Knowledge that Transforms

THE purpose of the following study is twofold: to examine the nature of the cycles in interest rates and to criticize the application of the periodogram method to economic statistics. As for the particular series under investigation, it may be said at once that no uniform periodicity is revealed. If we confine the analysis to those portions of the entire time interval which are fairly free of extreme deviations, we discover however a cyclical movement which conforms moderately well to the average found by the periodogram. This average cycle has a length of about 40 months, and its form may be described as that of a sine curve distorted so that the time interval from low to high considerably exceeds that from high to low. It is found that, because of the presence of a marked seasonal variation and of irregular extreme deviations in the economic statistical record, the blind application of the periodogram method is likely to yield results which are illusory. Even after the results are checked by the usual tests available for the criticism of the findings of a periodogram analysis, a thorough examination should be made of the actual cycles found by eliminating the seasonal variation from the original data. It is probable that the summation process involved in the periodogram method will be of considerable value in studies of the length and form of the cycles in those portions of a statistical record which are clearly free of irregular extremes, but it is very doubtful whether the attempt to analyze the form of the cycle by the Fourier method will assist in describing or understanding the cycle. The average form of the cycle is at best not very precisely determined, and its breaking up into a group of harmonic components of differing periods and amplitudes is certainly bewildering and probably furnishes no true insight into the nature of economic fluctuation.

THE Index of Trade, presented in the folded chart opposite this page, is designed to give a view of the combined fluctuations of trade, transportation, manufacturing activity, and industrial employment in the United States, month by month since I903. The Index is based upon representative statistics, but certain statistics are not available throughout the entire period and others, being expressed in terms of dollars, do not accurately reflect fluctuations in the physical volume of trade since the war. Consequently, it has been necessary to use somewhat different selections of statistics for the intervals I903-I5,9I5-I9, and I9I9-23, and all statistics expressed in terms of dollars have been excluded since I9I4. The actual overlapping curves for I9I5 and I9I9 are shown on the chart, the initial years of the intervals I9I5-I9 and I9I9-23 being plotted as dotted lines. All the statistical series utilized throughout were adjusted for long-time trend and seasonal influences before averages were taken, so that the resulting index depicts the percentage deviations from normal of the combined figures for trade, transportation, manufacturing activity, and industrial employment. The Index is an average of the corrected figures of diverse statistical series, such as bank clearings outside New York City, the value of imports of merchandise, gross earnings of leading railroads, production of pig iron in tons, and the relative number of wage earners employed in industrial establishments. The justification for combining such diverse series into a single average is twofold: economic and statistical. The economic reason for averaging the series is that together they reflect wholesale and retail transactions, domestic and foreign trade, the volume of transportation of all classes of goods, the activity of railroads, the volume of manufacture, and the purchasing power of wage earners. In other words these particular series are selected from the available data because they are themselves indices of economic conditions in trade, transportation, manufacture, mining, and agriculture, and then they are combined because, taken together, they touch business at many more points than does any one series alone. The statistical reasons for combining the series are: First, the major movements accompanying the ebb and flow of industrial activity of these series are very similar as to the timing of advance and recession2 and, therefore, the average fluctuates in a manner representative of that of each constituent series. Second, the minor irregular variations accompanying strikes, transportation congestion, and the like of the various series are very much moderated when a number of corresponding items are averaged. The average, therefore, may be expected to soften or iron out minor variations and to throw into relief the general ebb and flow of trade. For the period since the outbreak of the war in I9I4, as has been said, we have not included in the Index of Trade statistical series expressed in terms of dollars bank clearings, for instance -although we did include such series previous to I914. The reason for this difference of treatment of the two periods is to be found in the nature of price fluctuations before and after the war: previous to I914 fluctuations in the dollar amount of transactions reflected (approximately, not exactly) corresponding fluctuations in the physical amount of trade, but after I9I4 the violent rise and fall of commodity prices destroyed this correspondence. It is possible to utilize as the material for a continuous index of the physical volume of trade for the last eight years of violent price changes only such series as are expressed in physical units. For the intervals I9I5-I9 and I9I9-23, therefore, only such series are selected for the Index. In the choice of the current group of series still another criterion has been used: since promptness is essential, only those data are utilized which become available for each month by the middle of the following month.

Q UANTITATIVE analysis is disclosing the I fundamental significance of fluctuations in the physical volume of production. Earlier explanations of overproduction and underproduction were of a general and abstract character. For the most part, they dealt with the surface of things. Without interpretation, they charged overproduction when production was at a minimum, and underproduction, when production was at a maximum. Such blanket charges meant little. A more intensive and objective examination of production was needed. With the recent rapid increase of available statistical data and notable improvement of statistical methods, such investigation has finally become practicable. Much about cyclical fluctuations is to be learned from the records of any important line of production agriculture, forestry, mining, manufacture and from data given altogether in annual terms. But a large part of the ground has already been covered. The yearly records for agriculture (I879-I920), mining (I879-I9I9), and manufacture (I899-I9I9), were analyzed at length in the issues of this REVIEW from September, I920, to January, I92I; and in the REVIEW for July, I922, this analysis of the annual data was broadened and extended through I92I for agriculture and mining.' The more recent yearly records for manufacture still remain to be studied. More important still, the monthly production data now available in quantity call for comprehensive analysis. The present study will concern itself primarily with these monthly data.2 The monthly records of manufacturing output show better than any other data the real nature of cyclical changes in the volume of production. In the first place, agricultural data are inadequate because agricultural production assumes a yearly form, and records available only at yearly intervals do not give a clear picture of the character of the business cycle.3 During a single year business may pass through at least two distinct phases of the cyde. The opening of the year may find industry proceeding at top speed, and the close of the same year find it staggering into depression. A total or average for the year will disclose neither of these conditions. True, a general notion of ups and downs in the physical volume of production may be gathered from annual data, but the details of the picture are lost. Returns for shorter periods are necessary. Quarterly data hardly go far enough in this direction; weekly go farther than is necessary. Monthly data effect a satisfactory compromise and serve to throw all features of the business cycle into bold relief. From the nature of the case, then, data on agricultural production will not suffice. To secure monthly data, resort must be had to the records of manufacture and mining. Records of manufacturing output have the further advantage an advantage which they hold over the data of mining as well as agriculture of affording a recapitulation of the results

ships, nor have we ever intended to imply that they do. Lack of information even among the authorities on the subject as to the latent merits of employment data and the cyclical behavior of employment itself created a situation which called forth the emphatic statement of those points. Since I92I, employment statistics have received much attention. Numerous writers and observers of business conditions have come to use employment data of various kinds, often expecting too much from them and using them erroneously. Indeed the significance of employment fluctuations cannot fully be realized except by studying their relations to those of other economic factors. The purpose of the present article is briefly to review some of the most important relations. In doing so, we may first present our revised index of employment, to supplement the

This generalization is supported by a wide range of, annual statistics for Great Britain and the United States, and especially by monthly statistics of clearings, commodity prices, and interest rates for the two countries. The argument based upon the three factors just named is outlined in the following paragraphs: i. (a) MINOR CYCLES averaging 3j years (4o months) in length. Though single cycles may vary considerably from this average, an underaverage cycle is often followed by an overaverage cycle, and vice versa, so that the average of two or three consecutive cycles is closer than the single cycle to the general average length.