This PDF is a selection from an out-of-print volume from the National Bureau of Economic Research Volume Title: Business Annals Volume Author/Editor: Willard Long Thorp Volume Publisher: NBER Volume ISBN: 0-87014-007-8 Volume URL: http://www.nber.org/books/thor26-1 Publication Date: 1926 Chapter Title: Introduction: Business Cycles as Revealed by Business Annals Chapter Author: Wesley C. Mitchell Chapter URL: http://www.nber.org/chapters/c4636 Chapter pages in book: (p. 15 - 100) BUSINESS ANNALS INTRODUCTION BUSINESS CYCLES AS REVEALED BY BUSINESS ANNALS. BY WESLEY C. MITCHELL.' I. The Uses of Business Annals. For our work upon business cycles in the National Bureau of Eco- nomic Research we have made two collections of materials, which we think others will find useful. One is a collection of statistics. We are assembling in a single source-hook the widely scattered records of economic and social activities in Great Britain, France, Germany, the United States, and, to a limited extent, in other countries. We are going back to the earliest compilations of continuous series and coming down to the current year, making critical notes upon the character of the data, piecing together fragments, and presenting the figures by months or quarters whenever possible. When this collection is ready for use, we hope that we shall be enabled to publish it in two or more volumes. The second collection, consisting of what we call "business annals", is presented in this book. Though less bulky than the statistics, these annals cover a wider area and a longer period. For the statistical record shrinks rapidly as we go back 25, 50, and 100 years; few indeed are the series extending to the earliest decades covered by our British and American annals. Another difference is that the annals sketch a general situation of which the statistics show certain parts in 'The writer has received much help from his colleagues in the National Bureau of Economic Research, especially Frederick C. Mills and Willard L. Tborp. 15 16 BUSINESS ANNALS detail. Thus the two collections supplement each other :—one pre- sents the most precise records kept of specific economic or social processes, the other traces the fluctuations in economic and social fortunes at large. Together they sum up much of the experience which may be made to light the path into the future. The value of statistics for practical and theoretical uses is now widely recognized, and so too are the difficulties, dangers and limita- tions attending their use. Business annals are a less familiar form of record, and require an introduction showing how they are made, how far they may be trusted, and what they add to our knowledge. One merit of the annals, simplicity, is obvious at first glance. They tell a story, tell it in the most straightforward way, and tell it without the use of symbols or strange terms. Men who lack the time, patience, or technical training to wring the meaning from statistical' tables can use this record with ease. The story which the annals tell concerns the vicissitudes of economic fortune through which 17 countries have passed in periods which range from 36 t.o 136 years. In the fewest possible words they trace the fluctuations in manufacturing, construction work, employ- ment, domestic and foreign trade, prices, speculation, financial opera- tions, and agriculture, so far as the facts can be gathered from available sources. Thus the annals cover the grand divisions of economic activit.y. They also note the most important events of a non-economic sort which presumably influence economic activity— the making of war and peace, diplomatic strains, domestic disorders, changes of political administration and economic policy, droughts, floods, earthquakes, epidemics among men or cattle, and the like. Everyone who reads this story year after year for nation after nation will see running through all the episodes one of the problems of modern civilization. In no country covered by the annals—not in the most rapidly growing communities developing rich new lands, and not in the most conservative of old communities—.--does a period of economic prosperity ever last more than five or six years at a stretch. Each country has its seasons of prosperity, but. these seasons always end in seasons of depression. In their turn, the periods of depression yield to new periods of prosperity. Economic experience is made up of such alternations of good and ill fortune and the transi- tions from one to the other. The alternations are more marked or INTRODUCTION 17 more frequent in some countries than in others; but they occur everywhere. No country has yet learned to control them. Upon the vexed question how these fluctuations come about, the annals throw no direct light. That problem is reserved for another volume, in which the National Bureau hopes to make what contribu- tions it can toward a solution. But the annals establish certain facts about business fluctuations over a wider area and for a longer time than any other records. They show not only when periods of depression have begun in numerous countries, but also how long they have lasted, whether they have been severe or slight, confined to one country or spread over many. They show when revivals of activity have occurred and how these revivals have fared; they help us to find out how far the different branches of a country's business— particularly industry and trade, finance, and farming—have had common or diverse fortunes; they bring out the relations between the vicissitudes of business and the vicissitudes of politics and of natural processes. While these annals were compiled primarily to throw light upon business cycles, they will prove useful for many other ends. His- torians, political scientists, sociologists, and publicists frequently need to know the condition of business in certain times and countries. Statisticians