but it did not die out entirely. 29 Other approaches were also tried but none of them met with any great success. 30 Second, extensions of the concept of rent just discussed suggest themselves readily in view of the difficulty of drawing a logically tenable line between what objects are, and what are not, natural agents or, which is only another way of expressing the same thing, of agreeing on the defining characteristics of natural agents. Thus, violating his original definition of rent (income derived from the ownership of land and other free gifts of nature, Principles, p. 150) Marshall denied that mining royalties are rents. 31 Other writers were more impressed with the analogy. But he made no difficulty about the extension—if indeed it can be called an extension at all—of the concept of rent from rural to urban land. 32 Much more important, however, was one of his most felicitous creations, the concept of quasi-rent or ‘income from an appliance for production already made by man,’ that embodies the recognition of two facts that were particularly important in connection with the new theories of interest: the fact that any price paid for the services of capital goods is closely analogous to the price for the services of natural agents; and the fact that this analogy holds particularly for the short run and decreases with the increase in the length of the time to which a proposition is intended to apply. 33 Another class of extensions sprouted directly from Ricardian roots. A man who still persisted in seeing point in Ricardo’s emphasis upon ‘differential rent’ was likely to discover, as Bailey had done before, that such differentials were not confined to land. We have had occasion 34 to notice Mill’s, Mangoldt’s, and Walker’s interpretations of entrepreneurial gains as rents of differential ability. Marshall presented the general case for the latter concept though, in my 29 F.Oppenheimer’s book on David Ricardos Grundrententheorie (1909; 2nd ed., 1927) may be cited as an example. 30 See, e.g., Achille Loria’s work, Rendita fondiaria…(1880). 31 Moreover he did so on the untenable ground that royalties do enter into the price of the mineral mined in a sense other than that in which does the rent of agricultural land. 32 Also see Edgeworth’s work on the subject (republished in vol. I of his Papers) and Wieser’s Theorie der städtischen Grundrente (1909). The latter reads like an application of Ricardo’s theory of rural ground rent, Ricardo’s marginal land being replaced by the ‘peripherical’ urban land that yields no higher rent when used for building than it would in its optimal agrarian use. 33 There is, of course, no sharp dividing line between quasi-rent and rent proper. If we assume that the greater part of a landlord’s income is also quasi-rent, then we may say that, at any given moment, the bulk of capitalist income (in the Marxist sense) is quasi-rent. Also there is no sharp dividing line between quasi-rent and wages. A physician’s income is in part quasi-rent though usually classed with wages. The ‘rent’ or ‘quasi-rent’ yielded by a piece of land that has been reclaimed by the owner’s labor is, in part, at least, in the nature of wages. A little reflection will show that this is more than an otiose play with concepts. 34 [On Bailey, see above, Part III, ch. 4, sec. 3c; on the interpretation of entrepre neurial gains as rents of differential ability, see above, sec. 2b.] History of economic analysis 902 opinion, this only served to expose its emptiness. 35 Similarly, a man who has acquired Ricardo’s habit of deducing rent from the physical ‘law of decreasing returns from land’ may easily discover the ubiquity of this phenomenon wherever factors are being applied to a fixed quantity of one of them: 36 this amounts to generalizing Ricardo’s rent concept by generalizing Ricardian decreasing returns. If the fixed factor is plant and equipment— which may in fact be taken as fixed in the short run—we shall, after a certain point, observe decreasing physical returns to successive ‘doses’ of those factors that can be varied in the short run, and the Marshallian quasi-rent then appears as the exact analogue of the ‘Ricardian’ rent of land. 37 Third, 38 the aspect of the ‘Ricardian’ theory of rent that appealed most to policy- minded economists was the aspect suggested by the words Surplus or Residual. Strictly speaking these words, as applied to the rent of natural agents, had lost their meaning in the Jevons-Menger-Walras analysis, which was no longer under the necessity of explaining rent as a ‘leftover’ sui generis but was 35 Principles VI, ch. 5, 7 and ch. 8, 8. The statement in the text requires protection against two misunderstandings. First, it must not be understood to refer to the entire contents of the two paragraphs just cited, which contain many a profound remark. Second, it must not be understood to imply denial of the importance—which on the contrary I think paramount both for economic and for sociological analysis—of the wide range of variation of the ‘natural’ abilities of men. All I mean is that the theory of rent contributes nothing to our understanding of the role of supernormal abilities and that nothing is gained by calling their earnings rents except that by so doing we are enabled to show—which was indeed Bailey’s purpose—that differential fertility of different plots of land is entirely superfluous also in the theory of the income from ownership of natural agents. 