Technological
Change and
Employment

EWAN CLAGUE AND LEON GREENBERG*

OVER THE LONG RUN, technological change and increased productivity have contributed to economic growth and expansion of employment opportunities. In the short run, however, numerous dislocations have occurred, requiring transfer and adjustment of workers to new occupations and new industries. These short-run changes, accompanied in recent years by high rates of unemployment, have frequently led to the question "How much unemployment is due to technological change?" A precise answer to this question may be impossible, because unemployment results from various factors-declining demand and output, competition, laborsaving machinery, and product substitution, among others. These factors may operate simultaneously and yet affect different parts of the economy in different ways.

Innovation and technological change which result in the installation of laborsaving machinery and equipment usually are reflected in increased productivity (output per man-hour) of plants and

• Respectively, Commissioner of Labor Statistics and Chief, Division of Productivity and Technological Developments, Bureau of Labor Statistics. This article is based on a paper presented by the authors at the Twenty-First American Assembly, held on May 3-6, 1962.

1 The productivity estimates for most of these industries have not been previously published and are not available for publication. The individual industry estimates may be subject to some margin of error.

The year 1953 is used as a convenient starting point, partly based on availability of certain estimates. Results and conclusions would be roughly the same if another year of similar economic activity, say 1955, were used. Recession years such as 1954 should be avoided as terminal points. Data were not available for years subsequent to 1959.

Actually unit man-hours (i.e., man-hours required per unit of output), the inverse of output per man-hour, were used. The use of this inverse was necessary for direct comparison with percent changes in employment. For example, a 20-percent increase in output per man-hour is equivalent to a 17percent decrease in unit man-hours; the latter can be directly related to the percent decrease in employment.

4 To a certain extent, the relationship between unit man-hour and employment changes is affected by changes in annual hours worked, but on the average this was not significant during this period.

industries. If the rise in output per man-hour is not accompanied by an equivalent increase in output, employment (measured in terms of total hours employed) will decline. Product substitution, which can be viewed as another form of technological change, may result in increased output and employment in one industry but in decreased output and employment in another-although such substitution may not be accompanied by productivity gains. In both types of situations, workers who are displaced and new entrants to the labor force must seek job opportunities in other occupations or in other industries.

Employment and Output Per Man-Hour

Overall Relationship. In order to obtain some measure of the relationship between technological change and employment, an analysis was made of the productivity-employment experience of manufacturing industries during the postwar period. Estimates of productivity were derived for 201 industries (SIC 4-digit or combinations) for the period 1953-59, accounting for about 70 percent of total manufacturing employment and representing many different kinds of industrial activities. Estimates were also derived for 313 industries for the period 1947-57, accounting for over 95 percent of total manufacturing employment.1

The two time periods covered are overlapping, but dissimilar. The interval 1953-59 was selected in order to have the analysis cover a recent period. However, 1953-59 was a period in which total manufacturing employment declined. In order to determine whether the employment-productivity relationship would be different in a time of expanding employment, an analysis was also made of 1947-57, when total manufacturing employment increased. Output for total manufacturing increased in both periods, but rose more sharply in the early years than in the later years. The changes in productivity were correlated with the changes in employment, among the individual industries, between the first and last years of each of the time periods studied. During both periods some industries experienced decreases, others increases in employment. However, in each of the periods the industry-by-industry comparison resulted in a low correlation between changes in employment and changes in unit manhours.

3

The low correlation indicates that, on the average, productivity gains among industries were associated with both increases and decreases in employment over a period of years. Put another way, employment decreases occurred in industries with large productivity gains and those with small (or no) productivity gains. In the latter case, the declines in employment were associated with decreases in output.

Detailed Industry Comparisons. The overall correlation analysis obscures differences in the impact of technological changes occurring among the several hundred industries within manufacturing. Therefore, a more detailed examination was made for each of the periods to arrive at some estimate of the amount of "disemployment."

The term "disemployment" is used to refer to those declines in industry employment associated with increases in productivity (or the declines in unit man-hours). Quite clearly, some decreases in employment were associated with declines in output and not with productivity change. It is true that employment reduction in some plants and industries was due to technological substitution-for example, the replacement of coal by oil, natural fiber by synthetic fiber, rail transportation by the automobile, metals by plastics. This kind of change does not show up in the productivity figures unless it happens to be accompanied by reduced labor requirements.

