HomeLibraryClassroomTimeline

    Publishing Information

    Machinery and Unemployment

    By Leo Wolman

  1. CURRENT discussion of what has popularly come to be known as "technological unemployment" still appears to disregard the fact that measurement of the phenomenon, not to mention the explanation of it, is a vastly complex and technical matter. Measuring the changing productivity of industry and the efficiency of labor, on the one hand, and estimating the changing volume of unemployment and accounting for the causes of the change, on the other, are related economic problems on which a tremendous amount of work has been done during the past fifteen years. Beginning with the index numbers of the physical output of 1920, the work of E. E. Day, Woodlief Thomas, Carl Snyder, F. C. Mills, and many others in this field has made available a great fund of additional materials and constant refinements in the methods of interpretation. In the measurement of unemployment, likewise, rising popular interest in the subject and the eagerness of students to examine anew the effects of machinery have produced the invaluable studies on the volume of unemployment by M. B. Givens, Paul Douglas, and Ralph Hurlin, and the particular inquiries on displacement by machinery by Isador Lubin, Ewan Clague, W. J. Couper, and Elizabeth Baker, together with the remarkable series of investigations conducted since the early twenties by the staff of the United States Bureau of Labor Statistics.

  2. The material, then, on both the output of industry and the volume of unemployment exists now in greater abundance than ever before. But there is still wide difference of opinion. among close students as to precisely what has happened, say, in the last twenty or thirty years. These differences of opinion, moreover, are not the result of hair-splitting but of real difficulties which cannot be disposed of by any amount of argumentation. Exhaustive treatment of this whole question, assuming I were prepared and competent to undertake it, obviously cannot be attempted within the compass of an article. I shall, therefore, limit myself to the consideration of a few of the problems connected with measuring, first, output, ant then unemployment.

  3. At the outset there is the question of the appropriate general method of attack. The preparation of a list of laborsaving machines, or processes, however interesting such a list may be, cannot serve the purpose of adequately describing the experience of the past and estimating the probabilities of the future. In the analysis of this kind of problem, the economist, the statistician, and for that matter the engineer are concerned with aggregate and average changes and results. The collection of an exhaustive list would require enormous funds and the exercise of the most meticulous care in the job of investigation. The results of such an enumeration might well be spectacular, as they probably would have been during many periods in the industrial history of this country duce the Civil War. But it would be the sheerest accident if they faithfully and accurately reflected the aggregate trend.

  4. So far as the basic data for the measurement of aggregate changes in the physical output of industry are concerned, they are still defective in quality and in comprehensiveness. Statistical devices must frequently be employed which are of dubious value and uncertain meaning. Since, also, one obvious purpose of current investigation of these questions is the discovery of comparative historical developments--to determine, for instance, whether the current rate of industrial output is faster or slower than it was at some time in the past--it must be remembered that defects in the current data are often multiplied many times the farther back we go with the same or similar series. Long-time comparisons, consequently, must be accepted for what they are, as reflections of the best measurements possible under the circumstances and not of instruments of unquestioned precision. More detailed work on the earlier series, particularly with reference to the experience of specific industries, may, in my judgment, produce results greatly at variance with some of the conclusions now widely held.

  5. Finally, the economic characteristics of the last twenty years have in so many respects been unusual that no valid interpretation of any important feature of the period is possible without taking these unusual conditions at least partially into account. The rise of the automobile and kindred industries after the war, the great growth of purely war industries and their rapid demobilization between 1916 and 1920, and the dilution of labor attending the diversion of men from industry into the American army are a few of the more important factors affecting both the productivity of industry and the volume of unemployment during this whole period. In the face of forces of this magnitude, the events of the relatively few years from 1922 to 1929 may to a much greater degree represent the resolution of temporary forces, originating directly and indirectly in the war, than the unfolding of persistent long-time trends.

  6. With these cautions in mind, it should be possible to examine the data on the physical product of several categories of American industry with some appreciation of the limitations of the available measures and of the accidental or temporary character of what has happened. Before presenting illustrative data, I should say that inspection of a great mass of material leads me to the conclusion that the physical output of industry and the physical output per worker increased at a very rapid rate in only a few of the years after 1920 By how much this rate exceeded the rate of change in single ears in the past it is almost impossible to determine. But it is certainly the case that the conclusions drawn from the available data are often powerfully influenced, if not determined, by the purely technical fact of the length of the periods chosen for comparison and the character of the terminal years in each period.

  7. A recent and authoritative survey of the advance in agricultural technique is found in the chapter on agriculture in "Recent Social Trends," by O. E. Baker of the United States Department of Agriculture:

    The five years from 1922 to 1926 [he writes] are in several ways the most remarkable in the history of American agriculture. Agricultural production increased about 27 per cent, while crop acreage remained practically stationary and labor engaged in agriculture declined.... Nevertheless, production per person engaged in several types of farming has not increased as rapidly as is commonly assumed and there is a wide margin available for further advance. The increase in efficiency has been notable principally in the production of the small grain and hay crops. Cotton today is picked by hand, as it was a century ago, most of the corn is still husked or snapped by hand, and practically all the fruit is picked by hand, while much of the fruit has to be sprayed also, which was not done a century ago.

  8. Estimating the future effects of the universal (italics mine) mechanization of agriculture, Mr. Baker concludes that the average production per labor year would in the future be increased by 33 per cent "for the nation as a whole." "This," he states, "is about the same percentage increase as has occurred during the past thirty years."