36 Let us repeat that it was in this manner that J.B.Clark found his way toward a marginal productivity theory of distribution. At least this seems to be the natural inference from his ‘Distribution as Determined by a Law of Rent,’ Quarterly Journal of Economics, April 1891. But in order to be able to travel this way, he had to renounce community with Ricardo, for whose teaching it was essential to hold that the rent phenomenon is specific to the land factor. 37 More precisely, it appears as the exact analogue of the so-called second case of Ricardian rent, which refers not to the application of capital and labor to decreasingly fertile or more distant plots of land but to the application of successive ‘doses’ of capital and labor to the same plots of land. It should be observed, however, that the true importance of the quasi-rent concept is quite independent of this analogy. 38 I am not going to include here the ‘psychological rents’ that result from the ‘laws’ of decreasing marginal utility or increasing marginal disutility. One of them, consumers’ rent, will be considered later (see below, ch. 7, sec. 6 and Appendix). In addition, we may speak of a savers’ rent that may be derived, if we wish, from the fact that savers’ behavior can be described in terms of a marginal equality between the advantages of saving or consuming an additional dollar’s worth of resources, and that there is hence a surplus of advantage on the intramarginal dollars; and of a laborer’s rent (Marshall’s producers’ rent) that may be similarly derived from the balance, at the margin, of the advantages of another hour’s leisure or work which implies that there is a surplus of advantage on the intra-marginal hours of work. The validity of these concepts is one question; their value is quite another. In any case, these ‘rents’ must not be confused with those associated with ‘laws’ of physical returns. General economics 903 able to explain it directly and on the same fundamental principle with other types of income. But economists soon found that they might retain the surplus aspect all the same. Rent might be capable of being interpreted as the payment for the services of a requisite of production, but this payment was not necessary in order to call forth the corresponding service, whereas it was in the cases of the services of capital goods and of labor—a fact that seemed important for questions of welfare economics and of taxation. Marshall, shifting emphasis to this surplus aspect of rent, gave a lead toward expressing it by saying that the services of natural agents were ‘costless’ in the sense that in order to have them, society need not incur ‘real cost’ (disutilities of labor and saving). 39 But if we elaborate this unearned-surplus aspect of rent, we discover two things. In the first place, we discover, as we discovered before from another standpoint, that rent defined as a ‘surplus’ is no more confined to natural agents than is rent defined as a productivity income. Similar surpluses, that is, differentials over and above the payments that would be necessary to call forth the corresponding supplies of goods and services, are scattered all over the economic organism. Many workers, and not only movie stars, receive much more than the amount necessary to induce them to do what they are actually doing, and in many cases they would offer more service if they were paid less per service unit. Even if we carry the hypothesis of perfect competition as far as it is possible without getting ludicrously out of contact with facts, there are plenty of situations of advantage, some short-lived, others more durable, in which such surpluses are earned. Under conditions of monopolistic competition, let alone straight monopoly, 40 such situations must be still more frequent. Finally we may include gains from situations of advantage that are created by ‘collusion’ (contrived scarcities) or by specific 41 institutional pat- 39 The reader will observe that this is something different from saying that the payment for such services does not ‘enter into the price of products.’ 40 Again, the reader will perceive the essential difference between subsuming monopoly gains under surpluses of this kind and explaining the rent of land as a monopoly gain: the two have nothing to do with one another. But any difficulty the reader may have in realizing this bears witness against the advisability of calling these surpluses rent. It constitutes in fact a typical instance of unnecessary confusion being created for no better reason than a preference for terms that, like rent, have acquired derogatory associations. If it were not for this, it would be readily recognized that ‘surplus’ does all that is needed and that the term ‘rent,’ in this connection, is redundant. 