History of Economic Analysis part 117 docx

cism of its policies but also of any display of lack of sympathy with the party philosophy. It promoted party members and demoted Jews. If it did not insist on acceptable professions of faith, it welcomed them. In individual instances, the party or groups within the party and even the authorities went much further than all this implies. And in addition, the disturbing effects of the general conditions then prevailing upon research and study must be taken into account. However, though less so than in the field of the physical sciences, the bulk of professional work went on. In particular, nobody would have got into trouble in consequence of having worked out new theoretical or statistical tools. A work like Keynes’s General Theory could have appeared unmolested—and did. 4 It must not be forgotten that the creed of National Socialism was not primarily or essentially economic and that, hence, it was compatible not only with all kinds of technical economics but also with the advocacy of widely differing policies. 2. ITALY In Italy we find a similar situation, only much more pronounced. The Fascist regime resented criticism of its measures as much as or, since individual policies were much more closely associated with the leader personally, still more than was the case in Germany. 5 It also insisted either on a sympathetic attitude of economists or else on neutrality—perhaps the best way of expressing the situation is to say that what the government insisted on was absence of active hostility to Fascist principles. A few leading men—such as Ricci and Bresciani-Turroni—expatriated themselves, but most were not seriously disturbed. Purely scientific work was not interfered with at all. 6 Under these 4 See Carl Föhl, Geldschöpfung und Wirtschaftskreislauf (1937). The author tells us in the preface that his manuscript had been completed in December 1935. All the more interesting is the far- reaching though not complete parallelism between his argument and Keynes’s (see below ch. 5). 5 Hitler left individual measures, and especially economic measures, to his lieutenants, who stayed in positions of leadership for relatively long spells and were allowed to acquire reputations with the public and to develop policies of their own. Mussolini did not permit this. In consequence the economic policies of Fascism, even as regards details, tended to become in the public eye his own personal measures. 6 It is important to emphasize that even in treatises that took a professedly sympathetic attitude to the città corporativa the analytic parts did not differ from generally accepted economic doctrine and could have been written just as well by enemies of Fascism. As an example I mention Professor Luigi Amoroso’s Principii di economica corporativa (1938). The first two parts deal respectively with the theories of money and of equilibrium and are completely free from political implications, Fascist or other. Only the third part develops what might be termed the economic philosophy of Fascism—much of which, as formulated by Amoroso, would command the hearty approval of the modal American economist. As another example, I mention a course of lectures on economic policy or political economy that did not keep to the ivory tower of what we call here ‘purely scientific work’ and still displays not much constraint: Professor Giovanni Demaria’s La politica economica dei grandi sistemi coercitivi (1937). Economics in the ‘totalitarian’ countries 1122 conditions, scientific economics continued to move at what we have seen in Part IV was a high level, both within and without the Pareto school, until the war. Barring war effects there was no break and neither was there one after the fall of the regime. 3. RUSSIA But the case of Russian economics 7 in the Stalinist period differs from the German and Italian cases, not in degree but in kind. In the decade preceding that period—roughly between 1917 and 1927—this was not quite so. Opponents of the Soviet regime or even neutrals were indeed dealt with much more ruthlessly than were opponents of the National Socialist or Fascist regime. Scientific research itself, not only discussion of policies, was regimented in a manner unheard of in Germany or Italy, not only because of the nature and methods of the bolshevist administration but also because of two other reasons that contradicted and, nevertheless, reinforced each other. On the one hand, the Soviet creed, ideologically at least, was essentially an economic creed and slight deviations from the holy books, even if of a purely theoretical kind, acquired an importance that is difficult for us to understand; on the other hand, the bolshevist government very naturally exploited to the full the naïve emotionality of the ‘revolutionary people,’ who necessarily believed that, the millennium having arrived, there were no longer such things as ‘economic laws’ and hence no need for any economic analysis at all. In this situation discussion tended to be geared exclusively to the momentary wishes of the men who were, or were believed to be, at or near the helm, and such arguments as that a certain view was ‘reactionary’ or ‘leftist’—in fact, pure denunciation—began to replace scientific views. Nevertheless the break was not complete. Conversion to Soviet orthodoxy was made easier by the facts that Marxism, now the prescribed creed, had had a strong hold on Russian economists even before 1917 and that there is plenty of room for scientific analysis within the limits set by allegiance to its principles. So long as genuine Marxist stalwarts like Bukharin played some role, we might have still to state that, qualitatively more than quantitatively, genuinely analytic work was at an ebb, but there would be no reason to question that there was genuinely scholarly work—as the mere existence of the Marx-Engels Institute suffices to show. However there were other institutes, for example one for research in agricultural economics and another for business or cycle research, which for the time being enjoyed some freedom not only in collecting but 7 I repeat that I do not know Russian. The following remarks are based upon (a) such Russian works as are available in languages I know, especially English and German; (b) conversation with colleagues who do know Russian but must not be held responsible for the impression I received from it; and (c) secondary literature of very unequal value, of which most, though not all, has an anti-bolshevist bent. I mention only the item that I have found more useful and, so far as I can judge, more correct than any other: A.Zauberman, ‘Economic Thought in the Soviet Union,’ Review of Economic Studies (vol. XVI (1), 1948–49 and vol. XVI (2) and (3), 1949–50). [J.A.S. had apparently read only the first of the three articles by Zauberman.] History of economic analysis 1123 also in interpreting economic data. Kondratieff’s work, already mentioned, caused a great stir 8 and constitutes, so far as I can make out, the peak performance of the work produced by a considerable number of competent economists (Perwuschin, Oparin, Sokolnikoff, and others); this work, in spite of the sinister implications of the fact that some of the authors have not been heard of since, may be taken as proof that serious economics survived until the rigors of the Stalinist regime fully asserted themselves. Then the break occurred after all, and teaching as well as the work of the Institute of Economics of the U.S.S.R. was more and more reduced to descriptive treatment of the practical problems of the Soviet government and to mutual recriminations of slaves incessantly in fear for their lives. 9 We confine ourselves to two points that promise better things for the future. First, Soviet Russia inherited from pre-Soviet times an excellent tradition of work in statistical methods and their mathematical—mainly probabilistic—background. This work, much less exposed than economics to political attack, survived and continued to produce internationally recognized contributions. Second, it is evidently impossible to ‘plan’ investment without developing an apparatus, however primitive, for comparing alternative methods of carrying it out, even if the purpose itself is given by dictatorial command, and for comparing alternative investment purposes if there is some freedom of choice. But into any attempt to do so, actuarial norms and the concepts of value, marginal productivity, and interest enter by logical necessity. The task of Soviet economists was and is not to improve these concepts but to smuggle them in in such a manner as to hide their fundamental identity with the corresponding ‘capitalist’ concepts. 10 Advance 8 N.D.Kondratieff’s long-cycle theory has been published in several books and articles, some of which are available only in Russian. An abbreviated translation by Professor W.F.Stolper of a German article that Kondratieff himself believed to give the gist of his theory can be found in the Review of Economic Statistics, November 1935, and a survey by Mr. George Garvy of the controversy that arose about it (‘Kondratieff’s Theory of Long Cycles’) in the same journal, November 1943. This survey shows very well, on the one hand, the venomous ferocity with which controversies were carried on in that atmosphere and, on the other hand, the facts that scientific points of view were not absent and that scientific work was still possible. Kondratieff was exiled to Siberia in 1930. 