perhaps two competing railroads, built ahead of their time in that the amount of traffic now available for them is not sufficient to enable them to operate in a condition of least proportionate cost. It was brought out earlier that this is the usual situation in the early stages of a railroad’s development because of the large initial investment necessary to do any business at all, and the difficulty of expanding gradually until a somewhat larger stage of development is reached. This means, of course, that the entrepreneurs must be satisfied for the time being with less than the rate of return in a fully developed industry. But it does not mean that they should be denied the right to charge a monopoly price while such a condition exists, it being remembered that, by our hypothesis, such a price will not result in a net return which is as great as that of any well developed industry of the same risk. The entrepreneurs, in setting a rate in these early stages, will naturally look further than in the greatest immediate net return, taking into due consideration the fact that the lower the present rates, the quicker will the territory develop and a higher return be made possible through fuller utilization of plant and the resulting large decrease in proportional costs. Two railroads competing for a total amount of business which might, perhaps, be more than adequate to utilize most efficiently the plant of either, but is not as yet sufficient to utilize both most efficiently, would not, if their [15] managers acted rationally, cut rates in efforts to get all of the business. The rate set independently under rational competition would be exactly the same as the rate set by deliberate agreement – the rate which would yield the largest net return to both taken together. It would be reasoned that any reduction from this rate by either party would necessarily lead to the same reduction by the other and a consequent lowering of the earnings by both.

272. The Distribution of Common Costs. We are aided in the solution of this difficult problem by the conclusion reached in the first section of the paper. The costs are surely “distributed” the same in all industry. “Charging what the traffic will bear” is either always or never a result of free competition. Professor Pigou in his “Economics of Welfare” concludes that it can result only under monopoly. We shall endeavor to show that the practice inevitably accompanies free competition in all fields.

28The most direct answer to the question of how the common costs are to be distributed is that they cannot be and hence are not to be distributed at all. An appreciation of the full significance of this can lead to but one conclusion – that since there is no possibility of allocating the common costs to specific commodities, the forces of demand must be relied upon solely to determine how much the rate charged is to exceed the specific cost of hauling a commodity, and, further, that unless the forces of demand are allowed full play in the matter, economically incorrect results are sure to be realized. [16]

29From the fundamental law of economics that all the forces operative in price determination can be brought under the heads of demand (utility) and supply (cost), it must follow that unless a price is determined solely by the intensity of demand, cost has entered in to some extent. Applied to the problem at hand, this means that in any rate theory or practice, unless a rate on a commodity is greater than the specific costs by an amount determined solely by “what the traffic will bear”, cost has entered in (perhaps unconsciously) – a certain part of the common costs has been considered as belonging some way or other to individual commodities. Examples of this are numerous – theories of allocation on a basis, for instance, of weight, bulk, ton miles, or prime costs (Pigou; we have called these “specific” costs), and the practice of the Interstate Commerce Commission, in exercising its rate making authority, of allowing such matters as weight, bulk, desirability of traffic, etc., to exert an influence beyond that of their natural effect upon specific costs.

30The assertion that common costs cannot be allocated may seem to be a mere word quibble. It may be contended that if the “value” of the service is given full play, we are simply distributing costs on another basis – that of “what the traffic will bear”. But the point is so vital to correct theory that we must be a bit dogmatic in our insistence upon the adoption of a different terminology. Common costs are, by definition, common to a number of commodities. If any part of them ceased when a particular [17] commodity was not produced, it would not be a common cost, but would be a specific cost of that commodity. In segregating the common costs, we have already distributed all the costs that can be distributed and called them specific costs of the commodities to which they pertain. Our next problem is not one of distributing costs, but of fixing a price for the individual commodities such that their specific costs will be covered in each case and such that the total revenue yielded will cover the total cost incurred (including a reasonable profit). Such a rate will have no relation to the cost of individual commodities except that it cannot go below their specific costs. Indeed the phrase “cost of the individual commodity” is a misnomer, it is illegitimate, it conveys no meaning, except in the sense of the specific cost of the commodity.

31The fallacies into which one may be led by a failure to comprehend the significance of this proposition and by a consequent careless use of the term “cost” are well illustrated in Professor Pigou’s contention that under simple competition coal and copper could bear the same rate per ton mile. Throughout the famous controversy between Professors Pigou and Taussig in the Quarterly Journal of Economics there seems to be a failure to comprehend the real issue involved. They are both seeking an answer to the same question with which we are here concerned – what would determine the rates under free competition? Professor Taussig is right in his contention that although equal rates per pound (of copper and coal) might result “if each [18] sort of transport were conducted quite by itself”, they will not “necessarily or probably be the same on two such articles” when the same plant is used to haul both. But he does not prove his point, and he refuses to meet the issue on the use of the word “joint”, simply maintaining stubbornly that the costs cannot be allocated, disregarding what is evident upon a moment’s reflection, that the commodities are not hauled in fixed ratios, and hence that the costs are not joint in the sense that the term is usually used in economic literature. Professor Pigou establishes this latter point, but reasons that since the proportion of coal and copper hauled is not fixed, under simple competition the same rate for both would result, since a higher rate for one would lead to more of that commodity and less of the other being hauled, which would continue until the two were brought to equality.