41 By emphasizing the word ‘specific’ I mean to convey that mere reference to the general institutions of capitalist society, such as private property, covers an economist’s failure to explain a return more often than it affords explanation. This is why, much to the displeasure of critics of a certain type, competent economists of all times have always looked askance at work that uncritically used the phrase Institutional Rent. The case is different, of course, where the existence and the modi operandi of specific institutional factors can be fully established. Examples are afforded by protective duties (including measures of some of our states that in effect amount to protective duties), certain features of recent labor legislation, and so on. History of economic analysis 904 terns. A tendency has asserted itself in our own time to combine all such surpluses under the heading of rent. Though the incomes from ownership of natural agents are included, they only form a special case of rent in this sense, the theory of which has but little in common with the theory of rent surveyed in the first part of this subsection. In the second place, however, we also discover that part of the surplus gains under discussion divide up into two classes between which there is an analytically significant difference. Consider a natural agent, perfectly homogeneous in quality, perfectly divisible, and perfectly transferable between the different uses (industries) that it is capable of serving, and let us assume perfect competition all round. Each of these uses will then be governed by what we have called opportunity cost. The users of the factor may therefore have to pay all that its services are worth to them, in which case no surplus attaches to its use. 42 And the owners may receive, from each group of users, no more than they could receive from any other. They derive no surplus over opportunity cost, though the whole of their receipts may be, in another sense, a surplus over Marshallian real cost. 43 In another class of cases this is not so. It is unnecessary, I hope, to adduce examples in which owners of requisites of production, whether or not they be natural agents, earn surplus gains over opportunity costs: technological difficulties of ‘transforming savings’ into certain types of capital goods suffice to create, for the owners of the latter, gains over opportunity costs that are also gains over real costs which, at least in the short run, even otherwise quite unimpeded competition may be inadequate to remove. 44 The distinction between surplus gains over real costs that are also surplus gains over opportunity costs, and surplus gains over real costs that are not, is sufficiently important to be recommended to the reader’s attention, especially because the former do not, and the latter do, play an essential role in the process of allocation of resources. 45 42 This does not hold for marginal quantities if the intramarginal ‘surplus’ is absorbed by payments to other factors. [J.A.S. questioned this note and the sentence to which it applies.] 43 This proviso should suffice to negative any idea of using opportunity cost for apologetic purposes. 44 The wording of the sentence above is inspired by a passage in Pareto’s Cours (2nd vol., 745 et seq.). Pareto’s share in this generalization of an aspect of the old rent concept, which was to gain popularity in the Anglo-American literature of our own epoch, deserves to be mentioned in spite of doubts as regards the analytic progress thereby accomplished. Not that the facts envisaged by the Paretian theory of rent are unimportant. But I can see no point in grafting them upon the quite heterogeneous facts that are satisfactorily described by the marginal productivity theory. See, however, the work of one of Pareto’s disciples, Professor G.Sensini, La teoria della ‘rendita’ (1912). 45 See especially Joan Robinson, The Economics of Imperfect Competition (1933), ch. 8. Mrs. Robinson may be considered to be the chief authority on the ‘new’ theory of rent at which we have glanced. Four references on the subject of this subsection are added to enable interested readers to fill the many lacunae of our sketch: A.S. Johnson, ‘Rent in Modern Economic Theory,’ Publications of the American Economic General economics 905 (c) Wages. Longfield’s and Thünen’s old marginal productivity theory of wages was the new thing in the 1880’s and 1890’s and, at least among leading theorists, the accepted thing for the rest of the period and beyond it. Böhm-Bawerk’s amendment, namely, that the real wage rate, in perfect equilibrium and perfect competition, should be equated to the Discounted Marginal Product of Labor rather than to the Marginal Product of Labor gained some votes after 1910, in the United States mainly because Taussig threw the weight of his authority into the scale. 46 We need not stay long to discuss the types of wage theories that preceded the vogue of the marginal productivity analysis, partly because most of them did not amount to much and partly because we have learned the necessary minimum about them already. 