9 The bulletins of the Academy of Sciences of U.S.S.R. (Department of Economics and Law) contain comments on this state of things (though not in the terms above) that waver between condemnation and recommendation of a sterilized Marxism. But it seems questionable whether analytic work of the kind that is being turned out, with its childish attempts at rediscovering elements of economic logic without running into heresy, is really preferable to total suspension of such work. The importance of these attempts seems to be greatly overestimated by C.Landauer, ‘From Marx to Menger,’ American Economic Review, June 1944. Also see John Somerville, Soviet Philosophy (1946). 10 An almost pathetic example has been presented by Holland Hunter, ‘The Planning of Investments in the Soviet Union,’ Review of Economics and Statistics, February 1949. Professor Hunter there translated a chapter of a textbook by Professor Khachaturov on the Economic Principles of Railroad Transportation (1946), and ana- Economics in the ‘totalitarian’ countries 1124 on this line is difficult and slow owing to the incessant threat of denunciation that lurks behind the unfavorable reviews that publications of this kind seem to have received so far. However, there is in these publications some promise for the future, especially because it is safe to predict that denunciations of this type will go out of fashion: bolshevist economists are bound to discover in the end what Pareto and Barone realized half a century ago, namely, that there is an economic logic that has nothing specifically ‘capitalist’ about it. Nor is this all. National income accounting and budgeting techniques are rapidly developing in non-bolshevist countries—they can hardly be called ‘capitalist’ any more—and traditional economics will have to adapt itself to them. [J.A.S. intended, but did not actually write, a section on national income accounting in this concluding Part] These techniques and corresponding methods of analysis are still more obviously needed in the Soviet state. There thus exist two tendencies that, born of similar needs, have begun to assert themselves independently in Russia and elsewhere, especially the United States, and now tend to converge—as do so many others. Still, beyond this there is nothing to report for a history of economic analysis, and only he can wonder at this who, even at this point, has not grasped the purpose of this book. 11 lyzed the Coefficient of Relative Effectiveness of Investment under the aspect mentioned in the text in a most instructive manner. The reader finds other examples, though less completely worked out, in the [first of the three] articles by Zauberman, who gives prominence to the work of S.G.Strumilin. He also mentions (vol. XVI (1), p. 3n.) an early attempt made by Mrs. B.Khmielnitskaya to secure room for economic theory by defining it—quite sensibly—as the science of the norms of rational management of a socialist (‘organized’) society. It is amusing to note that in doing so she seems to have adapted for her purposes the German von Gottl’s concepts of idiography (the description of individual facts, ‘ideographic’ must be a misprint) and nomothesis (the generalizing statement of laws, which she changed into ‘normography,’ wisely replacing von Gottl’s ‘nomos’ by ‘norm’). 11 One of the causes of the scientific sterility of Russian economics in the period under survey was, of course, the fact that some of the rulers, in particular Lenin, Trotsky, and even Stalin, wrote so extensively and authoritatively on questions that ordinarily belong in the sphere of professional economics. Therefore even a reader who does grasp the purpose of this book and who knows how to distinguish economic anal ysis from political economy might still object that I do not report on the works of those three men or at least on the voluminous works of Lenin. The answer has been given already: however great their historical importance in other respects, as contribu tions to economic analysis they are negligible. History of economic analysis 1125 CHAPTER 4 Dynamics and Business Cycle Research LET US RECALL once more that here, as throughout this book, Dynamics means exclusively analysis that links quantities pertaining to different points of theoretic time— in the sense that has been repeatedly explained before—and not the theory of evolutionary processes that run their courses in historic time: it is practically coextensive with sequence analysis and includes period analysis as a special case, but it is not coextensive with the theory of economic growth or development, or ‘progress.’ 1 Thus defined dynamics is a genuinely new departure. We have indeed seen at various turns of our way, particularly in the case of Sismondi, that dynamic considerations in our sense intruded into economic analysis times out of number, chiefly by implication but also explicitly. But the exact core of economics was nevertheless static and was believed to constitute a self-contained body of doctrine, a body moreover that embraced all or almost all of the essential insights. This is obvious in the case of Walras. But it also applies to Marshall. 