32The error in this position does not reveal itself without some analysis. The doctrine is succinctly stated on page 265 of the “Economics of Welfare”, where he says: “If there are a number of competing sellers supplying transportation, or anything else, to several markets with separate demand schedules, and if the price in one of these markets is higher than in the other, it is necessarily to the interest of each individual seller to transfer his offer of service from the lower-priced market to the higher-priced market; and this process must tend ultimately to bring the prices in the different markets to a uniform level.” Evidently Professor Pigou is considering two commodities whose costs [19] (of transportation) are the same, and arguing that a higher rate for one than for the other is self destructive. The fallacy is, briefly, that the costs of hauling the commodities are not known. If he has taken cost to include a share of the common costs, he is assuming an allocation as having already taken place, whereas he is engaged in demonstrating the method by which it should take place. If he has taken cost to mean prime or specific costs, he again assumes what he is proving in order to prove it, but not so directly. An entrepreneur might conclude that since he was getting a higher price for one of two commodities whose specific costs were the same he was therefore making more profit on that commodity, but such a conclusion would be irrational. In order to know the profit on any commodity, he must know, in addition to its selling price, its cost; in order to know its cost, the common costs must have been allocated. The motives which operate to distribute common costs equally, thus depend for their operation upon a previous distribution of common costs. These logical fallacies render Professor Pigou’s conclusions erroneous.

33It may still be contended that an entrepreneur is surely not acting irrationally if he substitutes the production of a commodity yielding a higher price for that of another, the specific costs of the two being the same (the common costs could evidently not be influenced, for the specific costs would reflect all the differences). The answer is that nothing could be more rational than such a substitution, provided that the demand conditions for the two articles were such that it could be made. Entrepreneurs, in [20] deciding what proportions of goods involving common costs to produce and that prices to charge will see to it (1) that the specific costs are always covered, and (2) that the prices for individual goods are so adapted to the demands for them that the total return will exceed the total cost by a maximum, that is, that profits on the business as a whole will be greatest. (Competition of other entrepreneurs will, of course, keep these from going higher than a normal competitive level). Indeed, in any business where there are costs common to all the goods produced, the term “profits” can have meaning only when applied to the business as a whole. If cost of the individual commodity is a meaningless term under these conditions, as we have maintained, then surely profit on the individual commodity is just as meaningless.

34It is the important proviso above that Professor Pigou entirely overlooks, and which is essential to establishing his case. Even though two commodities of equal specific costs are yielding different rates, entrepreneurs are not going to bring the rates to equality, even under simple competition if the effect is to diminish their total profits. We have attempted to establish this by theoretical analysis. An examination of what actually happens in any recognized competitive field will surely substantiate the conclusions reached. A restaurant keeper, because he is making a higher percentage return over the prime cost of coffee than over that of pie does not stop selling pie, nor even raise the price of pie and lower that of coffee in an [21] attempt to bring the return on the two to equality because he knows that to do so would diminish his total profits. Nor are there any forces of competition that I have been able to discover that will bring the percentage of return over the prime costs of the two to equality in the restaurant business as a whole. To be sure, competition will force a single price for each commodity over the whole field. But that should not be confused with the matter under discussion.

35Before closing, one further refinement should be made. On pages 10 and 11 there was mentioned a necessary relativity in the use of the words “specific” and “common”. The theory, to be accurately stated, should take this into consideration. An illustration will perhaps bring out the point involved most clearly. The rates on the individual commodities hauled on a freight train from Omaha to Chicago must be sufficient in each case to cover their own specific costs, and sufficient in addition to cover the specific costs of hauling the freight train, which are common costs of the individual goods hauled. A lengthy statement of the refinement in all its ramifications is perhaps rendered unnecessary by our insistence that the specific costs (of anything) must always be covered.

36Our conclusions are, then, that under perfect competition, (1) the price of any commodity cannot go lower than its specific costs, (2) it cannot go higher than the price which will yield the largest total return over the total specific costs of all producers, (3) any individual [22] price may be anywhere between these limits, depending only upon the strength of the demand for it, and (4) the total return to the enterprise cannot exceed the normal competitive return, through the action of competing entrepreneurs. The last statement may seem to contradict (3). Of course the total return could hardly be affected without its parts being affected. But since it would be possible to have individual rates at both limits mentioned in (1) and (2), and still have (4) realized, it seems more accurate to state the theory as above.

37The truth of the conclusions is perhaps rendered more evident by showing that any other system of prices (rates) would be destroyed by free competition. Suppose an entrepreneur in any field firmly resolves to determine his specific costs and “distribute his overhead” according to some arbitrary plan, believing the results to represent cost, and setting his prices accordingly. He will find either (1) that he is not utilizing his capital most efficiently, and that others, “charging what the traffic will bear” are able to undersell him through a larger volume of business with the same amount of capital invested, or (2) that although his capital is utilized most efficiently in the sense of volume of physical business, competitors underbid him again because, by adapting their prices to the demand they secure a greater revenue from the same amount of physical business, and are forced to lower their prices by other competitors attracted to the field by the abnormal profits. Our entrepreneur will now find himself underbid with respect to a number of commodities and will be forced [23] to conform to the strength of the demand in order to stay in business. “Charging what the traffic will bear” in any business is only another way of saying “utilizing the plant most efficiently”. The truth of this makes elaborate proof superfluous. Naturally the entrepreneur who utilizes his plant most efficiently will be the one who survives.

38Throughout the discussion I have tried to bring out that “charging what the traffic will bear” is present wherever costs are necessarily incurred common to two or more commodities. The practice is not “theoretically incorrect but a practical necessity in the railroad industry”. It is an inevitable phenomenon of free competition. Of course the principle is especially important and especially in evidence in an industry like the railroads where the common costs are large. But its importance in industry in general should not be overlooked. It seems to me that there is room for some refinement in economic theory on this point.