of crises, and it seems to belong there quite as much as certain dates which Bouniatian admits as turning-points unaccompanied by severe financial strain, there are 16 cycles, ranging in length from about 4 to about 13 years, and averaging not quite 8 years. Of the cycles marked off by recessions, 22 are shown. Perhaps we should add recessions in 1814 after the first abdication of Napoleon, in 1861 when the American Civil War upset the cotton trade, and in 1870 when the Franco-Prussian War brought confusion to the financial markets. But in none of these cases does the evidence indicate a general slackening of trade. Even if these cases were counted, it would still appear that English business has experienced fewer recessions than American business during the same period of four generations. Hence English cycles have been longer on the average than American cycles. Taking the dates entered in the table we get an average *The dates thus marked show the crises recognized by Mentor Bouniatian, Les Crises Économiques, Paris, 1922, p. 43. Most authorities would include 1913, also, on the same grounds that lead Bouniatian to find crises in 1882 and 1900, although these years were not marked by severe financial strain. duration of 534 years in England, against 4 years in the United States. And the frequency of recessions has been diminishing in England, though not in this country. In the first half of the period covered (1793-1857), the average length of English cycles was nearly 5 years, in the second half (1857-1920) exactly 7 years. But these averages are even less a guide to business forecasting in England than in America. It is difficult to find any regular order in the lengths of the successive cycles in either Table 1 or Table 2. When we tabulate the frequency of English cycles according to duration we find less concentration at the mode than in the corresponding American table. From 1793-1920 there were 2 cycles about 2 years long (1829-31, and 1918-20) 215 2 2423 Four-year cycles are most common in England, three-year cycles in the United States. One-half of the English cases are 4-6 years in length, while two-thirds of the American cases are grouped at 3-5 years. On applying the same methods of analysis to the three other countries for which we have annals running back to the 1860's, 1850's, and 1840's, we find that in average duration their cycles are intermediate between the English and the American patterns. The average length works out as follows: 1838-1920-82 years France, 15 cycles, average length 51⁄2 years. England, (1837-1920), 12 cycles, average length nearly 7 years. United States, (1837-1920), 22 cycles, average length 34 years. 1848-1925-77 years Germany, 15 cycles, average length 5 years. England (1847-1920), 11 cycles, average length 6 years. years. 1866-1922-56 years Austria, 10 cycles, average length 5.6 years. England (1866-1920), 8 cycles, average length 634 years. United States (1865-1923), 15 cycles, average length not quite 4 years. If we split the periods covered by our annals for these three countries as nearly in the middle as possible, we find that the average length of business cycles has increased in France as in England, decreased in Germany, and remained constant in Austria as in the United States. The figures are: 3. FREQUENCY DISTRIBUTIONS OF THE MEASUREMENTS BASED ON THE ANNALS. A systematic summary of our evidence concerning the duration of business cycles is provided by the following exhibits. Table 3 is a companion piece to Tables 1 and 2. It shows the dates of recessions in fifteen countries as accurately as Dr. Thorp can determine them from the annals, and shows also the approximate duration of successive cycles reckoned to the nearest whole year. Chart III is a graphic version of Tables 1, 2 and 3. It uses lines of varying length to show the duration of business cycles in each of our countries, in chronological order. We can treat the observations upon the duration of business cycles assembled in this table and chart as the data of an historical inquiry, or as the data of a theoretical problem. In the first case we ask: What has been the duration of business cycles in the countries and during the periods for which we have annals? In the second case we ask: What expectations regarding the duration of business cycles are justified by the sample observations in hand? As historical data, our observations probably contain inaccuracies. Conceiving a business recession as a decline in economic activity which follows a period of expansion and spreads over most of a country's industries, we have sought to find and date every recession which occurred in certain countries during certain periods. On the basis of these recession dates, we have measured the duration of successive cycles to the nearest whole year. Finally, we have struck averages from these measurements. Mistakes may have occurred in any of these steps. We may have omitted some recessions; we may have included some cases which do not fit our definition of recessions; we may have blundered in measuring or averaging. But so long as we are trying merely to report what has taken place in the past, these doubts concerning the accuracy of our work are all that need concern us. The historical record is fixed; it has its unique features and interest; in studying it we can indulge in no speculations. A subtler problem and doubts of another order are presented when we treat our observations as data for drawing theoretical conclusions regarding the duration of business cycles at large. For this purpose, we must ask, not merely whether our observations are historically dependable, but also whether they constitute a representative sample of the phenomena measured. Are the observations sufficiently numerous? Are they sufficiently independent of each other? Ought we discard the observations upon cycles which we think have been cut short or prolonged by factors which have no organic relation to business activity? In the sense in which the term is used here-recurrences of prosperity, recession, depression and revival in the business activities of countries taken as units-the total number of past business cycles may well be less than a thousand. For business cycles are phenomena peculiar to a certain form of economic organization which has been dominant even in Western Europe for less than two centuries, and for briefer periods in other regions. And the average cycle has lasted five years, if we may trust our data. Of the whole number of cases to date, the 166 cycles we have measured form a significant fraction. By compiling business annals for Norway, Belgium, Switzerland, Denmark, Spain, New Zealand and Chile we could probably get additional observations as satisfactory as some of those already included. Perhaps we could trace business cycles in Greece, Egypt, Turkey, some of the Balkan States, possibly Mexico, and additional countries TABLE 3 DATES OF BUSINESS RECESSIONS IN FIFTEEN COUNTRIES: VARIOUS YEARS TO 1925 |