2 He no doubt added plenty of extra-static considerations, chiefly about growth but also about sequences, so much in fact that he may be said to have posited the task of future dynamic theory (see e.g. Principles, p. 519), just as he posited the task of future econometrics; but though he presented material, viewpoints, and desiderata, he did not cross the Rubicon. For the rest, we have noticed the suggestive pointers of Pantaleoni and Pareto, but there was no advance toward the goal to which they pointed. By the phrase, ‘crossing the Rubicon,’ I mean this: however important those occasional excursions into sequence analysis may have been, they left the main body of economic theory on the ‘static’ bank of the river; the thing to do is not to supplement static theory by the booty brought back from these excursions but to replace it by a system of general economic dynamics into which statics would enter as a special case. The realization of the fact that even a static theory cannot be fully developed without an explicit dynamical schema 1 This is repeated here because many modern authors do identify dynamics with the theory of growth. The chief authorities for our terminology are Frisch and Hicks. Prominent examples of another terminology are Harrod (see especially Towards a Dynamic Economics, 1948) and Stigler. An intermediate position is held by many, e.g. by Charles F.Roos (Dynamic Economics, 1934). I also repeat that my insistence on this distinction is due not to any wish to quarrel about words but merely to a wish to avoid confusion. 2 But not to Böhm-Bawerk, see above, Part IV, ch. 6, sec. 5. (Samuelson), 3 which we have noted before, is a first step in this direction, and if space permitted, a few others could be mentioned. 4 However, no attack on the whole front of Walrasian theory has as yet developed and the analogy with a building plot is still painfully apposite: an increasing number of workers see the new goal; but for the time being this is practically all, since H.L.Moore’s effort did not go substantially beyond comparative statics. More positive success has attended the efforts to ‘dynamize’ aggregative theory. [1. DYNAMIZING AGGREGATIVE THEORY: MACRODYNAMICS] This is understandable. On the one hand, aggregative theory that reduces truly innumerable arrays of variables to half a dozen or even less can evidently stand up much better than could a Walrasian system under the complications that are inseparable from even the simplest dynamic schema. As an illustration, consider so simple a dynamizing device as the introduction of lags. Offhand and until more powerful methods are invented than are available now, there is very little we can say when we give different time indices to all the quantities that enter the Walrasian system, except that it becomes unmanageable thereby. But this is no longer so if the only variables we have to take care of are ‘consumption,’ ‘investment,’ and a national income that is identically equal to current consumption plus current investment. Suppose we postulate arbitrarily that the consumption (C t ) of some period (t) equals a constant proportion (α) of the income of period (t−1), αY t−1 ; and that the investment (I t ) in period t equals a constant proportion (β) of the difference between current consumption and the preceding period’s consumption, β(C t −C t−1 ) or β(αY t−1 −αY t−2 ). Remembering that Y t ≡C t +I t , we get as, on the strength of grammar-school mathematics, it is easy to see: 5 Y t =α(1+β)Y t−1 −αβY t−2 . This is a homogeneous second order difference equation with constant coefficients, which is very easy to solve by an elementary technique that lies ready at hand and yields certain economically interesting results. The tempta- 3 See Foundations, Part II, especially ch. 11. I take the opportunity to submit that there and in Appendix B, also in ‘Dynamic Process Analysis,’ his contribution to A Survey of Contemporary Economics (H.S.Ellis, ed.), Professor Samuelson has performed a most meritorious pedagogical task; there is no better introduction to the meaning and techniques of modern dynamics. 4 Pressing considerations of space are, however, not the only reason for refraining: on the one hand, I do not wish to blur the main contours by details; on the other hand, I do not wish this sketch to degenerate into a bibliography. Going over the volumes of Econometrica is almost all that the reader needs in order to find the ropes. 5 This is (the gist of) the Hansen-Samuelson equation; see P.A.Samuelson, ‘Interactions between the Multiplier Analysis and the Principle of Acceleration,’ Review of Economic Statistics, May 1939, reprinted in Readings in Business Cycle Theory (selected by a Committee of the American Economic Association, G.von Haberler, chairman, 1944). Dynamics and business cycle research 1127 tion to avail oneself of so tremendous a simplification is almost irresistible and impervious to the objections that might be raised on theoretical grounds. 