MILLION DOLLARS

60

general upward trend of the curve during the years of prosperity immediately preceding the crisis of 1907 is noteworthy. The effects of the crisis are shown in the ensuing decline. It is to be noted, also, that these statistics show a seasonal fluctuation, with a peak in March, due presumably to the importing of merchandise for the spring trade, a sag in the summer, and

200

CHART I

180

100

140

120

another upward movement in the autumn caused by imports of merchandise for the holiday trade. (2) Exports of Merchandise. For judging business conditions, the export statistics of the United States 1 are much less useful than the import statistics. The export statistics are, in themselves, less reliable because of the greater percentage of error in the returns; they are not scrutinized by the customs inspectors and there is no adequate check upon the accuracy of the exporters'

The export statistics for Great Britain, on the contrary, are a particularly good index of conditions in that country, since the British manufacturers are so largely dependent upon foreign markets.

manifests. Furthermore, because of the predominance of raw materials and foodstuffs in our export trade, the volume of our exports depends largely upon conditions affecting demand from foreign countries. The move

ment does not necessarily indicate the strength or weakness of the domestic situation. The exports of manufactured goods tend to fall off with improvement in domestic demand and to increase during depression, when our manufacturers show their greatest interest in developing foreign trade. The course of the export trade, 1903-08, is also shown on Chart I. The marked seasonal fluctuation is due to the heavy exportation of raw cotton and other agricultural products during the late autumn and winter months.

Balance of trade statistics, which show the difference between imports and exports, seem to me to have little significance. There are so many invisible exports and imports that the balance of trade figures always involve a large element of uncertainty. How great is the foreign indebtedness upon which interest payments are due? Is the investment of foreign capital increasing or is the foreign indebtedness being paid off? What shipments of securities are being made? What transportation charges are to be paid? No record can be kept of all these transactions, which have just as much influence as the visible merchandise shipments upon foreign exchange rates and the movement of specie.

(3) Immigration. The statistics for immigration fluctuate in a general way with business conditions in the United States. An upward tendency was indicated, for example, during 1905, 1906, and 1907, and a marked reaction in 1908. The immigration figures are of especial interest to certain manufacturers, since they give some indication of the increase in the supply of unskilled labor. Their significance as a general index,

however, is lessened by the fact that the movement of immigrants adjusts itself only with more or less delay, according to information transmitted from this country to the foreigners before they leave their homes. The net immigration, that is the total number of immigrants less the number of emigrants, should be more significant; but the latter figures have been published only since July, 1907. A much more serious criticism of the use of immigration statistics as a business barometer is that they are influenced not only by conditions in the United States but by industrial, social, and political conditions in the countries whence the immigrants At best these statistics could not show a very close approximation to actual changes in business conditions in this country. At the present time, in consequence of the European war, all comparisons have become inconclusive.1

come.

(4) Bank Clearings. Because of the wide-spread custom of making payments by check, bank clearings give a fairly accurate index to the volume of business transactions. Altho influenced by general changes in prices, by bank consolidations, and by the spread of the check-using habit, bank clearings show approximately how much business is being done at any one time. As a business index, the bank clearings for the United States exclusive of New York City are more significant than the total clearings. The New York clearings, which constitute about one-half of the total clearings for the country, are so affected by the volume of speculative transactions upon the Stock Exchange that they should at least be considered separately. The clearings in other cities where stock exchanges are located are not a sufficiently large proportion of the total to necessitate

1 Immigration statistics are currently published in numerous periodicals and also in the Monthly Summary of Commerce and Finance.

their exclusion. Bank clearings are not subject to wide fluctuations and do not indicate what is likely to take place in the future, but they do show in a general way what is taking place. The clearings statistics as reported by Bradstreet's, The Commercial and Financial Chronicle, and Dun's Review differ slightly in detail but approximately agree.

(5) Railroad Gross Earnings. Railroad traffic fluctuates with the amount of business being done in the community. As an index to the volume of traffic, since tonnage figures are not currently available,1 railroad gross earnings are commonly used. Statistics for net earnings show the general financial condition of the roads, but are far less useful for general purposes than the gross earnings. The latter are in the same class as bank clearings, showing what is taking place but foretelling little of the future.

Because of the delay which occurs in securing reports from some of the companies, the total earnings for all the roads in the country cannot advantageously be used in studying business indices. It is necessary, therefore, to take the earnings for a representative group of roads. In the Commercial and Financial Chronicle statistics for the earnings of a group of roads are given monthly. These statistics are usually made up from preliminary returns and are thus, to some degree, subject to revision. The most serious difficulty however, which prevents the use, except for casual observations, of such compilations as those of the Commercial and Financial Chronicle, is that the make-up of the group continually changes. The number of roads included varies from month to month, yielding totals

1 For a few years the American Railway Association has published a monthly bulletin, "Statement of Freight Car Balance and Performance," which gives, amongst other things, the ton miles of freight carried, but these bulletins appear several months late.

which usually can be compared only with the preceding month or with the corresponding month of the preceding year. Mr. Babson presents on his desk sheet a useful monthly table of the total gross earnings of ten railroads, always including figures for the same roads.

(6) Idle Cars. From January, 1908, to November, 1914, the American Railway Association issued semimonthly reports on the number of idle freight cars. Since February 1, 1915, monthly reports have been issued. Altho these reports have probably been of assistance to railroad officials by furnishing a guide to traffic demands and by enabling them to secure a better balance of car supply, I am disposed to think that the statistics are much less reliable as a business index than has been commonly believed.

In the first place, the number of roads reporting has varied. On April 1, 1914, the number of roads reporting was 190; on June 1, 176; on October 1, 204; and on November 1, 192. Similar variations appear for other months. Further, in making any long time comparisons, the change in the capacity of the cars is also to be considered. But neither of these factors is so fundamental as the irregularity in the number of new cars added from year to year. The statistics for the number of freight cars idle cannot show the fluctuation in the volume of traffic and, hence, the amount of business done, when the number of cars available for service itself fluctuates irregularly. The number of idle cars depends not only upon the number actually in use, but also upon the number of new cars added and of old cars scrapped. The variations in the number of cars in service are shown by the following table, compiled from the bulletins of the American Railway Association. The wide divergencies in the number of new cars added during these years vitally affect the number of cars idle at any one time;