47 Suffice it then to Association, 3rd series, III, November 1902; B.Samsonoff, Esquisse d’une théorie générale de la rente (1912); the contributions on the rent of land, especially that of F.X.Weiss, ‘Die Grundrente im System der Nutzwertlehre,’ in Wirtschaftstheorie der Gegenwart (III, ed. Hans Mayer, 1928– 32); and Gerhard Otte, Das Differentialeinkommen im Lichte der neueren Forschung (1930), a work that clarifies many hazy points and incidentally establishes the emptiness of the concept of differential rents all the more effectively because the author does not mean to do so. 46 Böhm-Bawerk’s amendment, just like the analogous amendment in the case of rent, is merely a consequence of his theory of interest. How far it is invalidated by the argument from synchronization is therefore not a separate question: it is answered as soon as we have answered it for interest. But another question arises here: how far should Böhm-Bawerk’s (and Taussig’s) theory of wages be interpreted as a rehabilitation of the wage-fund theory? We may indeed arrange our terminology in such a way as to bring out a strong similarity (which Taussig was inclined to do). And we may interpret so much into Ricardo, McCulloch, and J.S.Mill as to make them ‘forerunners.’ On the whole, however, I am inclined to think that this blurs the essential lines of doctrinal development instead of bringing them out. In Böhm-Bawerk’s structure, ‘capital’ plays so different a role in relation to wages and joins forces with so many elements which the wage-fund theorists did not see that it seems more confusing than enlightening to stress the tenuous affinity that no doubt exists. In any case, if we are not content with Jevons and Rae as predecessors, it is the affinity with Senior and Marx that should be emphasized rather than the one with the wage-fund theorists properly so called. 47 The wage-fund controversy has been followed into the period under survey in our description of the issue in Part III (see ch. 6, sec. 6f). For the Marxist theory of wages, see Part III, ch. 6, secs. 2 and 6. The use of the bargaining-power element in the explanation of wages coincides of course with its use in the explanation of ‘profit’ Though the wage theorists other than the marginal productivity theorists added points of value here and there (for example they inquired with care into the relation between wages and hours on the one hand and workmen’s performance on the other; see L. Brentano’s influential contribution under that title, English trans., Hours and Wages in Relation to Production, 1894), for the most part they went on discussing the ‘classic’ problems. Two American works, however, took a higher flight. One was F.A. Walker’s Wages Question (1876), which expounded the ‘residual-claimant’ theory. The idea was really not quite new: in substance Senior had already had it (Outline, p. 185 et seq.). But Walker worked it out and propagated it in his popular textbooks. It may be best conveyed by contrast with Ricardo. Ricardo, as we know, first eliminated rent from the price problem so as to be left with ‘profits’ plus wages. Then History of economic analysis 906 recall that most of these wage theorists went on killing the wage-fund theory—some of them in the mistaken belief that they were thereby gaining a point for labor—and that, with practical unanimity, they held that wages were not paid out of capital but out of consumers’ income (George, Walker, Sidgwick, Brentano, and many others). Though this argument, as we also know, rests upon a misunderstanding of the wage-fund theory, it should be noticed that it did, in fact even though not in intent, pave the way for the marginal productivity theory. Let us cast at least a perfunctory glance on the victorious onward march of this analysis as applied to wages, neglecting all minor points. Jevons’ statement in the brilliantly original Chapter 5 of his Theory must be mentioned first. 48 Menger’s presentation is, however, fully equal to it in spite of being still more incomplete. Walras’ earlier formulation is somewhat impaired by the fact that his constant production coefficients, like Wieser’s, exclude the possibility of taking account of relations of substitution between labor and other requisites of production within each firm. Marshall established the marginal productivity analysis of wages in England, succeeding more completely than he seems to have wished. But Edgeworth’s various contributions should not be forgotten (see especially his paper on the ‘Theory of Distribution,’ 1904, republished in Papers Relating to Political Economy, vol. I). Among other things, he exploited the new catallactics for the purpose of treating special cases of wage determination. A particularly felicitous idea of his was to invoke the theory of international values in order to elucidate the relation between employers and employees—treating them by analogy with different nations that trade with one another—or between non-competing groups of workmen. Wicksteed and especially Wicksell then greatly improved the Austrian theory. The development in the United States was largely independent of the con- he went on to determine wages independently (by equating them to the minimum of existence), and profits remained as a residual (unless indeed we credit him with an abstinence theory, which