6 No wonder, then, that the early 1930’s were fertile in such aggregative schemata—R.Frisch’s macrodynamics. 7 Not all of them were mathematically exact of course—a fuller survey would have to mention several important ones that were presented by non-mathematical economists such as Professor von Hayek. It should be carefully observed that this drive toward macrodynamics was in itself quite independent of any wish for a closer alliance of economic theory with statistical figures: macrodynamics would have asserted its claims even if theorists’ attitudes to statistics had not changed at all as compared with the preceding period, and as a matter of fact several writers who displayed no symptoms of such a change in attitude were just as anxious as anyone can be to secure the advantages of aggregative simplification. [2. THE STATISTICAL COMPLEMENT: ECONOMETRICS] But, on the other hand, the equally strong drive toward a numerical economics, an economics that would be statistically operational, is also a dominant factor in our scientific situation. And this factor, however independent of the desire for a simplification of the pattern of economic theory per se, also favors macrodynamic methods. For with few exceptions the aggregative variables—particularly if their number be augmented by price levels and interest rates—are easy to identify with our most important time series. As an outstanding example, which displays both tendencies closely united and which constitutes so important an element in the economic research of our time that it cannot be omitted from any sketch of it, however brief, I mention the work of Tinbergen. 8 His numerous aggregative schemata, most of which use many 6 It is convenient to defer consideration of these objections and of an important qualification that should be added to them. But let us note at once that our example also illustrates the fact that economists are prone to yield to the further temptation to improve upon the situation by introducing additional simplifications: in our example, it is not only the reduction in the numbers of variables which simplifies things but also the postulate that the coefficients are constant; if they were not, the equation would not be so easy to handle. 7 We mention here only one example, namely Professor Frisch’s own schema presented in his powerful paper, ‘Propagation Problems and Impulse Problems in Dynamic Economics,’ Economic Essays in Honour of Gustav Cassel (1933). The reader will find many others in Tinbergen’s survey article quoted in the next footnote. 8 In the long list of Professor J.Tinbergen’s publications the one that is perhaps most suitable to serve as an introduction, to American and English readers, of his theoretical and statistical methods is his Statistical Testing of Business-Cycle Theories: I, A Method and its Application to Investment Activity; and II, Business Cycles in U.S. 1919–32 (League of Nations, 1939). Still more useful as a survey of what may be called now the earlier work in dynamics is his article ‘Suggestions on Quantitative Business Cycle Theory,’ Econometrica, July 1935. Both titles are unduly unassuming. Just now the reader should ignore the specific reference to business cycle research—the reason for which will be explained presently—and accept the first as a treatise on, and the latter as a survey of, general dynamics. History of economic analysis 1128 more variables to start with than do those of other authors, are in the first instance set up on the basis of purely theoretical considerations that are extremely simple—so much so that it is perhaps more enlightening to speak of common-sense considerations: they embody, in a system of (almost always) linear equations with constant coefficients, the definitions of obviously important aggregates (definitional equations); the relations that common sense suggests should subsist between them (balance equations); and relations that are supposed to describe the behavior of classes of households and firms (behavior or ‘decision’ equations). 9 This involves the fundamental principle that construction of the theoretical set-up should precede the statistical work: the relations themselves are not suggested by statistical observations; they are postulates and not results. 10 Statistical figures are to ‘explain’ the numerical values of some variables by given numerical values of others by the method of multiple correlation—a process which also eliminates those ‘explanatory’ variables whose partial regression coefficients indicate the insignificance of their influence. The system is then, by process of successive substitutions, reduced to ‘final’ equations that are held to depict the economic mechanism. 