At the same time, decreases in industry employment do not necessarily mean that workers are laid off and become unemployed. In some cases, employment reductions are achieved through normal attrition-deaths, retirements, and quits.

This analysis is intended to suggest the magnitude of the productivity-disemployment relationship and, it is hoped, indicate some directions for further work in attempting to measure this relationship. (See statement of summary and limitations at end of article.)

For purposes of this analysis, the industries were divided into two broad groups-those with employment decreases and those with employment increases. For the latter group there is, of course, no net loss in employment associated with productivity gains (although there may have been such an association among occupational groups or individual plants in the industry).

The industries with employment decreases were divided into three subgroups:

1. Those where the percent decreases in unit man-hours equaled or exceeded the percent decrease in employment. The entire decrease in employment was associated with the decrease in unit man-hours. (Output in these industries remained stable or increased.)

2. Those where the percent decrease in unit man-hours was less than the percent decrease in employment. Part of the employment decline was disemployment-that part which is equivalent to the decline in unit man-hours. The remainder of the employment decrease was associated with a decline in output.

3. Those where unit man-hours increased. The decrease in employment was not associated with unit man-hours but was associated entirely with a decline in output.

The industries with employment increases were also divided into two subgroups:

1. Those with a decrease in unit man-hours. The increase in employment was associated with increased output.

2. Those with an increase in unit man-hours. The increase in employment in these industries was associated with the increase in unit manhours, or output or both.

1953-59. In the period 1953-59 (when total manufacturing employment declined about 6 percent), the 201 industries covered by the analysis showed a net decline of 775,000 in total employment. However, 58 industries showed an employment decline of 287,000, associated entirely with declines in unit man-hours; 55 industries had a decline of 657,000 in employment of which 309,000 was associated with a decline in unit manhours and the remainder with a decline in output. In 21 industries, there was a decline of 288,000 in employment associated with decreases in output (table 1). Thus, there was total disemployment of 596,000, and an additional decline of 635,000 (as rounded) associated with decreased output. It is estimated that for total manufacturing the figures would be 745,000 and 795,000, respectively.

There was, apparently, no special concentration among the 113 industries of disemployment. All types of industries experienced this employmentunit man-hour relationship-durables and non

durables-food, textiles, metalworking, machinery, and others. They included small and large industries, ranging from those which had fewer than 5,000 employees to those with over 500,000 in 1953.

The industries with employment increases and those where the employment decreases were associated with decreased output also varied. They included durables and nondurables, cutting across almost all of the major industry groups, and ranged from small to large.

A striking illustration of the nature of the displacement problem shows up when the detailed employment and unit man-hour analysis is applied to production workers (table 2). Of the 201 sample industries, there were 135 in which a decline of 820,000 production worker jobs was associated with decreases in unit man-hours. This yields an estimate for total manufacturing of 1,131,000 production worker jobs which disappeared during the period 1953-59, or nearly 200,000 per year. It is obvious that the disemployment figure for production workers (1,131,000) was substantially higher than that for all employees (745,000).5

As with all employees, there was no particular industrial pattern to the employment and unit man-hour relationships for production workers. Various kinds of industries were represented in the different groups and they varied in size.

ployment for 201 industries (SIC 4-digit or combinations) and employment for all manufacturing from the Bureau of the Census.

1947-57. The relationships described in the previous paragraphs cover a period (1953-59) when output showed very little gain and employment declined. A similar analysis was also prepared for 1947-57-when output rose over 40 percent and employment of production workers increased about 6 percent (table 3).

Estimates for 313 industries were derived for the 1947-57 period. Out of this total, 153 showed disemployment of 876,000 production workers. For total manufacturing, the figure would be about 883,000 production workers, or nearly 90,000 per year. At the same time, there was a decline of about 228,000 production workers (23,000 per year) associated with decreases in industry output. In other words, even in a period of rapidly rising output, accompanied by a small net increase in employment (of production workers) for total manufacturing, there were many industries in which there was a decline in employment associated with declines in unit manhours or, to a much lesser extent, with decreases in output.