we now do not do). Similarly, Walker determined the other shares in distribution independently so as to be left with wages as a residual. Opponents (Taussig, e.g.) pointed out that this clashes with the facts of the modern wage contract. But the decisive theoretical objection is in the methodology that such a theory involves, to wit in the very attempt to determine independently any elements of a system of interdependent elements. The other work was F.W.Taussig’s Wages and Capital (1896, London School Reprint, 1932). It must be mentioned here and not among American contributions to the marginal productivity theory, because in 1896 its author had not yet accepted the latter. In fact, so far as the book is concerned, he had not even noticed it: there is not even mention of the name of Thünen. Its claim to historic importance rests upon his, in great part original, attempt to graft Böhm-Bawerkian doctrine upon the ‘classic’ system. But it is recommended to the attention of the reader also for another reason: it is a masterly performance in a style of theoretical reasoning that has completely gone out of fashion. By perusing it, the reader, besides learning a lot, will be able to acquire an idea what this style was like at its best. 48 It is interesting to note, however, that it does not introduce any time discount on the marginal product of labor. General economics 907 temporaneous development in Europe. The marginal productivity theory, in a very advanced version that took full account of the relations of substitution among productive factors and came ‘close to the modern concept of a marginal rate of substitution,’ sprang ready-made from the brain of Stuart Wood, whose two papers on the subject should assure his position in the history of analytic economics: 49 ‘A New View of the Theory of Wages’ (Quarterly Journal of Economics, October 1888 and July 1889) and The Theory of Wages’ (Publications of the American Economic Association, IV, 1889). Simultaneously with the latter (i.e. in the same volume of the Publications) J.B. Clark published his marginal productivity theory of wages, ‘The Possibility of a Scientific Law of Wages.’ In 1892 appeared H.M.Thompson’s Theory of Wages. Taussig, thereby joining the ‘marginalists,’ introduced Böhm-Bawerk’s amendment into American wage theory (‘Outlines of a Theory of Wages,’ Proceedings of the American Economic Association, April 1910). For the rest, I shall confine myself to mentioning three standard works of our own epoch which are all based upon the marginal productivity theory of that period. The first is P.H.Douglas’ Theory of Wages (1934), which will have to be mentioned again as one of the boldest ventures in econometrics ever undertaken. The second—whose great merits are somewhat impaired by not quite adequate handling of the tools of theory—is J.W.F.Rowe’s Wages in Practice and Theory (1928). The third, so far as theory is concerned, by far the most significant Marshallian performance in the field, is J.R.Hicks’s Theory of Wages (1932). These stepping stones will carry the reader to the beginning of the Keynesian controversies. Since consideration of some of the more delicate questions about marginal productivity must be postponed to the next chapter, this is really all that needs to be said for the moment (see, however, the subsection on Labor Economics below). But in view of the fact that misunderstandings persist to this day concerning the nature and value of the marginal productivity analysis in its specific application to wages, the reader will perhaps condone or welcome, as the case may be, the following explanatory comments in spite of the repetitions some of them involve. First, then, let us recall what has been said above about the difference between Longfield’s and Thünen’s marginal productivity and that of Jevons and Menger. Longfield’s and Thünen’s concept is the one revived by Stuart Wood and the one that is commonly used now. The current textbook simply says that, in perfect equilibrium and perfect competition, the money wage rate of every kind of labor equals the physical marginal increment of the product due to the ‘last’ increment of labor applied (marginal product of labor) multiplied by the equilibrium price of the product. But with Jevons and Menger and also with Marshall, this was not the basic concept. Their basic concept was the increment of satisfaction individual consumers experience from that increment 49 Justice, but not more than justice, has been done to his performance by Professor Stigler (‘Stuart Wood and the Marginal Productivity Theory,’ Quarterly Journal of Economics, August 1947), to whom the statement above in quotes is due. History of economic analysis 908 of product. 50 Only a theory that uses this concept is a genuine imputation theory of wages and really should be distinguished from simple marginal productivity theories that do not use it. But both yield the same results, of course, and if we do not care for the ‘deeper meanings’ that Jevons and Menger believed the imputation theory reveals, we may deduce the usual formula for the competitive wage rate without using that concept. 