11 In itself, every step in this procedure is open to serious criticisms, about which no more can be said than that they should not blind us to the greatness of this pioneer effort. Since most of these criticisms are of a statistical nature, the statistical work of Frisch—partly taken account of by Tinbergen—and his group should at this point be mentioned again, particularly the work of Haavelmo, who during his brief sojourn in the United States, without holding a teaching position, exerted an influence that would do credit to the lifetime work of a professor. 12 In any case, however, the economist who accepts macrodynamics as it stands, with or without its statistical complement, may speak of conquest already achieved—and not only of a developing attack and of the increasing clearness of a goal, which is all that we are able to record in the matter of dynamizing the Walrasian or Paretian system. [3. THE INTERACTION OF MACRODYNAMICS AND BUSINESS CYCLE RESEARCH] Exactly as macrodynamics has been and is being propelled by the specifically econometric drive—the tendency toward reasoning in terms of statistical figures—so both the theoretical and the numerical components of macrodynamic work have been propelled by the preoccupation with business cycle 9 For examples, see the work cited in the preceding footnote. It should be remembered that Tinbergen’s publications of this type begin (so far as I know) in 1934. 10 This is the fundamental difference between the methods of Tinbergen and those of W.C.Mitchell, whose methods will be touched upon below. 11 I feel in duty bound to offer my apologies to Professor Tinbergen for this sort of report. But I hope that he and the reader will prefer even these jejune sentences to the bare reference to his works which not every reader can be trusted to follow up. 12 The bulk of his teaching has been embodied in a number of papers published in Econometrica. But see especially Trygve Haavelmo, ‘The Probability Approach in Econometrics,’ Supplement to Econometrica, July 1944 (Cowle’s Commission Papers, New Series 4). Dynamics and business cycle research 1129 problems. This preoccupation is, as we have already seen, a salient characteristic of our time. From the foregoing analysis of the factors that produced macrodynamics and in particular statistical macrodynamics, we may infer that this development would have occurred even if there were no such thing as the particular kind of fluctuations that are commonly identified as business cycles. From what has been said earlier in this Part and in Part IV we may infer that preoccupation with the phenomena of business cycles would have increased, as compared with the times before 1914, even if modern macrodynamics had not emerged. But it is obvious that both developments were bound to reinforce one another and that, on the one hand, the methods, materials, and results of business cycle research encompassed more and more of general economics and, on the other hand, the methods, materials, and results of modern macrodynamics evolved principally with a view to serving business cycle research, 13 so much so that reference to business cycles intrudes even into the titles of many macrodynamic publications of much wider range. 14 It is now easy to formulate more precisely the nature and results of this interaction. We have seen in Part IV, Chapter 8, that all the fundamental ideas concerning the phenomenon of business cycles were present before 1914. 15 What our period added, besides critical development of these ideas, was in the first instance the new wealth of data and new statistical methods of handling them. Even the econometric program, barring the ‘higher’ mathematics, had been carried out by such outstanding students as Juglar, Mitchell, and Spiethoff. 16 But incomparably greater possibilities have offered themselves since 1919. Some writers were content to use whatever figures the arms of their analytic apparatus could grasp. An outstanding example is Professor Pigou, whose Industrial Fluctuations (1st ed. 1927), though remaining a ‘theoretical’ work, nevertheless, owing to the new material, differs greatly from the kind of work that an economist of the same type would have produced before 1914. Others displayed a tendency to plunge into the statistical material directly and to 13 A simple way for the reader to realize this is to glance over Professor H.M. Somers’ ‘Classified Bibliography of Articles on Business Cycle Theory,’ appended to the volume of Readings in Business-Cycle Theory, to which reference has been made already, or to some of the other bibliographies mentioned there (p. 444), particularly the one by Professor R.A.Gordon. 14 This is why, speaking of macrodynamics per se and not intending to speak of business cycle research specifically, I had, nevertheless, in a previous footnote, to quote two of Tinbergen’s works whose titles carry this connotation. The reason why, with what might seem uncalled-for pedantry, I insist on this point is simply that it is essential to a correct diagnosis of the modern scientific situation. 