dealing with time series will find that the annals provide an illuminating background for their special problems. Journalists who have to comment upon current developments, business men whose plans stretch into the future, indeed everyone who would control his expectations by experience, may profit by this condensed record of what has happened in the recent past. Those who are concerned with the spread of factory production and business enter- prise over the world can see how far the economic fluctuations in Russia, Brazil, Argentina, South Africa, British India, Japan and China are being assimilated to the economic fluctuations in Western Europe and North America. The annals show us also how the great commercial nations share in each other's prosperity and suffer from each other's reverses—a matter which merits far more attention than it commonly receives in discussions of national policy. II. The Scope of the Annals. A few earlier efforts have been made to provide a record of the sort here set forth; but most of them have been restricted in scope 18 BUSINESS ANNALS and buried in technical books, journals or official documents.1 Fuller accounts of business conditions in various countries, for various periods, have been included in many of the books dealing with the history or the theory of crises and depressions. But the reader who wished a wide view has had to do much hunting and piece together such scraps as he could collect. There have been gaps he could not fill. Nor has he been safe in trusting all that he found. Errors com- mitted in certain early investigations have been taken over by writers upon the theory of business cycles, and supplemented by fresh errors due to the forcing of a stereotyped pattern upon variable facts. Those who have taken the precaution to consult two authorities have fre- quently ended in doubt; for the chronologies of business crises and depressions are seldom precisely alike.2 To settle such doubts by appeal to the sources has been a laborious undertaking, for which few have had the time. How much labor is required to compile a record which merits confidence can be inferred from the lists of official documents, reports, periodicals, pamphlets and books cited by Dr. Thorp and his assistants ZThe use of condensed annals has been much commoner in political history than in economics. Quasi-annals are often found paralleling statistical tables, as in the "Annual Review" in the December number of the Investor's Monthly Manual, 1865 to date, London, and in Pixley and Abell's Silver Cwcular, London. Annals for England for the period 1816-1841 were compiled by Thomas Michael Sadler, and reported in his biography by Robert M. Seeley, published m 1842. Annals covering the period 1821-1830 are given by William Smart in his most helpful Economic Annals of the Nineteenth Century, London, 1917, vol. ii, p. 571. The most extensive British annals are found in the Final Report of the Royal Commission on Poor Law and Relief of Distress, 1909-1910 (Great Britain, Parliamentary Papers, 1909, vol. xxxvii, part vi, section 144, footnote). These annals cover the years from 1815 to 1907. Unfortunately, no source or authority is given, and a cloud is cast upon their reliability by several glaring inaccuracies, such as the placing of "Baring's crash" in 1894 instead of 1890. For the United States, the longest table of annals .is the table covering 1781 to 1922 in Otto C. Lightner's History of Business Depres.szons, New York, 1922, p. 123. These annals are phrased in meteorological terms, such as "stormy," "fair," "tornado." A brief but early series of annals, covering the years 1821 to 1832, is presented by the fiery William M. Gouge in his Short History of Paper Money and Banking in the United States, 4th ed., New York, 1840. Annals covering 1856 to 1893 appear in a broadside published by the Financial Graphic. More detailed annals for 1860 to 1922 are included in Business Barometers by Roger Babson, 16th ed., Wellesley Hills (Mass.), 1923. At least two attempts have been made to compile comparable annals for several countries. The business annals of the United States, England, Germany and France, 1889-1911, are included in Wesley C. Mitchell's Business Cycles, Berkeley (Calif.), 1913, p. 88. The annals for the same four countries for 1867-1880 are presented by Warren M. Persons, P. M. Tuttle, and Edwin Frickey in the Review of Economic Statistics, preliminary volume ii, supplement, July, 1920.—Note by Willard L. Thorp. 2For example, the years of American depression listed by Mr. George H. Hull (Industrial Depressions, New York, 1911, pp. 50, 51) do not tally with the periods discussed by Mr. Otto C. Lightner in his History of Business Depressions, New York, 1922. INTRODUCTION 19 as sources which they have used in dealing with the 17 countries covered.' Were all the publications which have been searched for information entered, the lists would be much longer. The work involves not merely much ransacking of sources likely and unlikely, but also careful weighing of the evidence. Pamphlets on questions of the day are valuable, particularly for our early years; but they often betray bias in the reporting of certain conditions. When the compiler finds differences in the views of contemporary observers, he may not be able to accept either view, or he may accept both; for sometimes the observers have different industries or different parts of a country in mind. The American annals have been carried back to 1790, the first year after the adoption of the Constitution. To make possible inter- national comparisons from the beginning of the American record, British annals have been compiled for the same period of 136 years. The lack of reliable sources, and, in Germany, lack of economic unity make it difficult to go back of 1840 in France, 1853 in Germany, and 1867 in Austria. The annals for these