51 In many important cases of applied wage theory we do not even need the usual formula but may treat the determination of the wage rate simply as a matter of supply and demand. And this is why Fleeming Jenkin must now be added to the list of the builders of modern wage theory (Part III, ch. 6, sec. 6f). He used nothing but the simple supply and demand apparatus—taking everything for granted that may be behind it—and was nonetheless able to derive important results, for example, about the possibilities of trade-union policy. But an important limitation of his considerable performance should be noticed at once, especially because it carries over to the Marshallian analysis of wages. An analysis that uses the simple demand-supply apparatus is essentially Partial Analysis, that is to say, it takes as independently given the factors that determine the demand and supply schedules. As we shall see, this is inadmissible in the case of so important an element of the economic system as is labor as a whole. To illustrate the point, let us for a moment consider the most obviously practical implication of this. So long as we operate with given demand and supply schedules which do not change when wage rates change, then we shall ordinarily have a single equilibrium rate of wages such that any increase of it creates (or increases) unemployment. And practically all economists of this period would have subscribed to the latter proposition, even for a general increase of wage rates. 52 50 Denoting by Ui the total satisfaction of consumer i, by x j quantity of the commodity j he consumes, and by L the labor that goes into this commodity, the concept in question is given by This, barring the partials, is Jevons’ expression which recurs in Marshall’s Principles. It should be re-emphasized that the marginal utilities as well as the marginal physical products involved are individual marginal utilities and marginal products of individual firms. No question of social evaluation comes in to add to our troubles, though Wieser’s and Clark’s expositions seem to suggest this. Nor need we meet any such thing as a marginal social product. This concept has indeed been introduced by Professor Pigou as a tool of his welfare economics and discussed by Professor Edgeworth (see his paper, ‘The Revised Doctrine of Marginal Social Product,’ Economic Journal, March 1925). But this was a special construction for a special purpose and has no place in the explanatory theory of wages now under discussion. 51 This usual formula (marginal physical product times equilibrium price), of course. does not apply to any cases other than that of pure or perfect competition in all factor and product markets. See below, ch. 7. 52 The specifically Böhm-Bawerkian version reads like this: if in a state of equilibrium an increase of wage rates is imposed upon the system, then another and longer ‘period’ of production becomes the most profitable one; if this is adopted, however, the existing subsistence fund suffices only for a smaller number of workmen; hence General economics 909 Second, recalling what has been said about the formal character of the marginal productivity theory, let us ask ourselves how far this theory provides a ‘causal’ explanation of wage rates. On the one hand, it is clear that in order to enable it to explain any particular level of wages that we observe in any given place at any given time, it is necessary to feed into it the particular facts of that place and time; and these facts, such as the available amounts of complementary factors, and not the margins of productivity may then be called the true or ultimate causes of that wage rate. On the other hand, it is equally clear that wage rates, being elements in a system of interdependent magnitudes, are simultaneously determined with all its other elements so that even in pure theory— that is, irrespective of the facts of any particular case—they cannot be said to depend upon a margin of productivity as if this were an ultimate datum. However, this is all that Marshall can have meant when he wrote that wage rates are determined at the margin and not by the margin. But this argument only parallels Marshall’s argument about marginal utility—about the three balls resting against one another in a bowl—and admits of a similar reply. 53 In any case, it does not reduce the value of the marginal productivity theory as a tool for solving wage problems. 54 Third, he who wants to use the marginal productivity of labor as an explanatory principle and as a tool for solving wage problems must, of course, understand it and acquire some experience with it. If he fails to fulfil these conditions, difficulties will crowd upon him which, human nature being what it is, he will turn into so many objections, especially if, suspecting apologetic traps, 55 he dislikes the theory in the first place. But for the period under discussion there was an excuse for this. The theory was not only not developed in a manner that would have shown its usefulness—as it stands out, for example, in Hicks’s Theory of Wages (1932)—but in many cases it was also faultily formulated. Some economists even had difficulty in seeing the difference between the marginal product of labor and the product of marginal (least efficient) labor. Others seem to have believed that the marginal produc- the rest become unemployed. Observe that this argument far transcends the simple supply and demand argument; and also that it is intended to hold only for an imposed increase of wages and not for one that results from an increase in the subsistence fund. 53 Its formulation is left to the reader as a useful exercise. 