  9. In the bituminous coal industry, output per man per day increased a fraction less than 100 per cent in the forty years from 1890 to 1930. During the last decade, 1920 to 1930, or one-fourth of the whole period, per capita output measured in the same way rose slightly more than 25 per cent. The productivity of labor in the anthracite industry, however, actually declined from 1920 to 1930. Viewing the whole mineral situation from the standpoint of productivity and costs, F. G. Tryon, in the most valuable summary of this question it has been my good fortune to see (in "Recent Social Trends"), forecasts the future in these terms:

    Considering minerals as a whole, the outlook in the next ten years is for ample supplies available at declining cost.... In the long-time outlook [however] the outstanding facts are the growing difficulties of mining and the prospect of an ultimate increase in cost. The tendencies are unmistakable, and the experience of England shows how early in the exploitation of a mineral resource the stage of increased cost may arrive.

  10. On the railroads the per capita output of labor increased 14 per cent from 1899 to 1909; 25.5 per cent from 1909 to 1919; and 2 per cent from 1920 to 1930. Here, then, is the measurable record of productivity in three basic industries. While the changes recorded since the war are noteworthy and afford substantial evidence of technological progress, they are still far from representing an unprecedented or revolutionary development.

  11. What the really difficult problems of measurement ant interpretation are, is most clearly revealed in estimating the course of total ant per capita output in the heterogeneous group of manufacturing industries. In this group the per capita output of labor as measured by one competent student increased 7 per cent from 1899 to 1909; declined 1 per cent from 1909 to 1919; and then increased 49 per cent from 1919 to 1929. In view of these data, which have in reality been the source of much of the current belief in the widespread displacement of men by machinery, the rapid advance in the efficiency of labor after the war appears unquestionable. But there are many good reasons, technical and logical, for accepting this conclusion with considerable hesitation.

  12. In the first place, the rate of the post-war rise in the productivity of manufacturing labor is in part the result of an apparently slow rate of increase in productivity in earlier periods. This condition can be illustrated in two ways. Among the four groups of industries here considered, the productivity of manufacturing industries showed the slowest rate of increase from 1899 to 1925. Thus the increases in this period were 53 per cent in agriculture, 56 in railway transportation, 99 in mining, and 42.5 in manufactures. Between 1909 ant 1919, furthermore, the per capita output of labor actually declined in manufactures, whereas it increased substantially in each of the three other groups. It requires no profound knowledge of statistical method to see how the decline in productivity before 1919 helps to explain part of the rapid increase in manufacturing output after 1919.

  13. The post-war rise in manufacturing productivity was also neither continuous nor uniform. During the years from 1923 to 1929 there were only two occasions when the annual increase was striking, one from 1924 to 1925 and the other from 1927 to 1928. In all other years of this period the changes were so small as to be hardly noticeable. This is an important fact to bear in mind before the conclusions drawn from the experience of so brief a span of years are accepted as unmistakable evidence of a powerful and permanent trend. Professor Frederick C. Mills, for instance, in a recent encyclopedic study of post-war and pre-war economic changes ("Economic Tendencies"), concludes that the annual rate of increase in per capita physical output in industry was twice as great in the post-war as in the pre-war period. For this comparison Mr. Mills chose 1922-29 as the post-war period, and 1901-13 as the pre-war. Although the choice of periods must always be to some extent arbitrary, it is clear that the results derived by Mr. Mills are in a measure determined by certain formal features of the periods he has elected to use. The pre-war period is thirteen years long, the post-war only eight. In economic series longer periods are almost certain to include more years of business recession than shorter periods; and in years of business recession per capita output usually, though not always, declines. If, then, the post-war period were extended to include the years since 1929, all of which have so far been years of business depression, the results would obviously be quite different.

  14. These comments on the measurement of productivity apply with added force to the estimates of the volume of unemployment in this country. The specific studies of the displacement of men by machines are meager samples of a vast and complicated economic system. The observed decline since the close of the war in employment in mining and railroading is in large part due to the competitive rise of new industries which have absorbed a great amount of labor. In the manufacturing industries, where the situation is confused by the influences of the war, the official figures for 1929 show that the numbers employed in manufacturing increased from 1909 to 1929 by 38 per cent, while during the same period the population of the country, ten years of age and over, increased by precisely the same amount. For the rest, every novice in economic statistics knows that it has been quite impossible to derive anything like an accurate measure of the volume or rate of unemployment in the United States, particularly since it has been only since 1928 that anything like comprehensive employment data have become available for several of the most important American industries.

  15. Direct estimate, finally, of the extent of technological unemployment must be made by measuring the joint effects on employment of changes in the volume of business activity and in the efficiency of labor. So far as I know, the only attempt to apply this method is to be found in an article by David Weintraub of the National Bureau of Economic Research in the Journal of the American Statistical Association for December, 1932. It should prove valuable in this connection to present one of his major conclusions. "So far as manufacturing industries are concerned," Mr. Weintraub states, "the evidence based on the comparatively short period 1920-31," does not "warrant the statement that the increased output per man-hour has resulted in a permanent displacement of workers. In fact, the analysis indicates that during the period 1920-26 a number greater than that displaced by the increased output per man-hour was absorbed by the increased physical volume of output," while in the years 1926-29 only 0.1 per cent of those displaced through greater efficiency failed to be reabsorbed in manufacturing industry.