As in the period 1953-59, employment increases and decreases were scattered among many industries throughout manufacturing, and again

This does not mean that output per man-hour or unit man-hour indexes for production workers are superior in any general sense to those for all employees, or vice versa. It is, however, an indication of the usefulness of various kinds of productivity measures for manpower analysis.

there was no particular industrial pattern to the decreases in employment associated with decreases in unit man-hours.

Implications for Future

The foregoing analysis has yielded two estimates of employment decreases which were associated with decreases in unit man-hours. The average annual rate was nearly 200,000 for the period 1953-59, when output showed very little gain and production worker employment in total manufacturing declined. The rate was 90,000 per year for the period 1947-57, when output and production worker employment both increased. This indicates that disemployment occurs even when there is a large increase in output but apparently at a lower rate than when output is stable, declining, or rising slowly.

It is generally expected that during the next decade output of the economy will grow at a faster pace than it did in the 1953-59 period. If the same expectation is applied to manufacturing, it does not appear unreasonable that gains in output per man-hour in many industries will be associated with decreases in employment amounting to at least 100,000 workers per year, on the average. (The estimate could be larger if the shift from production worker to nonproduction worker jobs is accelerated.) An additional reduction can be expected in certain manufacturing

industries where declines in employment will be more directly related to decreases in output.

A similar analysis could not be made for the nonmanufacturing sector because detailed data on output or employment, or for matching the two, are not available for many industries. On the whole, employment and output in nonmanufacturing increased more than in manufacturing, while output per man-hour increased less, for the two periods studied. On the other hand, there are nearly twice as many workers in nonmanufacturing as in manufacturing. In addition, some evidence suggests that there has been substantial disemployment in some nonmanufacturing industries (e.g., mining, transportation). Similar trends can be expected in the future, although not necessarily in the same industries. Tentatively, the magnitude of disemployment in nonmanufacturing (excluding agriculture) is estimated to be on the same order as that for manufacturing.

Summary and Limitations

Thus, decreases in employment associated with increased productivity in nonagricultural industries might amount to about 200,000 workers per year during the next decade assuming that output grows more rapidly than in the 1953-59 period. This is a rough approximation of disemployment between industries, but it suggests a possible magnitude and indicates the kind of exploration that

might be made to measure the volume of technological displacement or disemployment. For example, further attempts might be made to do a plant-by-plant analysis within industries.

The estimates of disemployment cannot be equated with unemployment, because some of the disemployed workers may have quit, retired, or died. In other words, these are estimates of the decline in the number of jobs available in certain groups of industries. On the other hand, these figures are only partial indicators of total technological displacement, for two major reasons:

1. The figures reflect only net disemployment between industries. They are the net result of changes in employment occurring among many plants-some with increases, others with decreases and may not reflect the sum total of disemployment which took place among all plants. For example, in those industries where employment increased, there were probably a number of plants in which employment declined. In these cases, while job opportunities in the industry as a whole did not decline, the occupations and their location may have changed.

2. The figures on disemployment do not reflect the impact of technological substitution, where plants or industries lose markets to technologically progressive competitors, to new products, and to new materials. This kind of technological change results in decreased output in some industries and

is partly accounted for in figures previously cited as decreases in employment associated with decreases in output.

It is also possible that the method of analysis, using the net change between terminal years (1953 and 1959, 1947 and 1957), could mask disemployment that may have occurred during the intervening years.

Finally, this analysis has not dealt with the question of overall demand and growth in output and productivity of the total economy and their relationship to the total labor force and unemployment. For example, with current civilian employment at about 67 million, if output per man-hour in the total economy (private and government) were to increase at a rate of 3 percent a year, output of goods and services (gross national product) would have to increase fast enough to provide about 2 million jobs per year in order to retain the same level of employment. If output per man-hour were to increase at a lower or higher rate, increases in gross national product would have to increase at a corresponding lower or higher rate. In future years, as total employment becomes a larger figure, the same rates of increase in output per man-hour would require increases in gross national product providing for a larger number of jobs. In addition, the economy must expand enough to provide jobs for over 1 million net additions to the labor force each year.