15 This also applies to Professor von Hayek’s theory, if it is permitted to link it up with that of Professor von Mises. If it is not, I apologize. 16 Spiethoff’s preliminary presentation of his business cycle analysis as a whole did not, it is true, appear before 1923, and his comprehensive work, as well as its translation into English, is still expected. But this delay was and is caused by the heroic attempt to master vast materials single- handed. As regards Mitchell, the reference above is to his book of 1913. History of economic analysis 1130 scrap the existing apparatus as well as the existing explanatory hypotheses. We may illustrate this tendency by two instances that in other respects have very little to do with one another, namely the work of the Harvard Committee (W.M.Persons) and the work of Mitchell. The Harvard University Committee on Economic Research, presided over by Charles J.Bullock and chiefly directed by Warren M.Persons and W.L.Crum, embarked upon extensive historical-statistical investigations and developed important time series, but owes its international renown—its methods being discussed, copied, and further developed almost everywhere, especially by E.Wagemann’s Berlin Institute—to the ‘three-curve barometer,’ a revised version of which the reader finds authoritatively described in the April 1927 number of the Review of Economic Statistics (‘The Construction and Interpretation of the Harvard Index of Business Conditions’). No analysis of its method is possible here. We must confine ourselves to indicating the fundamental principle and to adding three remarks which the reader is urgently requested to bear in mind. The principle is to correlate time series which common sense indicates to be particularly relevant, after having ‘eliminated’ from them seasonal variations and the ‘secular trend,’ so that cycles are given as a residual (for details see W.M.Persons, ‘Correlation of Time Series,’ Rietz’s Handbook of Mathematical Statistics, 1924, ch. 10). The remarks I wish to add are these: (1) The statistical methods used by the Harvard Committee are in the light of later, and even of contemporaneous, developments of ‘higher’ statistics exposed to serious objections. But this must not induce us to overlook the impulse that both further compilation of statistical figures and the development of statistical method derived from that pioneering venture; or to overlook the fact that there was a rough common sense about those methods that would go some way toward justifying their results as approximations, if anyone cared to undertake the task. (2) If critics erred in failing to give due weight to the historical importance of that venture, they erred still more in that part of their criticism which was directed against the forecasting value of the barometer. The fact is that the barometer curves indicated the approaching break in 1929 clearly enough—the trouble was that the interpreters of the curves either would not believe their own methods or else would not take what they believed to be a serious responsibility in predicting depression. (3) The constructors of the Harvard Barometer emphasized for the benefit of their readers and also believed themselves that they were not using any of that discredited and discrediting monster, economic theory. Professor Persons was quite prone to reply to theoretical objections by pointing to the hundreds of correlation coefficients that had been figured out under his direction. As a matter of fact, however, they did use a theory that was all the more dangerous because it was subconscious: they used what may be termed the Marshallian theory of evolution. That is to say (if we neglect the important but in this connection secondary correction for seasonal variations, one of their most lasting contributions) they assumed that the structure of the economy evolves in a steady or smooth fashion that may be represented (except for occasional changes in gradient, ‘breaks’) by linear trends and that cycles are upward or downward deviations from such trends and constitute a separate and separable phenomenon. This is an error which we shall have to mention again presently. But though erroneous, this view constitutes a theory, or the backbone of one. The little methodological controversy on the subject of Dynamics and business cycle research 1131 . cism of its policies but also of any display of lack of sympathy with the party philosophy. It promoted party members and demoted Jews. If it did not insist on acceptable professions of faith,. the work of the Institute of Economics of the U.S.S.R. was more and more reduced to descriptive treatment of the practical problems of the Soviet government and to mutual recriminations of slaves. points of view were not absent and that scientific work was still possible. Kondratieff was exiled to Siberia in 1930. 9 The bulletins of the Academy of Sciences of U.S.S.R. (Department of Economics