three countries are based upon drafts revised by foreign scholars, for whose generous cooperation the National Bureau expresses its hearty thanks—Professor Albert Aftalion of Paris, Dr. Robert R. Kuczynski of Berlin, and Dr. F. A. von Hayek of Vienna. To show the geographical spread of business fluctuations in recent times, it seemed desirable to add several other countries to the five included in the long-range studies. With 1890 as the starting point, Italy, the Netherlands, Sweden and Russia were chosen to represent diversified conditions in Europe. Dr. Robert F. Foerster has inspected the Italian annals, while the Russian annals were submitted to Professor N. D. Kondratieff, Director of the Con- juncture Institute of Moscow, and his colleagues Messrs. A. L. Vains- tein and M. B. Ignatieff. Next, three large English-speaking colonies on three continents were included—Canada, Australia and South Africa. For expert help upon the records of the last mentioned coun- try we are indebted to Dr. E. H. D. Arndt of Transvaal University College. To represent South America, Argentina and Brazil seemed fittest. Finally, the greatest of the Oriental civilizatioiis were in- cluded—British India, Japan, and China. See the bibliography, Chapter xix, below. 20 BUSINESS ANNALS III. The Trustworthiness of the Annals. Despite the care exercised by the National Bureau's staff and the generous cooperation of foreign scholars, the compilers dare not believe the annals to be free from error. Indeed one who is fresh from the puzzling task of weighing many pieces of testimony, each tinted by its sponsor's personal equation, is likely to feel more uncer- tainty about his conclusions than does the reader. The compiler develops a standard of judgment more exacting than one who has not spent months upon such work. Results of the sort here put down in crisp phrases are as much subject to a margin of error as are statistical summaries. The analogy goes further. While no one acquainted with census enumerations (to take a familiar example) believes that we have a strictly accurate record of the growth of popu- lation in the United States since 1790, no competent critic doubts that. the official figures give most valuable approximations to the unknown truth. These approximations are believed to be closer in some census years than others—the enumeration of 1870 for example seems to have been seriously defective in the Southern States. Fur- ther, the later results are certainly fuller than the earlier ones, and probably contain a lower percentage of error. So here. There are doubtless some errors in our results; these errors are doubtless more serious in some years than in others; the reliability of the results is less for the early periods than for the later periods in the same coun- tries, and less for countries like China, Russia, or Italy than for countries like Germany, England and the United States. It is more difficult to detect the general drift of affairs in a country like British India with a wide diversity of conditions than in countries like Brazil or South Africa, where business seems to be dominated by a few well recorded factors. But, though keenly conscious of t.he fallibility of our sources and of our judgments concerning them, we believe that the annals form a valuable approximation to the truth. Fortunately, there is a way of testing two samples of our results objectively. For varying periods of American and British experience, statisticians have compiled what are called "indexes of general business activity," or "indexes of the volume of trade." If these series deserve their names, the fluctuations which they show in economic' activity and the business changes which our annals describe may be expected to run similar courses. While Dr. Thorp and his assistants have made some use of INTRODUCTION 21 statistical tables in compiling their annals, and while the writers whose observations constitute their sources have done likewise, it is by no means a foregone conclusion that the annals and the indexes of business activity will agree closely. For the data used in making the indexes cover a much narrower range of economic activities than • are represented in the annual business reviews of such sources as consular reports, the London Economist, Raffalovich's Marché Financier, or the Financial Retriew. Moreover, in so far as consuls, or editors, or our own compilers have used statistics in drawing their conclusions, they have used the data in unadjusted, or but slightly altered form. The statisticians who make business indexes, on the other hand, subject their data to an elaborate series of transforma- tions. They compute and eliminate secular trends; often they elimi- nate also seasonal variations; in some cases they seek to eliminate the effects of price fluctuations. When they are combining several series, they may reduce the fluctuations of each to units of its standard deviation, and "weight" their average by the use of some- what dubious data. As a final step they often "smooth" their curves. All these operations are quite different from those which a financial editor performs when he passes through his mind reports from many cities and many industries, and sets down his broad conclusion con- cerning the course of business as a whole. The statistical operations are more objective and more precise; but they deal with more limited data, and deal with them in a more circumscribed and mechanical fashion. To all acquainted with the making of the two types of summaries, indeed, it will be clear that a comparison between the annals of busi- ness and the statistical indexes of general business activity is quite as much a test of the latter as of the former. The makers of the statistical indexes are usually careful to point out the limitations of their results, and eager to compare them with the results of other investigations. They