54 See, e.g., the treatment of the wage-minimum problem in Pigou’s Wealth and Welfare (1912). 55 It is hoped that it is unnecessary to go into this again. I should perhaps add, however, that the large majority of economists who defend increases in monetary wage rates or resist reductions have very little to fear from admitting the correctness of the theory. For their arguments will in general be based upon assertions of fact that have nothing to do with it. Few if any will care to espouse arguments that really conflict with it so soon as the common-sense complement of it is brought home to them in each individual case. Moreover, the fact that the use of the marginal productivity theory is a proposition about a rate of wages that would prevail in perfect equilibrium and perfect competition in itself suffices to show the vast expanse of territory the theory leaves uncovered. tivity theory of wages breaks down if increased wage incomes or reduced hours increase the efficiency of labor. 56 Fourth, in consequence of this we find that many labor problems continued to be treated by means of the tools that had served the ‘classics.’ This holds in particular for the machinery problem. It received plenty of attention but analysis rarely rose above the old arguments pro and con the ‘compensation theory.’ Such as it was, this discussion on History of economic analysis 910 technological unemployment provides, however, one of the answers to the Keynesian indictment that the theorists of that period knew of no unemployment other than ‘frictional’: for technological unemployment, even if essentially temporary so far as the effects of any individual act of mechanization is concerned, may evidently become a permanent phenomenon through being incessantly re-created. 57 The purely theoretical question of full employment in perfect equilibrium and perfect competition will be considered in the next chapter, and nothing need be added to what has been said before on the supply of labor. 6. THE CONTRIBUTION OF THE APPLIED FIELDS * We have repeatedly noticed that the economists of the period, or most of them, approached questions of economic policy, or many of them, in a new spirit. In this section we shall not dwell on this fact again, but rather hunt for the contributions to analysis that resulted from their preoccupation with practical questions. In all cases, these preoccupations advanced scientific knowledge mainly by increasing our command over facts. Gains for our analytic apparatus, though of course not absent, were much smaller than they might have been. We shall briefly survey the more promising fields (except money and cycles, which are considered in ch. 8). But we shall not consider developments in the field of business economics (business administration, Privatwirtschaftslehre), including accounting and ‘actuarial science.’ It has been emphasized from the first that there is really no better reason for separating it from general economics than that a large 56 The theory of substitution, Marshall’s exposition of it notwithstanding, was far from being common property even by the end of the period. This is the reason why Charles J.Bullock’s paper on ‘The Variation of Productive Forces’ (Quarterly Journal of Economics, August 1902) in spite of various shortcomings deserves to be put on record as a major contribution. 57 We are free, of course, to define the concept of frictional unemployment so widely as to include technological unemployment and also the other types of unemployment that were recognized— mainly: unemployment from imperfections of competition; unemployment from monetary causes; and unemployment from business fluctuations, whatever their cause—but then the indictment loses its force for, thus defined, friction is no longer an obviously inadequate explanation of the observed facts of unemployment. In particular, the indictment should not have been directed against Pigou’s Theory of Unemployment (1933). For the period, see especially W.H.Beveridge, Unemployment (1909). * [The following section was unfinished and untyped at the time of the death of J.A.S.] General economics 911 . for economic and for sociological analysis of the wide range of variation of the ‘natural’ abilities of men. All I mean is that the theory of rent contributes nothing to our understanding of. Bailey, see above, Part III, ch. 4, sec. 3c; on the interpretation of entrepre neurial gains as rents of differential ability, see above, sec. 2b.] History of economic analysis 902 opinion,. cases of the services of capital goods and of labor—a fact that seemed important for questions of welfare economics and of taxation. Marshall, shifting emphasis to this surplus aspect of rent,