BRITISH EXPORTS AND THE BAROMETER

Two of the most familiar diagrams in economics are that setting out the value of the exports of United Kingdom produce per head of the population in each year, and that setting out the general level of wholesale prices as recorded by Sauerbeck's index-number. Each of these two curves shows a large and somewhat irregular fluctuation, in general accord with the other indications of industrial activity and depression, such as the bank rate, unemployment percentages, pauperism, marriage rate, consumption of beer, and number of larcenies.

The general agreement is subject to marked exceptions in particular years, and the length of the separate fluctuations varies greatly (successive minima of export values coming at intervals of six, twelve, six, eight, seven and eight years). Moreover, each of the two curves shows general changes of level, accomplished over long periods and blurring the effect of the cyclical fluctuations. Thus export values per head fluctuated about the same general level from 1820 to 1840; the general level then rose rapidly for the next twenty-five or thirty years, remained stationary for the next thirty, and started to rise rapidly again. from 1900 to 1913.

in

The correlation between the curves of export values and of prices is to some extent due simply to the presence of a common element in each. Values depend on price as well as quantity; so far as the Sauerbeck index is a true index of prices, any change of it tends to be reflected in the export values. An increase in the latter may mean merely an increase of prices and not in the volume of exports. A great many attempts have been made at various times to correct the statistics of exports (and imports) by excluding the effect of price variation and determining the actual quantities of exports (and similarly of imports) in each year. So far as I know, however, attention has not hitherto been drawn to the results of simply dividing the export values per head for each year, by the corresponding Sauerbeck number for that year, and setting out the quotients in the form of index-numbers. These new index-numbers are given in the attached table and chart, for the period from 1820 to 1913. They yield a curve of rather remarkable character and regularity. A ruler laid along

the line joining the depressions of 1847 and 1893 practically passes through two other points of depression, 1862 and 1877-8, and shows above it three successive double waves, 1847 to 1862, 1862 to 1877-8, and 1877-8 to 1893, each of almost the same length (between fifteen and sixteen years), and the same character (with a minor depression about half-way). Fifteen years back brings us to another depression in 1832; sixteen years forward to the last one in 1909. Each of these periods shows, like the three central ones, a double wave (with depressions about halfway through); the only difference is that in each case the general tendency of the graph is altered, and that the half-way depressions appear relatively deeper. Broadly, we see in the whole period of seventy-seven years, from 1832 to 1909, five main fluctuations of practically the same length and character. A cycle of somewhere between 152 and 15'4 years would exactly fit all the depressions.

The size of the waves increases from the beginning to the end of the period covered by the chart; this does not, however, indicate any greater violence of fluctuations, as the whole scale of the export trade is growing. If the figures were expressed logarithmically, the waves would be far more nearly equal.

A periodic fluctuation of this regularity is altogether exceptional in economic records. Its closest parallel is to be found in the seasonal fluctuations of trades, such as building and printing, which are directly dependent on climate and the social habits to which the climate gives rise. It does not suggest sunspots, partly because the length of the fluctuation is different, but even more because it is far more regular than the sunspots. But it does suggest the operation of some physical factor -in the earth or its atmosphere-with a periodicity almost as strong as that of the tides. It will be desirable to consider first what it is precisely that the new index-numbers can be taken to represent.

They do not represent the volume of exports. To get that we should have to divide the export values by a price index-number based only upon the prices of such classes of goods as are exported.1 The new index-numbers, in so far as the Sauerbeck index is correct, represent simply the purchasing power of British exports. Since this purchasing power is used in general only to

1 Comparison of the new index with the estimates of export volumes from 1890 onwards, as given in the third Fiscal Blue Book, shows general but not complete agreement. The movements of the new index, though strictly it represents only the purchasing power of the British export trade, correspond also, as a rule, to variations in the scale of the trade.

purchase, the fluctuations of the index-numbers should reproduce fluctuations in what is purchased by British exports. Broadly, British foreign trade amounts to sending out manufactured articles and coal in order to purchase on the one hand gold and on the other hand crops of all kinds (grain, cotton, sugar, jute, oil-seeds) or livestock (and their produce) which are dependent on such crops. The proportion of manufactures among the imports, though growing, is still small. The suggestion here put forward for consideration is :

1. That the British export figures treated in the way described record indirectly changes in the world's output of gold and of growing crops, the new index tending to rise as this output increases, and to fall as it diminishes.

2. That the apparent periodicity of the figures is a true one, and arises from some physical cause or combination of causes, whether terrestrial or super-terrestrial, having a regular fluctuation in a period of between fifteen and sixteen years, or some factor thereof, and causing a corresponding fluctuation in the productivity of the earth.