recognize (1) that the original data are subject to varying margins of error; (2) that the technical methods of eliminating secular trends, seasonal fluctuations, and the effects of price variations are far from perfect; (3) that the residuals left in time series by these eliminations contain not merely the cyclical fluctuations, but also the effects of random factors peculiar to the series used. Even if a statistician had relatively abundant raw mate- riáls to work up, he would not claim that his results formed a strictly accurate record of changes in business conditions. In his 22 BUSINESS ANNALS eyes the best results he can get remain approximations, limited by the errors of the underlying data and the uncertainties of his technical methods. But the most serious limitation is that the statistician who seeks to cover a considerable period can find but few time series fit for his purpose. The indexes of general business or volume of trade which run back of the great war must be made on one of two plans. Either they must be records of a single type of activity—like Mr. Carl Snyder's "clearings index of business"—or they must be made by averaging the fluctuations of groups of series which themselves change from time to time like the American Telephone and Tele- graph Company's "index of general business conditions," or Professor Warren M. Persons' "index of trade." Now, no single type of transactions—not even such an inclusive type as the volume of cheeks cleared in all reporting towns outside of New York—can be taken to represent all the important phases of business activity. The payments made by check in the towns which have no clearing houses, and the payments made in coin and paper money may undergo fluctuations which differ in amplitude and timing from the fluctuations of clearings. That is merely a doubt concerning the faithfulness with which clearings represent changes in total pay- ments. Far more important is the certainty that the volume of payments made by check within a given period undergoes fluctuations materially different from the fluctuations which are taking place in the volume of goods produced, shipped, or consumed, and different from the fluctuations in employment, the disbursing of income, and the purchasing of consumers' goods. Yet the latter processes are quite as much a part of the fluctuations of business as is the former. The indexes made by averaging the fluctuations of several series represent a wider range of activities. But the, activities which can be included are those for which a statistical record happens to have been made for a relatively long period—not the activities which a statis- tician would choose were he planning an index. Moreover, the changes in the lists of series which are available for successive decades raise grave questions about the comparability of the results for the earlier and the later years covered. Finally, there are puzzling ques- tions about the interpretation of a composite made by averaging the fluctuations of series so different as (say) price indexes, values of goods imported, and tons of pig iron produced. INTRODUCTION 23 What we have in our business annals and our indexes of general business conditions, then, are different approaches to the problem of recording the fluctuations of economic activities—approaches each. of which has its uncertainties as well as its merits. We cannot expect them to agree perfectly. When they disagree we cannot say that the discrepancy necessarily means error in one or all; it may mean merely that the different activities reflected by the various approaches really did not change in quite the same way. But if we find a general consilience among the results we shall feel increased confidence in the reliability of both approaches, and may regard the occasional discrepancies as presenting genuine problems from the study of which fresh knowledge may be gained. The charts which follow offer as graphic a comparison as can well be made between our annals and the leading American and British indexes of general trade which cover considerable periods. In the column for each year is entered a brief characterization of business conditions drawn from the annals, and above are plotted the index curves.' The curves show cyclical fluctuations above and below the moving base traced by the monthly ordinates of the secular trends of the time series used (corrected when necessary for seasonal varia- tions). Since these ordinates are assigned the value of zero or 100 in the computations, they fall in the chart upon a horizontal line, which may be called the base. In studying the charts, we must bear in mind that they do not do full justice either to the statistical method of presenting changes in. business or to the annals. It is a commonplace that. no statistical average represents adequately t.he array of data from which it is computed. Just so, the catchwords used to summarize the annals do not represent adequately Dr. Thorp's records. Much more than the charts show can be learned by examining the series combined to make the indexes of business conditions, and by reading the fuller form of t.he annals. In confining our comparison to the most abstract and symbolic summaries of the two sets of materials, we are imposing a severe test of conformity. 1For the methods followed in making the two American indexes used in Chart 1, see M. C. Rorty, "The Statistical Control of Business Activities," Harvard Business Review, January, 1923, vol. 1, pp. 154-166, and Carl Snyder, "A New Clearings Index of Business for Years," Journal of the American Statistical Association, Septem- ber, 1924, vol. xix, pp. 329-335. For the recent items in the two series, we are indebted respectively to Mr. Seymour L. Andrew, Chief Statistician of the American Telephone and Telegraph Company, and to Mr. Snyder of the Federal Reserve Bank of New York.