One part of this hypothesis can at once be tested. The output of gold is known and its history is simple. After remaining at a low level up to about 1845, the world's production of gold then increased eight or nine times by 1852-3, remained at its new level till about 1892-3, when it began a new rapid rise which, subject to a relapse in the time of the South African War (1900-3), continued to the outbreak of the Great War in 1914. In so far as an increased output of gold simply goes into the circulation and contributes to raise prices, it is presumably already discounted in the new index (for this represents a fraction with the Sauerbeck price-number as denominator). A substantial proportion of the gold output, however, serves industrial purposes and is thus a commodity whose output should affect the new index.

Simple inspection suggests at once that there is a connection between the output of gold and the new index-number. The first great rise in the world's output of gold (from 1845) at once turned upwards the course of British exports as shown in the new index. The second great rise from 1893 onwards repeated the effect. The South African War, from 1899-1902, checked the output of gold and its export abroad; this check is reflected in the depression of the new index curve. In the last years before the Great War, as gold steadily increased, exports, even as corrected for change of prices, continued to soar. It would no doubt be

possible to make arithmetical allowance for the variation in the gold output, and so eliminate its influence, leaving a curve which, on the hypothesis advanced above, would record simply the fluctuation of the world's crops. It is clear that the elimination of gold would have the effect of raising the curve relatively in the years 1899-1902, and lowering it in the period subsequent to 1895; the last sixteen-year period would thus fall into line with the rest of the curve. It would also bring the general course of the curve before and after 1845 more nearly into line.

For the next stage in the examination the necessary material is at the moment lacking, or has not been collected in a form allowing its use. It is indeed a matter of common observation that the Indian peasant's demand for cotton piece goods is directly dependent on the success or failure of his harvest. But there are no records of the crops of the world as a whole, or even of a large part of them, covering a long period of years, and making possible the construction of a crop curve to set beside the curve of exports. For the moment I must be content to cite a few familiar and significant facts as to particular countries. In India famines have occurred in 1861, 1877, 1891-2 and 1907-8, that is to say, in each case, at, or just before, one of the minimum points of the curve. There have also been other famines, some bearing a similar relation to other depressions in the curve, e.g., 1854, 1866, 1897, and some for which no such connection is apparent. In the United States, for which continuous records prior to 1866 are lacking, there have been very well marked shortages in the wheat crop in 1893 and 1907, with a less well marked one in 1876. The corn crop and the cotton crop have generally shown a depression at or about the same time. Here, again, there are other cases of deficient harvest, some corresponding with minor depressions in the export curve, some showing no apparent connection. Clearly the harvests of each particular country, or of any one kind, are each affected by many different causes. All that can be said for the moment is that the crop results are, on the whole, consistent with the suggestion of a single general cause affecting the whole world at intervals of fifteen to sixteen years. They give hope that by a wide enough collection of data the local and sporadic fluctuations would be flattened out and that a characteristic common fluctuation would be revealed.

It may be suggested that if the hypothesis under consideration is correct (that the British exports to the world as a whole, corrected by the general index of prices, are determined by the crops of the world as a whole), then a similar relation should be found between the exports to any particular country and the

crops there. Some provisional calculations as to India and the United States of America suggest that some relation of this kind can be found there. But the application of the present method to the trade and production of a single country is difficult, for several reasons. In the first place, though the Sauerbeck numbers may, and probably do, afford an adequate index of price changes for all commodities taken together, they may not be at all accurate in respect of the commodities sent to, or purchased from, a single country. In the second place, trade is international, and is fluid under the influence of capital seeking investment.

The United Kingdom, throughout the period under review, has been a wealthy country which was exporting capital and developing the rest of the world. The British manufacturers, in a year when they export more than usual, may use this power not to bring the increased supplies to this country, but to transfer it from one country to another. The fluctuation shown in the export figures is visible also in the imports, but is not nearly so well marked. The important thing to study is not the imports, whether from particular countries or as a whole, to the United Kingdom, but the total world production. The value and validity of the new method (if any) lies in its generality; one should not expect to find it applicable, without many adjustments, to the case of particular countries.

The relation suggested as being shown in the chart is between British exports as a whole and the agricultural productivity of the world as a whole. In a specially fruitful year (when the crops of the world as a whole are above the average) the producers of the crops, or those to whom they sell, will make a specially large demand for articles which they do not produce, i.e., manufactured goods. They may get these from the United Kingdom, and so stimulate exports to themselves; or, in any particular country they may get the goods from the manufacturers in that country. The latter manufacturers will in that case be less able than before themselves to make for other markets, and will thus increase the demand upon the United Kingdom and the exports from the United Kingdom, though not necessarily to their own country. Once the manufactures of any country are so large a proportion of the total manufactures of the world, as is the case with the United Kingdom, one may expect to find fluctuations in agricultural production reflected in them directly or indirectly. At bottom, the curve shows the relation not between the United Kingdom and the world, but between the manufacturing activity of the world (as measured by the British sample) and its agricultural activity.

No. 117.-VOL. XXX.